WO2017002673A1 - 分散型機器異常検出システム - Google Patents
分散型機器異常検出システム Download PDFInfo
- Publication number
- WO2017002673A1 WO2017002673A1 PCT/JP2016/068411 JP2016068411W WO2017002673A1 WO 2017002673 A1 WO2017002673 A1 WO 2017002673A1 JP 2016068411 W JP2016068411 W JP 2016068411W WO 2017002673 A1 WO2017002673 A1 WO 2017002673A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- distribution
- distribution information
- unit
- management
- distributed
- Prior art date
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 93
- 238000009826 distribution Methods 0.000 claims abstract description 305
- 238000004891 communication Methods 0.000 claims abstract description 145
- 238000005259 measurement Methods 0.000 claims abstract description 35
- 238000004364 calculation method Methods 0.000 claims abstract description 19
- 238000007726 management method Methods 0.000 claims description 232
- 230000005856 abnormality Effects 0.000 claims description 135
- 230000002159 abnormal effect Effects 0.000 claims description 53
- 230000010354 integration Effects 0.000 claims description 22
- 239000013598 vector Substances 0.000 claims description 21
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 5
- 239000006185 dispersion Substances 0.000 claims description 2
- 238000000034 method Methods 0.000 description 29
- 238000010586 diagram Methods 0.000 description 22
- 230000008569 process Effects 0.000 description 16
- 239000011159 matrix material Substances 0.000 description 14
- 238000012545 processing Methods 0.000 description 13
- 230000000694 effects Effects 0.000 description 12
- 230000006870 function Effects 0.000 description 12
- 230000004048 modification Effects 0.000 description 12
- 238000012986 modification Methods 0.000 description 12
- 230000002093 peripheral effect Effects 0.000 description 10
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 9
- 230000006866 deterioration Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 3
- 230000007774 longterm Effects 0.000 description 3
- 230000004931 aggregating effect Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000005755 formation reaction Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 229910001416 lithium ion Inorganic materials 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/005—Testing of electric installations on transport means
- G01R31/008—Testing of electric installations on transport means on air- or spacecraft, railway rolling stock or sea-going vessels
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/34—Testing dynamo-electric machines
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/40—Testing power supplies
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24136—Monitor load state of battery
Definitions
- the present invention relates to a distributed device abnormality detection system that collects physical quantity data of a large number of devices such as devices including sensors and detects abnormal devices based on the collected physical quantity data of the devices.
- a secondary battery such as a lithium ion battery has a small battery capacity, current input / output value, and voltage in a single cell, so by combining a large number of battery cells in series or in parallel, a large capacity, a high input / output value, and Used as a high-voltage battery system, for example, as a battery system mounted on a railway vehicle, it is connected in series so as to be 600V for driving or driving assistance, and the high output required to drive the motor There are things that are configured to get.
- Patent Document 1 reads out the internal resistance calculated based on the measured charge / discharge current, voltage, charging rate, and temperature of each battery cell in order of time series for each battery cell.
- An abnormality detection method is described in which a battery cell in which a change in the internal resistance of the battery is separated by a certain value or more compared to a change in the entire battery cell is detected as an abnormal cell.
- battery status data is acquired by an information exchange adapter including a data collection module and a data transmission module, and is transmitted to a data center of a deterioration degree calculation subsystem via a data communication network.
- a method for obtaining the degree of deterioration of a battery is described in which the degree of deterioration model is updated and the degree of deterioration is measured.
- the battery pack management device provided in the battery pack determines an abnormality based on the battery data of each battery cell stored in the short term, and the battery group management device that manages a plurality of battery packs has a long term.
- a failure detection system is described in which an abnormality is determined based on battery data of each battery cell stored.
- Patent Document 1 in order to use the internal resistance change of the entire battery cell, it is necessary to collect the internal resistance values of all the battery cells. Therefore, even if the method described in Patent Document 2 is applied, for example, when the system includes a large number of cells, the collection device and the analysis processing device are very expensive in terms of communication amount and calculation amount. It can be done. In order to cope with this, for example, when managing the system by dividing it into a plurality of subsystems, only information from a relatively small number of batteries can be used, and cell abnormalities may occur due to individual differences in battery cells, variations in temperature, charge state, etc. There was a problem of being hidden.
- the system is divided into a battery group management device and a battery pack management device, the battery group management device is in charge of long-term abnormality determination, and the battery pack management device is in charge of short-term abnormality determination.
- the battery group management device performs the long-term abnormality determination in an integrated manner, when the system includes a very large number of cells, the collection device and the analysis processing device are very expensive in terms of communication amount and calculation amount. Problems that can become things are not solved.
- the object of the present invention is to solve the above-described problems, and in a distributed device abnormality detection system that monitors a plurality of devices and detects an abnormality, the communication amount and the calculation amount can be reduced as compared with the prior art. Is to provide a distributed device abnormality detection system.
- the distributed device abnormality detection system is a distributed device abnormality detection system for monitoring physical quantities of a plurality of devices of substantially the same type and detecting an abnormality of each device, A plurality of device management devices connected to the plurality of devices and managing the devices; A management server device capable of communicating with the plurality of device management devices; Each of the device management devices is A first communication unit that communicates with the management server device; A measurement unit that repeatedly measures the physical quantity of the device; A distribution information generator for calculating distribution information of the device from the measured physical quantity of the device; A distribution comparison unit that calculates a difference between the distribution information of the device generated by the distribution information generation unit and the integrated distribution information of the entire device distributed from the management server device via the first communication unit; , An abnormality determination unit that determines whether the device is abnormal based on a difference between the calculated distribution information, The management server device A second communication unit that communicates with the plurality of device management devices; A first distribution integration unit that calculates integrated distribution information obtained by integrating the distribution information for each device based on the distribution information
- the distributed device abnormality detection system in the distributed device abnormality detection system that monitors a plurality of devices and detects an abnormality, the communication amount and the calculation amount can be reduced as compared with the prior art. It is possible to provide an inexpensive distributed equipment abnormality detection system capable of
- It is a flowchart which shows the control processing of the management server apparatus 104 of FIG. 7 is a table showing an operation example of the distributed device abnormality detection system 101 according to the second embodiment, and is a table showing the front and back of the input and output of the coin 1 and its peripheral distribution when the combined distribution of the input and output is considered. is there.
- 7 is a table showing an operation example of the distributed device abnormality detection system 101 according to the second embodiment, and is a table showing the front and back of the input and output of the coin 2 and its peripheral distribution when the combined distribution of input and output is considered. is there.
- FIG. 7 is a table showing an operation example of the distributed device abnormality detection system 101 according to the second embodiment, and is a table showing the front and back of the input and output of the coin 3 and its peripheral distribution when the combined distribution of input and output is considered. is there. It is a table
- FIG. 10 is a block diagram illustrating a configuration example of a device management apparatus 103 according to a third modification of the first embodiment. It is a block diagram which shows another structural example of the distributed type apparatus abnormality detection system which concerns on the modification 1 of Embodiment 2.
- FIG. It is a block diagram which shows the structural example of 103 A of apparatus management apparatuses which concern on the modification 2 of Embodiment 2.
- FIG. It is a block diagram which shows the structural example of the apparatus management apparatus 103B which concerns on the modification of Embodiment 3.
- FIG. 1 is a block diagram showing the overall configuration of a distributed device abnormality detection system 101 according to Embodiment 1 of the present invention.
- a distributed device abnormality detection system 101 according to Embodiment 1 is connected to a plurality of devices 102 that are substantially the same type, and is provided with a device management apparatus 103 provided for each device 102.
- the management server device 104, the relay server device 106, and the communication line 105 that connects the device management device 103, the relay server device 106, and the management server device 104 are configured.
- the plurality of devices 102 that are substantially the same type have a physical quantity that is a state value such as normal or abnormal because the plurality of devices 102 perform substantially the same operation.
- the plurality of devices 102 being of substantially the same type means devices having the same configuration in which an abnormality can occur in the same manner and performing the same operation.
- a relay server apparatus 106 provided in one or more stages controls a plurality of device management apparatuses 103 and lower relay server apparatuses 106. However, the relay server apparatus 106 is omitted, and the device management apparatus 103 and the management server are omitted.
- the device 104 may be directly connected by the communication line 105.
- the device 102, device 102, device management device 103, management server device 104, and relay server device 106 in the distributed device abnormality detection system 101 are respectively a computer (control means) such as a digital computer and a communication. It includes a circuit (communication means).
- each device 102 is, for example, a railway power conversion device, a storage battery, a brake device, or a motor.
- the device management apparatus 103 measures a discrete value x representing a physical quantity measured for the device 102.
- the distributed device abnormality detection system 101 detects the device 102 that is clearly different from the others by comparing the physical quantities x of a large number of devices 102, at least some elements of the physical quantity x are some devices. Assume that 102 is affected by the abnormality.
- the obtained distribution P t is affected by the device 102 in which the abnormality has occurred. For this reason, it is possible that the normal device 102 is determined to be abnormal and the abnormal device 102 is determined to be normal.
- the distributed apparatus abnormality detection system 101 uses the method described below.
- the entire plurality of devices 102 is divided into, for example, three or more (or a plurality of) groups, and information is aggregated for each group to form a distribution.
- the distribution obtained by aggregating the information of the devices in group m is written as P g, m .
- the total number of groups is M
- all M P g, m are distributed to all devices.
- the device n repeatedly measures the physical quantity of the device itself, for example, in a predetermined cycle, and configures a distribution Pe, n that the measured physical quantity follows.
- the probability that the distribution P g, m is not affected by the abnormal device 102 is (1-r )
- is the number of elements of N (m).
- (1-r) N (m) can be greater than 0.5.
- a majority of the distributions P g, 1 ,..., P g, M are not affected by the abnormal device 102 with high probability.
- the degree of difference takes a small value with respect to the majority distribution P g, m , and is arbitrary for the distribution P g, m influenced by a small number of abnormal devices 102.
- the degree of difference takes a large value for the majority distribution P g, m and takes some arbitrary value for the distribution P g, m affected by a small number of abnormal devices 102.
- the abnormal device 102 can be detected by comparing the median value of the different degrees with a predetermined threshold value.
- the abnormality rate r is sufficiently small, the grouping may be omitted and the information of all the devices 102 may be aggregated to constitute a distribution.
- the distribution P g, m estimated from the measured values of the device 102 is expressed by the following equation.
- the probability distribution can be expressed by a vector of k elements. Assuming that the total number of groups M is fixed, each device 102 randomly selects a group m to which the device 102 belongs, and the m-th column is the vector Pe, n and the other columns are 0 in the upper relay server device 106. And a number vector such that only the m-th element is 1.
- the relay server device 106 or the management server device 104 calculates the sum of the matrix transmitted from the device 102 or the lower relay server device 106 managed by the relay server device 106 or the lower relay server device 106 and the sum of the number vectors. It transmits to the server device 104.
- the management server device 104 calculates distributions P g, 1 ,..., P g, M by dividing each column of the matrix by the corresponding element of the number vector. In this way, by performing the addition processing in each relay server device 106 in a distributed manner, the distribution P g, 1 ,..., P g, M can be obtained without performing large-scale calculation processing in the management server device 104. it can. Further, the communication amount between any of the devices 102, the relay server device 106, and the management server device 104 is proportional to (K + 1) ⁇ M integers, and the communication amount is the total number of devices 102. Even if it increases, it does not change.
- distributions P g, 1 ,..., P g, M obtained in the management server device 104 are distributed to each device 102, and an abnormality determination unit 307 (described later) of the device management device 103 connected to each device 102. 3) calculates the degree of difference from its distribution from the distributed distributions P g, 1 ,..., P g, M , takes the median value, compares it with a predetermined threshold value, Is detected.
- the distributed device abnormality detection system 101 will be described using a simple example.
- the device 102 is a coin device (hereinafter referred to as “coin” in this specification) that outputs a value of 1 or 0 corresponding to the front and back sides that appear by throwing as a physical quantity.
- coin hereinafter referred to as “coin” in this specification
- normal coins in this example do not necessarily mean coins that appear with equal probability.
- the distributed apparatus abnormality detection system 101 detects a small number of coins that appear front and back with an appearance probability different from the appearance probability, assuming that most coins have the same appearance probability. .
- FIG. 2 is a table showing an operation example of the distributed device abnormality detection system 101 of FIG. FIG. 2 shows a case where the distributed device abnormality detection system 110 according to the first embodiment is applied to seven coins.
- the second line of FIG. 2 shows the appearance probability distribution of the front and back of each coin.
- Each device management apparatus 103 can obtain the appearance probability distribution Pe, n by counting the number of times the coins of the device 102 managed by the device management device 103 appear.
- the management server device 104 assigns groups to which the devices 102 belong so that the number of groups becomes sufficiently average. The assigned group is shown in the third line of FIG.
- each device management apparatus 103 generates a matrix and a number vector such that each column is the appearance probability distribution Pe, n of the device 102 that it manages.
- a matrix and a number vector that are the appearance probability distribution Pe, n of the device 102 are shown in the fourth and fifth lines in FIG.
- the relay server device 106 and the management server device 104 add the matrix and the number vector.
- the distribution [P g, 1 P g, 2 P g, 3 ] for each group can be obtained. .
- the degree of difference (difference) between distribution information can be measured by relative entropy.
- P g, m ) between distribution information can be calculated by the following equation.
- the sum relating to the physical quantity type k may be taken for k such that the distribution P g, m (x k ) relating to the physical quantity x k is not zero.
- the relative entropy is 0 if the distribution Pe, n and the distribution Pg, m are the same, and the non-zero pattern in the table is obtained below.
- Pattern A degree of difference D (P e, n
- Pattern B degree of difference D (P e, 4
- Pattern C degree of difference D (P e, 4
- Calculating the median relative entropy for the group it can be determined that the coin 4 is abnormal because only the coin 4 has a non-zero value of 0.2310, and has the highest median relative entropy.
- P g, 1 ) is affected by the abnormal device 102, The following formula.
- the coin 4 and other coins have the same degree of difference. However, since the group 2 and the group 3 are not affected by the abnormal device 102, it can be determined that the coin 4 is abnormal by taking the median value.
- an abnormality cannot be determined with the threshold value 0.005 and the threshold value 0.2.
- it can be determined when grouping is performed. That is, the detection power is improved, which is not very effective in the simple device 102 as in this example, but has a significant effect in a device having a more complex physical quantity.
- the number of groups M does not always need to be fixed. For example, prior to processing, the number of all devices 102 (total number of devices) is totaled using communication via the communication line 105, and the management server device 104 determines the number M of groups based on the total number of total devices.
- the distributed group number M is distributed to each device 102, and each device 102 may determine its own group m based on the distributed group number M.
- each device management apparatus 103 has only a unique ID having a sufficient number of bits, and the relay server apparatus 106 or the management server apparatus 104 generates groups 1,..., M sufficiently randomly based on each ID.
- the group m may be determined by using such a hash function.
- Ethernet registered trademark
- SHA-2 SHA-2
- the device management apparatus 103 transmits the median value to the relay server apparatus 106 or the management server apparatus 104 again via the communication unit 304 (see FIG. 3). Then, the relay server device 106 and the management server device 104 collect statistical information about the median device 102, and the management server device 104 determines a threshold value based on the statistical information. Distribution to the relay server device 106 or the device management device 103, and the device management device 103 may determine whether the device 102 is abnormal based on the threshold value.
- the threshold value determination method is appropriately selected so as to obtain a desired detection rate and false detection rate in consideration of the urgency of the failure to be detected and the magnitude of the influence. For example, for a failure with a high degree of urgency that should be detected immediately, a threshold value including erroneous detection is set.
- FIG. 3 is a block diagram showing the configuration of the device management apparatus 103 of FIG. 3, the device management apparatus 103 includes a measurement unit 301, a device ID memory 302, a distribution information generation unit 303, a communication unit 304, a distribution comparison unit 305, a robust average calculation unit 306, and an abnormality determination unit 307. And a display unit 308.
- the communication unit 304 is connected to the management server device 104 or the relay server device 106 via the communication line 105 and can communicate with the management server device 104 or the relay server device 106.
- the measurement unit 301 measures the physical quantity x k of the device 102 and outputs it to the distribution information generation unit 303.
- the device ID memory 302 holds a unique device ID for the connected device 102 and outputs information of the communication ID to the communication unit 304.
- Distribution information generating unit 303 counts the number of observations for each physical quantity x k, counted physical quantity generated by the communication unit 304 and the distribution of the distribution information of the device 102 the number of observations for each x k by dividing the total number of observations Output to the comparison unit 305.
- the communication unit 304 transmits the device ID held in the device ID memory 302 and the distribution information of the device 102 generated by the distribution information generation unit 303 to the relay server device 106 or the management server device 104.
- the distribution comparison unit 305 includes distribution information of at least three groups distributed from the relay server device 106 or the management server device 104 via the communication unit 304, and distribution information of the device 102 generated by the distribution information generation unit 303.
- Relative entropy (difference) is calculated according to the equation (2) and output to the robust average calculation unit 306.
- the distribution comparison unit 305 includes at least three groups of distribution information distributed from the relay server device 106 or the management server device 104 via the communication unit 304, and the distribution of the devices 102 generated by the distribution information generation unit 303.
- the relative entropy (difference) with the information may be calculated according to the equation (2) and transmitted to the relay server device 106 or the management server device 104 via the communication unit 304.
- the robust average calculation unit 306 calculates the median value of the group of the relative entropy calculated by the distribution comparison unit 305 as the robust average of the device, outputs the calculated robust average to the abnormality determination unit 307, and outputs the communication unit 304.
- the abnormality determination unit 307 compares the robust average of the device 102 calculated by the robust average calculation unit 306 with the threshold value delivered from the relay server device 106 or the management server device 104 via the communication unit 304, and the median Is greater than the threshold, the device 102 is determined to be abnormal.
- the abnormality determination unit 307 determines that the device 102 is abnormal, the abnormality determination unit 307 displays the determination result on the display unit 102A of the device 102 or transmits the determination result to the relay server device 106 or the management server device 104 via the communication unit 304.
- the distribution information for example the number of observations for each physical quantity x k, may be two numbers of sets of the total number of observations.
- the relative entropy may be a Kolmogorov-Smirnov distance, for example.
- the median may be, for example, a pruning average or a quartile average.
- FIG. 4 is a block diagram showing the configuration of the relay server device 106 of FIG.
- the relay server device 106 includes a group assignment unit 401, an upper communication unit 402, a lower communication unit 403, a distribution integration unit 404, a distribution distribution unit 405, a statistical information collection unit 406, a threshold value And a distribution unit 407.
- the lower-level communication unit 403 is connected to the relay server device 106 or the device management device 103 via the communication line 105 and can communicate with the relay server device 106 or the device management device 103.
- the upper communication unit 402 is connected to the management server device 104 or the relay server device 106 via another communication line 105 and can communicate with the management server device 104 or the relay server device 106.
- the group allocation unit 401 includes the total number of groups received from the management server device 104 or the upper relay server device 106 via the higher-level communication unit 402 and the device management device 103 received via the lower-level communication unit 403. Based on the ID, each device management apparatus 103 is assigned to a group using a hash function or the like. In addition, the group assignment unit 401 distributes the total number of groups to the lower relay server device 106 via the lower communication unit 403. Next, the distribution integration unit 404 generates, for example, a weighted average of the distribution information of the device management apparatus 103 and the lower relay server apparatus 106 received via the lower communication unit 403 according to the number of devices to generate integrated distribution information.
- the data is transmitted to the management server device 104 or the upper relay server device 106 via the upper communication unit 402.
- the distribution distribution unit 405 transmits the distribution information for each group distributed by the management server device 104 or the upper relay server device 106 via the higher-level communication unit 402, to the device management device 103 or the lower-level via the lower-level communication unit 403.
- the statistical information collection unit 406 integrates the robust average of the devices calculated by the device management device 103 and the robust average calculation unit 306 of the lower relay server device 106 received via the lower communication unit 403, and obtains the integrated data.
- the data is transmitted to the management server device 104 or the upper relay server device 106 via the upper communication unit 402.
- the threshold distribution unit 407 transmits the threshold distributed by the management server device 104 or the upper relay server device 106 via the upper communication unit 402 to the device management device 103 or the lower relay server via the lower communication unit 403. Delivered to the device 106.
- the statistical value of the statistical calculation by the statistical information collection unit 406 may be, for example, the robust average histogram, or the average, variance, kurtosis, skewness, maximum value, minimum value, median value, maximum value of the robust average. It may be a statistical value including a frequent value or a combination of the plurality of statistical values.
- the threshold value may include a plurality of threshold values representing the degree of abnormality.
- FIG. 5 is a block diagram showing the configuration of the management server device 104 of FIG.
- the management server device 104 includes a group assignment unit 501, a communication unit 502, a distribution integration unit 503, and a threshold value determination unit 504.
- the group allocating unit 501 determines that the number of devices 102 included in each group is inequality
- the total number of groups is determined so as to satisfy, and based on the device ID received from the device management apparatus 103 via the communication unit 502, the device management apparatus 103 is allocated to a group using a hash function or the like, and data of the group allocation result Is delivered to the relay server device 106.
- the distribution integration unit 503 integrates the distribution information of each group received from the relay server apparatus 106 or the device management apparatus 103 via the communication unit 502, and transmits the data of the integrated distribution information of the group as a result of the integration to the communication unit 502.
- the threshold value determination unit 504 determines a threshold value for distinguishing between abnormal and normal from the statistical data received from the relay server device 106 or the device management device 103 via the communication unit 502, and the determined threshold value.
- the value data is distributed to the relay server device 106 or the device management device 103 via the communication unit 502.
- FIG. 6 is a sequence diagram showing a communication procedure of the distributed device abnormality detection system 101 of FIG.
- the management server device 104 determines the number of groups, and distributes the determined number of groups to the relay server device 106.
- the relay server device 106 assigns each device to a group based on the number of groups distributed from the management server device 104.
- the device management apparatus 103 measures the physical quantity of the device 102, generates distribution information, and transmits it to the relay server device 106.
- the relay server device 106 integrates the received distribution information for each group, and transmits it to the management server device 104.
- the management server device 104 integrates the received distribution information for each group and distributes it to the relay server device 106.
- the relay server device 106 further distributes the distributed distribution information for each group to the device management device 103.
- the device management apparatus 103 calculates relative entropy from the distributed distribution information for each group and the distribution information of the device 102 managed by itself, calculates the robust average based on the relative entropy, and calculates the robust average data. Transmit to the relay server device 106.
- the relay server device 106 calculates the received robust average statistical value, and transmits the calculated robust average statistical value data (hereinafter referred to as statistical information) to the management server device 104.
- the management server device 104 integrates the received statistical information, determines a threshold value for distinguishing between abnormality and normality of each device 102 based on the integrated result, and distributes the determined threshold value to the relay server device 106. .
- the relay server device 106 further distributes the distributed threshold value to the device management device 103.
- the device management apparatus 103 determines whether or not the device 102 is abnormal from the distributed threshold value and the robust average calculated by itself.
- the relay server device 106 on the left side of the two relay server devices 106 illustrated in FIG. 6 is a relay server device belonging to the group of the device 102 and the device management device 103 illustrated in FIG.
- the relay server device 106 on the right side is a relay server device belonging to the group of the device 102 and the device management device 103 not shown in FIG.
- FIG. 7 is a flowchart showing a control process of the device management apparatus 103 of FIG. Each process in FIG. 7 is executed corresponding to each processing unit 301 to 307 in FIG. Each process in FIG. 7 is allowed to be replaced as long as the process dependency is not broken.
- the device management apparatus 103 may be implemented as software of a microcontroller, or may be implemented as an integrated logic circuit.
- the device ID is transmitted in step S1 of FIG. 7, the physical quantity is measured in step S2, the distribution information is generated in step S3, the distribution information generated in step S4 is transmitted, and the distribution information for each group is determined in step S5. Receive. Next, the distribution information for each group received in step S6 is compared, the robust average is calculated in step S7, the calculated robust average is transmitted in step S9, and the threshold is received in step S9. In step S10, it is determined whether each device 102 is abnormal based on the threshold value, and the control process ends.
- FIG. 8 is a flowchart showing a control process of the relay server device 106 of FIG. Each process in FIG. 8 is executed corresponding to each processing unit 401 to 407 in FIG. Each process in FIG. 8 is allowed to be replaced as long as the process dependency is not broken.
- the relay server device 106 may be implemented as, for example, software of a microcontroller or may be implemented as an integrated logic circuit.
- step S11 of FIG. 8 the device ID is transmitted, the number of groups is received in step S12, the group is assigned in step S13, and the number of groups is distributed in step S14.
- distribution information is received in step S15, distribution information is integrated in step S16, and the integrated distribution information is transmitted in step S17.
- distribution information for each group is received in step S18, distribution information for each group is distributed in step S19, a robust average is received in step S20, statistics are calculated in step S21, and statistics calculated in step S22 are calculated.
- the amount is transmitted, the threshold value is received in step S23, the threshold value is distributed in step S24, and the control process is terminated.
- FIG. 9 is a flowchart showing the control processing of the management server device 104 of FIG. Each process in FIG. 9 is executed corresponding to each processing unit 501 to 504 in FIG. Each process in FIG. 9 can be replaced as long as the dependency of the process is not broken.
- the management server device 104 may be implemented as, for example, software of a microcontroller or may be implemented as an integrated logic circuit.
- the device ID is received in step S31 of FIG. 9, the group is assigned in step S32, the number of groups is distributed in step S33, the distribution information of each group is received in step S34, and the distribution information of each group is received in step S35.
- the distribution information of the group integrated in step S36 is distributed, the statistics are received in step S37, the threshold is determined based on the statistics in step S38, and the threshold determined in step S39 Is delivered and the control process is terminated.
- the communication line 105 is configured using a communication line such as Ethernet (registered trademark) (Ethernet).
- Ethernet registered trademark
- the tree structure constituted by the device management apparatus 103, the relay server apparatus 106, and the management server apparatus 104 is logical communication using, for example, the Internet protocol (Internet Protocol) constructed on the Ethernet (registered trademark) (Ethernet).
- Internet Protocol Internet Protocol constructed on the Ethernet (registered trademark) (Ethernet).
- an abnormality of the device 102 can be detected by comparing a large number of devices 102 of substantially the same type. Can be detected. Also, by performing the detection process in a distributed manner by the device management apparatus 103 and the relay server apparatus 106, it is possible to manage a very large number of devices 102 without using a powerful computer or broadband communication means. Further, by grouping a plurality of devices 102, even if there is an abnormal device 102, the statistical information of the normal device 102 is correctly extracted, and the abnormal device 102 is based on the statistical information of the device 102. Can be detected.
- FIG. 12 is a block diagram showing an overall configuration of a distributed device abnormality detection system 101 according to the first modification of the first embodiment.
- the configuration having the relay server device 106 configured in multiple stages has been described.
- the relay server apparatus 106 is omitted and a large number of device management apparatuses 103 and management server apparatuses 104 are directly connected. You may connect by the communication line 105.
- FIG. Even in such a configuration, by calculating distribution information for each assigned group, it is possible to obtain an effect that can be managed without using a powerful computer and / or broadband communication means. .
- FIG. 13 is a block diagram illustrating another configuration example of the distributed equipment abnormality detection system according to the second modification of the first embodiment, and a storage battery in which the configuration described in the present embodiment is mounted on the railway vehicles 200 and 200.
- An example of a configuration when applied to 102B (an example of the device 102) is shown.
- the distributed device abnormality detection system 101 is connected to a plurality of storage batteries 102 ⁇ / b> B, which are substantially the same type of devices, and the device management apparatus 103 provided for each storage battery 102 ⁇ / b> B.
- a management server device that controls the inside of the type device abnormality detection system 101, and a communication line 105 that connects the device management device 103, the relay server device 106, and the management server device 104.
- the relay server device 106 is provided in different railway vehicles 200 and 200 of the same train train 201, and supervises the device management device 103 provided in the railway vehicles 200 and 200.
- the some storage battery 102B is disperse
- the currents in the plurality of storage batteries 102B can be regarded as equal currents. Due to the difference in internal resistance of the storage battery, the response of the voltage to the current may be different in the abnormal storage battery as compared with the normal storage battery. As an example, even if the current is the same, the terminal voltage may be different in an abnormal storage battery compared to a normal storage battery. As another example, the time constant of the voltage change with respect to the current change may be different in the abnormal storage battery as compared with the normal storage battery.
- FIG. 14 is a block diagram illustrating a configuration example of the device management apparatus 103 according to the third modification of the first embodiment, and illustrates an example of a configuration when the device management apparatus of FIG. 3 is applied to a storage battery abnormality detection system.
- FIG. 14 only the measurement unit 301 is extracted from the components of the device management apparatus 103 to clarify the configuration.
- the measurement unit 301 includes a voltage measurement unit 141 that measures a physical quantity of the storage battery 102B, and a voltage determination signal generation unit 142.
- the voltage measuring unit 141 measures the terminal voltage of the storage battery 102B as a physical quantity. For example, the terminal voltage of the storage battery 102 ⁇ / b> B is converted to the input voltage level of the AD converter by a transformer or the like, then measured by the voltage measuring unit 141, and the measured voltage is output to the voltage determination signal generating unit 142.
- the voltage determination signal generation unit 142 is obtained, for example, by determining whether the measured voltage is equal to or higher than a predetermined upper limit value, lower than a predetermined lower limit value, or between the lower limit value and the upper limit value.
- the abnormality determination unit 307 (FIG. 3) of the device management apparatus 103 detects the abnormal storage battery 102B based on the degree of difference in the appearance probability of the voltage determination signal.
- the configuration example of FIG. 14 can detect an abnormality in the storage battery 102B with a low-cost configuration when it can be considered that the usage conditions such as the charging rate and the deterioration state of the storage battery 102B are uniform among the plurality of storage batteries 102B. This is an effective configuration in a certain point.
- FIG. 1 The second embodiment according to the present invention is applied when the distributed device abnormality detection system 101 according to the first embodiment can define and measure the input / output data (input / output value) of the device 102 instead of the physical quantity of the device 102.
- the device 102 that is expressed by a function f different from the majority of other devices 102 is detected.
- the input x varies from device to device and does not necessarily follow the same distribution.
- the distributed device abnormality detection system 101 in the present embodiment will be described using a simple example as in the first embodiment. Assume that the device 102 is a coin and there are three coins 1, 2, and 3. Consider the front and back of the coin before throwing as input and the front and back of the coin as a result of throwing as output.
- FIG. 10A to 10C are tables showing an operation example of the distributed device abnormality detection system 101 according to the second embodiment, and FIG. 10A shows the input and output of the coin 1 when the combined distribution of input and output is considered.
- FIG. 10B is a table showing the input and output of coin 2 and its peripheral distribution when considering a combined distribution of input and output
- FIG. 10C is a table showing the distribution of the input and output. It is a table
- FIG. 11A to 11C are tables showing an example of the operation of the distributed equipment abnormality detection system 101 according to the second embodiment.
- FIG. 11A shows the front and back of the input and output of the coin 1 when the conditional probability is considered.
- 11B is a table showing the peripheral distribution
- FIG. 11B is a table showing the front and back of coin 2 when the conditional probability is considered, and its peripheral distribution
- FIG. 11C is the coin 3 when the conditional probability is considered.
- the coins 1 and 2 are normal and the coin 3 is abnormal.
- the difference between the conditional probabilities is measured by relative entropy.
- an abnormality determination having desirable properties is realized by measuring the expected value D (P e, n
- the distributed device abnormality detection system 101 is exactly the same as that of the first embodiment except that the formula (2) is changed to the formula (11).
- the first embodiment corresponds to the case where the input is not defined and only the output is defined and the physical quantity can be measured in the present embodiment, and the effects described in the first embodiment can be obtained.
- FIG. 15 is a block diagram illustrating another configuration example of the distributed equipment abnormality detection system according to the first modification of the second embodiment.
- the configuration described in the present embodiment is applied to the storage battery 102B mounted on the railcar 200. An example of the configuration when applied is shown.
- a plurality of railway trains 201 configured are traveling on different routes. Each train train 201 differs from the train 201 described in FIG. 13 in the following points.
- the relay server device 106m has an upper level and a lower level.
- the relay server device 106m has an antenna 106a and has a wireless communication function with the management server device 104m.
- the management server device 104m is installed in the data center.
- the management server device 104m has an antenna 104a and has a wireless communication function with the relay server device 106m.
- the currents in the plurality of storage batteries 102 ⁇ / b> B can be regarded as different currents in different formations.
- FIG. 16 is a block diagram illustrating a configuration example of the device management apparatus 103A according to the second modification of the second embodiment.
- the measurement unit 161 in FIG. 16 includes a current measurement unit in addition to the configuration of the measurement unit 151 in FIG. 162, a current determination signal generation unit 163, and a data integration unit 164.
- the current measurement unit 162 measures the current of the storage battery 102 ⁇ / b> B as a physical quantity and outputs the measured current to the current determination signal generation unit 163.
- the current is measured by, for example, inserting a minute resistance on the current path, measuring the voltage at both ends, and dividing by the resistance value.
- the current determination signal generation unit 163 is obtained, for example, by determining whether the measured current is a predetermined upper limit value or more, a predetermined lower limit value or less, or between the lower limit value and the upper limit value.
- a current determination signal that is a ternary value is output to the data integration unit 164.
- the upper limit value and the lower limit value are determined so that, for example, the presence / absence of current and the sign can be determined.
- the data integration unit 165 outputs integrated data by integrating the input voltage determination signal and current determination signal. By regarding that the input is the current determination signal and the output is the output determination signal, the abnormal state of the storage battery 102B can be detected.
- the configuration example of FIG. 16 is an abnormal condition among a plurality of storage batteries 102B mounted in different formations in a state where use conditions such as the charging rate and deterioration state of the storage battery 102B can be regarded as being uniform between the storage batteries 102B. This configuration is effective in that the storage battery 102B can be detected.
- Embodiment 3 The third embodiment according to the present invention is applied when the physical quantity measured by the device 102 is not a finite set in the distributed device abnormality detection system 101 according to the first embodiment.
- the physical quantity x of the device 102 is represented by, for example, an integer value, a real value, or a vector value random variable.
- This random variable may be a physical quantity expressed by an instrumental constant such as an internal mass, length, time, current, temperature, substance amount, and luminous intensity measured by the instrument 102, or a combination of these instrumental constants.
- the physical quantity x of the device 102 may use the communication amount of the device 102, the occupancy rate of the arithmetic device or the storage device, the transition state in the software, or the like. Further, the physical quantity x of the device 102 may be replaced with a feature quantity calculated from the various physical quantities and states. In particular, it may be a quantity that is subjected to nonlinear transformation in order to treat the quantities as random variables that follow a theoretically easy-to-handle distribution. The difference between the distributions of the random variables of the real value or the real vector value can be measured by the relative entropy D (P e, n
- the probabilities pe, n , pg, m are density functions of the corresponding probability distributions.
- the probabilities p e, n , pg, m are density functions of the corresponding probability distributions.
- a data amount proportional to the number of the devices 102 is required to express the distribution, which is a method not suitable for managing a very large number of devices 102.
- a certain theoretical distribution is assumed as the distribution P e, n , P g, m , and the density p e, n , pg, m is estimated by estimating the parameters of the theoretical distribution from the observed values of the device 102 .
- m is calculated.
- a Gaussian (Gauss) distribution as a theoretical distribution
- the following expression, distribution and mean ⁇ e, n ⁇ e, n A density function of a Gaussian distribution may be used.
- K is 1 in the case of a real number, and is the dimension in the case of a real vector.
- D P e, n
- any theoretical distribution may be selected in accordance with the device 102.
- the parameter is the degree of freedom d
- the degree of freedom d can be estimated by taking the average of the observed values.
- the statistics of observations to be used are not only the average and variance of the observations, but also kurtosis and skewness, higher moments, maximum values, minimum values, etc. may be used. Can be efficiently integrated in the device. Even in the case of a random variable that takes an integer value or an integer vector value, the present embodiment can be applied in the same manner by replacing the integral of the density function with an infinite sum of probabilities.
- the method described in this embodiment may be used when the physical quantity of the device 102 is a finite set as in the first embodiment. In this case, since the distribution information is aggregated into a small number of parameters, it is possible to improve the transmission / reception of the distribution information.
- the distributed device abnormality detection system 101 is (1) Instead of the device management apparatus 103 and the relay server apparatus 106 integrating the distribution information P e, n , P g, m and transmitting them to the management server apparatus 104 or the upper relay server apparatus 106, the average ⁇ e , N and the distribution ⁇ e , n are integrated and transmitted to the management server device 104 or the upper relay server device 106; (2)
- the distribution comparison unit 305 is the same as the distributed apparatus abnormality detection system 101 described in the first embodiment of the present invention except that the expression (14) is used instead of the expression (2).
- the effect described in the first embodiment can be obtained. Further, by applying the method described in this embodiment to the device 102 in which the physical quantity of the device is a finite set, distribution information can be efficiently transmitted and received.
- FIG. 17 is a block diagram showing a configuration example of a device management apparatus 103B according to a modification of the third embodiment, and a configuration when the configuration described in the present embodiment is applied to the storage battery 102B mounted on the railcar 200. An example is shown.
- the device management apparatus 103B in FIG. 17 is characterized in that a measurement unit 171 is provided instead of the measurement unit 301, as compared with the device management apparatus 103A in FIG.
- the measurement unit 171 has a configuration in which the voltage determination signal generation unit 142 and the current determination signal generation unit 163 are deleted from the measurement unit 161.
- the data integration unit 164 instead of the ternarized voltage determination signal and current determination signal, voltage measurement values and current measurement values that are not a finite set are integrated and output as integrated data.
- the configuration example of FIG. 17 is an effective configuration in that an abnormal storage battery 102B can be detected in a state where it can be considered that usage conditions such as a charging rate and a deterioration state of the storage battery 102B are different between the storage batteries 102B. is there.
- Embodiment 4 In the distributed device abnormality detection system 101 according to the first embodiment, the fourth embodiment of the present invention can measure the input / output data (input / output values) of the device 102 instead of the physical quantity of the device 102, and This is applied when the output data (input / output values) is not a finite set.
- This input / output data (input / output value) is represented by a random variable of an integer value, a real value or a vector value, for example.
- This random variable may be a physical quantity expressed by a device constant such as the mass, length, time, current, temperature, substance amount, and luminous intensity of the device 102 or a combination of the device constants.
- it may be the communication amount of the device 102, the occupation rate of the arithmetic device or the storage device, the state in the state transition in the software, and the like. Further, it may be a feature quantity calculated from the various physical quantities and states. In particular, it may be a quantity that is subjected to nonlinear transformation in order to treat the quantities as random variables that follow a theoretically easy-to-handle distribution.
- the input distribution P e, n (x) is a Gaussian distribution with mean ⁇ e, n, x , variance ⁇ e, n, xx , and conditional distribution P e, n ( y
- x) with respect to the input x of the output y is given by the following equation and the Gauss distribution of the variance-covariance matrix.
- P g, m ) can be calculated analytically by the following equation.
- the overall configuration of the distributed device abnormality detection system 101 in the present embodiment is the same as that in Embodiment 2 except that Expression (18) is used instead of Expression (11).
- the input / output data (input / output value) of the device 102 not the physical quantity of the device 102, can be measured and the input / output data (input / output value) is not a finite set.
- the effect described in Form 1 can be obtained.
- the distribution information can be efficiently obtained by applying the method described in the present embodiment to a device whose input / output data (input / output values) of the device is a finite set. Can be sent and received.
- FIG. 17 is a combination of the configuration with the measurement unit 171 shown in FIG.
- the storage batteries 102B mounted on the railway vehicle 200 in a state or physical quantity that can be regarded as different between the plurality of storage batteries 102B, such as the charging rate and / or deterioration state of the storage battery 102B
- the distributed device abnormality detection system is a distributed device abnormality detection system for monitoring physical quantities of a plurality of devices of substantially the same type and detecting an abnormality of each device, A plurality of device management devices connected to the plurality of devices and managing the devices; A management server device capable of communicating with the plurality of device management devices; Each of the device management devices is A first communication unit that communicates with the management server device; A measurement unit that repeatedly measures the physical quantity of the device; A distribution information generator for calculating distribution information of the device from the measured physical quantity of the device; A distribution comparison unit that calculates a difference between the distribution information of the device generated by the distribution information generation unit and the integrated distribution information of the entire device distributed from the management server device via the first communication unit; , An abnormality determination unit that determines whether the device is abnormal based on a difference between the calculated distribution information, The management server device A second communication unit that communicates with the plurality of device management devices; A first distribution integration unit that calculates integrated distribution information obtained by integrating the distribution information for each device based
- an abnormality of the device can be detected by comparing a number of similar devices.
- the distributed apparatus abnormality detection system is the distributed apparatus abnormality detection system according to the first aspect.
- the device management apparatus A device ID memory holding an ID unique to the device; Based on distribution information of each group distributed for a plurality of groups configured by dividing the plurality of devices, further comprising a robust average calculation unit for calculating a robust average of differences between distribution information,
- the device ID memory transmits the ID of the device to the management server device via the first communication unit,
- the distribution comparison unit calculates a difference between the integrated distribution information of at least three groups distributed from the management server device via the first communication unit and the distribution information of the generated device, respectively.
- the abnormality determination unit determines whether the device is abnormal based on the calculated robust average,
- the management server device The number of groups is determined based on the total number of the plurality of devices, and the group to which each device connected to its own device belongs is determined based on the device ID received from the device management device via the second communication unit.
- a second group assigning unit that The second communication unit distributes the number of groups to the device management apparatus;
- the first distribution integration unit integrates distribution information of devices received from the device management device via the second communication unit for each group, and transmits the information to the device management device via the second communication unit. It is characterized by delivering.
- the distributed apparatus abnormality detection system is the distributed apparatus abnormality detection system according to the second aspect.
- the distribution comparison unit transmits the difference between the calculated distribution information to the management server device via the first communication unit, or the robust average calculation unit determines the difference between the calculated distribution information.
- the abnormality determination unit determines whether or not the device is in an abnormal state based on the calculated robust average and a threshold distributed from the management server device via the first communication unit,
- the management server device A statistical information collection unit that calculates a difference between distribution information transmitted from the plurality of device management apparatuses or a robust average statistic of a difference between the distribution information; and A threshold value determination unit that determines a threshold value based on a difference between the distribution information received from the device management apparatus via the second communication unit or a statistic of a robust average thereof;
- the second communication unit distributes the threshold value determined by the threshold value determination unit to the device management apparatus.
- the distributed apparatus abnormality detection system is the distributed apparatus abnormality detection system according to the second aspect.
- the measurement unit repeatedly measures an input value to the device and an output value from the device,
- the distribution information generation unit generates a distribution of input values to the measured device, and a combined distribution of input values to the device and output values from the device,
- the distribution comparison unit calculates a first conditional distribution relating to a certain input from the combined distribution of the input value to the generated device and the output value from the device, and transmits the first communication from the management server device to the first communication.
- the input / output data (input / output value) to the device can be measured instead of the state of the device, the difference in input for each device is excluded, and the device It is possible to detect its own abnormality.
- a distributed device abnormality detection system is the distributed device abnormality detection system according to one of the second to fourth aspects.
- the measurement unit repeatedly measures an integer value, a real value, or a vector value as a physical quantity of the device or an input / output value of the device,
- the distribution information generation unit calculates a predetermined statistical value of the physical quantity or input / output measurement value of the repeatedly measured device, estimates a parameter of a predetermined theoretical distribution based on the calculated statistical value
- the distribution comparison unit calculates a difference between the distribution information from the estimated theoretical distribution parameter and the integrated distribution parameter distributed from the management server device via the first communication unit,
- the first distribution integration unit integrates predetermined statistical values of physical quantities or input / output measurement values of the device received from the device management apparatus, and sets predetermined integrated distribution parameters based on the integrated statistical values. Is estimated.
- the distributed equipment abnormality detection system is the distributed equipment abnormality detection system according to the fifth aspect, wherein the predetermined statistical values are average, dispersion, moment, maximum, minimum, median, or those It is the combination of these.
- the distributed device abnormality detection system is the distributed device abnormality detection system according to the fifth or sixth aspect, wherein the measured value of the state of the device or the input / output of the device is measured by the device.
- the integer value, real value, or vector value obtained as a result of performing linear or non-linear conversion on the integer value, real value, or vector value is a feature.
- the distributed apparatus abnormality detection system is the distributed apparatus abnormality detection system according to the first aspect. And further comprising at least one relay server device capable of communicating with the plurality of device management devices or lower relay server devices and capable of communicating with the management server device or higher relay server device,
- the relay server device A lower communication unit that communicates with the plurality of device management devices or the lower relay server device; An upper communication unit that communicates with the management server device or the upper relay server device; Based on the ID of the device received from the device management device via the lower communication unit and the number of groups distributed from the management server device or the higher relay server device via the higher communication unit A first group assigning unit for determining a group to which each device connected to the device belongs,
- the distribution information of devices received via the lower communication unit from the device management device or the lower relay server device is integrated for each group, and the higher communication unit is transferred from the management server device or the higher relay server device.
- a second distribution integration unit that distributes integrated distribution information for each group distributed via the lower communication unit to the device management apparatus or the lower relay server apparatus;
- the integrated distribution information or the parameters of the integrated distribution information distributed from the management server device or the upper relay server device via the upper communication unit, the device management device or the lower relay server via the lower communication unit A distribution delivery unit for delivery to the device;
- a statistical information collection unit that calculates a difference between distribution information transmitted from the plurality of device management apparatuses or a robust average statistic of a difference between the distribution information; and Threshold distribution for further distributing the threshold distributed from the management server device or higher-level server device via the higher-level communication unit to the device management device or lower-level relay server device via the lower-level communication unit And a section.
- high-performance information communication can be achieved even when the system includes a very large number of devices by performing distribution integration and distribution in the relay server device in a distributed manner.
- a system can be constructed at low cost without using equipment.
- a distributed device abnormality detection system is the distributed device abnormality detection system according to any one of the first to eighth aspects, wherein each of the devices includes a power conversion device, a storage battery, It is a brake device or a motor.
- each device is a power conversion device, a storage battery, a brake device, or a motor.
- the device management apparatus is A device management apparatus of a distributed device abnormality detection system for monitoring physical quantities of a plurality of devices of substantially the same type and detecting an abnormality of each device, each connected to the plurality of devices, A plurality of device management devices for managing devices,
- the integrated distribution information obtained by integrating the distribution information about the physical quantity of each device of at least three groups configured by dividing the plurality of devices for each group, and the difference between the distribution information of each device, respectively A distribution comparison unit to be calculated;
- a robust average calculator that calculates a robust average of differences between distribution information based on the distribution information of each group;
- an abnormality determining unit that determines whether or not the device is abnormal based on the calculated robust average.
- the device management apparatus According to the device management apparatus according to the tenth aspect, even when the operation of the device cannot be completely predicted in advance, the abnormality of the device is detected by comparing a number of similar devices with each other. be able to.
- the distributed device abnormality detection system even when the operation of the device cannot be predicted completely in advance, by comparing a plurality of similar devices with each other, It is possible to detect abnormalities.
- a system can be built at low cost without using high-performance information communication devices. Is possible. Therefore, while reducing the amount of communication when collecting battery information such as current, voltage, and temperature for a large number of devices, the amount of calculation can be reduced, and the distribution that can be realized with an inexpensive collection device and analysis processing device A type device abnormality detection system can be provided.
- 101 distributed device abnormality detection system 102 device, 102A display unit, 102B storage battery, 103, 103A, 103B device management device, 104 management server device, 104a antenna, 105 communication line, 106, 106m relay server device, 106a antenna, 141 Voltage measurement unit, 142, voltage determination signal generation unit, 161 measurement unit, 162 current measurement unit, 163 current determination signal generation unit, 164 data integration unit, 171 measurement unit, 200 railway vehicle, 201 railway train, 301 measurement unit, 302 equipment ID memory, 303 distribution information generation unit, 304 communication unit, 305 distribution comparison unit, 306 robust average calculation unit, 307 abnormality determination unit, 401 group allocation unit, 402 upper communication unit, 403 lower communication unit, 404 distribution integration unit, 4 5 distribution distributing unit, 406 statistics collection unit 407 threshold distribution unit, 501 Group assignment unit, 503 distribution integration unit, 504 threshold determining unit 600 the data center.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Physics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Automation & Control Theory (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Algebra (AREA)
- Operations Research (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Tests Of Electric Status Of Batteries (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
Description
前記複数の機器にそれぞれ接続され、前記各機器を管理する複数の機器管理装置と、
前記複数の機器管理装置と通信可能な管理サーバ装置とを備え、
前記各機器管理装置は、
前記管理サーバ装置との通信を行う第1の通信部と、
前記機器の物理量を繰り返し計測する計測部と、
前記計測された機器の物理量から、前記機器の分布情報を計算する分布情報生成部と、
前記分布情報生成部により生成された機器の分布情報と、前記第1の通信部を介して前記管理サーバ装置から配信された機器全体の統合分布情報との間の差異を計算する分布比較部と、
前記計算された分布情報間の差異に基づいて機器が異常か否かを判定する異常判定部とを備え、
前記管理サーバ装置は、
前記複数の機器管理装置との通信を行う第2の通信部と、
前記機器管理装置から前記第1の通信部を介して送信された機器毎の分布情報に基づいて、前記機器毎の分布情報を統合した統合分布情報を計算する第1の分布統合部とを備え、
前記機器管理装置は、前記生成された機器毎の分布情報を、前記第1の通信部を介して前記管理サーバ装置へ送信し、
前記管理サーバ装置は、前記計算された前記統合分布情報を、前記第2の通信部を介して前記機器管理装置に配信することを特徴とする。
図1は本発明の実施の形態1に係る分散型機器異常検出システム101の全体構成を示すブロック図である。実施の形態1に係る分散型機器異常検出システム101は、図1に示すように、実質的に同種である複数の機器102にそれぞれ接続されかつ各機器102毎に設けられた機器管理装置103と、システム全体を統括する管理サーバ装置104、中継サーバ装置106、並びに、機器管理装置103、中継サーバ装置106及び管理サーバ装置104を接続する通信回線105を備えて構成される。なお、実質的に同種である複数の機器102とは、複数の機器102が実質的に同様の動作を行って、正常又は異常などの状態値である物理量を有する。ここで、複数の機器102が実質的に同一の種類であるとは、異常が同様に起こりうる同様の構成を有して同様の動作を行う機器をいう。図1において、1段又は多段に設けられた中継サーバ装置106が複数の機器管理装置103及び下位の中継サーバ装置106を統括するが、中継サーバ装置106を省略し、機器管理装置103と管理サーバ装置104を直接に通信回線105によって接続してもよい。
本発明に係る実施の形態2は、実施の形態1に係る分散型機器異常検出システム101において、機器102の物理量ではなく機器102の入出力データ(入出力値)を定義および計測できる場合に適用する。すなわち、機器102の出力yが入力xに対する関数y=f(x)と表現できると見做せるとき、他の大多数の機器102と異なる関数fで表現されるような機器102を検出する。ただし、入力xは機器毎に多様であり、必ずしも同一の分布に従うとは限らないとする。
(1)図1に記載した構成と同様に中継サーバ装置106mが上位と下位が存在している。ここで、中継サーバ装置106mはアンテナ106aを有して、管理サーバ装置104mとの無線通信機能を有する。
(2)管理サーバ装置104mがデータセンターに設置されている。ここで、管理サーバ装置104mはアンテナ104aを有して、中継サーバ装置106mとの無線通信機能を有する。
(3)計測部151に代えて、図16の計測部161を備える。
本発明に係る実施の形態3は、実施の形態1に係る分散型機器異常検出システム101において、機器102で計測される物理量が有限集合ではない場合に適用する。このような機器102において、機器102の物理量xは例えば整数値、実数値あるいはベクトル値の確率変数で表される。この確率変数は機器102にて計測される内部の質量、長さ、時間、電流、温度、物質量、光度などの機器定数、あるいはそれらの機器定数の組み合わせによって表現される物理量であってよい。また、機器102の物理量xは、機器102の通信量、演算装置や記憶装置の占有率、ソフトウェア内部の遷移状態等を用いたものであってもよい。さらに、機器102の物理量xは、前記様々な物理量、状態等から計算される特徴量で置き換えたものであってもよい。特に、前記諸量を理論的に扱いやすい分布に従う確率変数として取り扱うために非線形な変換をかけた量であってもよい。実数値あるいは実ベクトル値の確率変数の分布同士の異なりは、実施の形態1と同様に、次式の相対エントロピーD(Pe,n|Pg,m)で測ることができる。
(1)機器管理装置103及び中継サーバ装置106が、分布情報Pe,n,Pg,mを統合して管理サーバ装置104又は上位の中継サーバ装置106に送信する代わりに、前記平均μe,nと分散Σe,nを統合して管理サーバ装置104又は上位の中継サーバ装置106に送信することと、
(2)分布比較部305において式(2)の代わりに式(14)を用いることを除いて、本発明第1の実施の形態に記載の分散型機器異常検出システム101と同様である。
本発明に係る実施の形態4は、実施の形態1に係る分散型機器異常検出システム101において、機器102の物理量ではなく機器102の入出力のデータ(入出力値)を計測でき、かつ前記入出力のデータ(入出力値)が有限集合ではない場合に適用する。この入出力のデータ(入出力値)は、例えば整数値、実数値あるいはベクトル値の確率変数で表される。この確率変数は、機器102の質量、長さ、時間、電流、温度、物質量、光度などの機器定数、あるいは当該機器定数の組み合わせによって表現される物理量であってよい。また、機器102の通信量、演算装置や記憶装置の占有率、ソフトウェア内部の状態遷移における状態等であってもよい。さらに、前記様々な物理量、状態等から計算される特徴量であってもよい。特に、前記諸量を理論的に扱いやすい分布に従う確率変数として取り扱うために非線形な変換をかけた量であってもよい。
第1の態様に係る分散型機器異常検出システムは、実質的に同一の種類の複数の機器の物理量を監視し、前記各機器の異常を検出するための分散型機器異常検出システムであって、
前記複数の機器にそれぞれ接続され、前記各機器を管理する複数の機器管理装置と、
前記複数の機器管理装置と通信可能な管理サーバ装置とを備え、
前記各機器管理装置は、
前記管理サーバ装置との通信を行う第1の通信部と、
前記機器の物理量を繰り返し計測する計測部と、
前記計測された機器の物理量から、前記機器の分布情報を計算する分布情報生成部と、
前記分布情報生成部により生成された機器の分布情報と、前記第1の通信部を介して前記管理サーバ装置から配信された機器全体の統合分布情報との間の差異を計算する分布比較部と、
前記計算された分布情報間の差異に基づいて機器が異常か否かを判定する異常判定部とを備え、
前記管理サーバ装置は、
前記複数の機器管理装置との通信を行う第2の通信部と、
前記機器管理装置から前記第1の通信部を介して送信された機器毎の分布情報に基づいて、前記機器毎の分布情報を統合した統合分布情報を計算する第1の分布統合部とを備え、
前記機器管理装置は、前記生成された機器毎の分布情報を、前記第1の通信部を介して前記管理サーバ装置へ送信し、
前記管理サーバ装置は、前記計算された前記統合分布情報を、前記第2の通信部を介して前記機器管理装置に配信することを特徴とする。
前記機器管理装置は、
前記機器に固有のIDを保持する機器IDメモリと、
前記複数の機器を分割して構成された複数のグループについて配信された各グループの分布情報に基づいて、分布情報間の差異のロバスト平均を計算するロバスト平均計算部をさらに備え、
前記機器IDメモリは前記機器のIDを、前記第1の通信部を介して前記管理サーバ装置へ送信し、
前記分布比較部は、前記第1の通信部を介して前記管理サーバ装置から配信された、少なくとも3つのグループの統合分布情報と、前記生成された機器の分布情報の差異をそれぞれ計算し、
前記異常判定部は、前記計算されたロバスト平均に基づいて前記機器が異常か否かを判定し、
前記管理サーバ装置は、
前記複数の機器の総数に基づいてグループ数を決定し、前記機器管理装置から前記第2の通信部を介して受信した機器のIDに基づいて自身の装置に接続する各機器が属するグループを決定する第2のグループ割当部をさらに備え、
前記第2の通信部は前記グループ数を前記機器管理装置に配信し、
前記第1の分布統合部は、前記機器管理装置から前記第2の通信部を介して受信した機器の分布情報をグループ毎に統合し、前記第2の通信部を介して前記機器管理装置へ配信することを特徴とする。
前記分布比較部は前記計算された分布情報間の差異を前記第1の通信部を介して前記管理サーバ装置に送信し、もしくは、前記ロバスト平均計算部は前記計算された分布情報間の差異に基づいてロバスト平均を計算して前記第1の通信部を介して前記管理サーバ装置に送信し、
前記異常判定部は、前記計算されたロバスト平均と、前記管理サーバ装置から前記第1の通信部を介して配信されたしきい値に基づいて、当該機器が異常状態か否かを判定し、
前記管理サーバ装置は、
前記複数の機器管理装置から送信された分布情報間の差異又は当該分布情報間の差異のロバスト平均の統計量を計算する統計情報収集部と、
前記機器管理装置から前記第2の通信部を介して受信した前記分布情報間の差異又はそのロバスト平均の統計量に基づいてしきい値を決定するしきい値決定部をさらに備え、
前記第2の通信部は、前記しきい値決定部によって決定されたしきい値を前記機器管理装置へ配信することを特徴とする。
前記計測部は、前記機器への入力値と前記機器からの出力値を繰り返し計測し、
前記分布情報生成部は前記計測された機器への入力値の分布と、前記機器への入力値と前記機器からの出力値の結合分布を生成し、
前記分布比較部は、前記生成された機器への入力値と前記機器からの出力値の結合分布から、ある入力に関する第1の条件付き分布を計算し、前記管理サーバ装置から前記第1の通信部を介して配信された統合分布情報から、前記入力に関する第2の条件付き分布を計算し、前記第1の条件付き分布と前記第2の条件付き分布との間の差異を計算し、前記第1の条件付き分布と前記第2の条件付き分布との間の差異の、前記分布情報生成部によって計算された機器への入力値の分布に関する期待値を、分布情報間の差異として計算することを特徴とする。
前記計測部は、前記機器の物理量又は前記機器の入出力値として整数値、実数値又はベクトル値を繰り返し計測し、
前記分布情報生成部は前記繰り返し計測された機器の物理量又は入出力の計測値の所定の統計値を計算して、前記計算された統計値に基づいて所定の理論分布のパラメータを推定し、
前記分布比較部は、前記推定された理論分布のパラメータと、前記第1の通信部を介して管理サーバ装置から配信された統合分布のパラメータから、当該分布情報間の差異を計算し、
前記第1の分布統合部は、前記機器管理装置から受信した前記機器の物理量又は入出力の計測値の所定の統計値を統合し、前記統合された統計値に基づいて所定の統合分布のパラメータを推定することを特徴とする。
前記複数の機器管理装置又は下位の中継サーバ装置とそれぞれ通信可能であってかつ、前記管理サーバ装置又は上位の中継サーバ装置と通信可能な少なくとも1つの中継サーバ装置をさらに備え、
前記中継サーバ装置は、
前記複数の機器管理装置又は前記下位の中継サーバ装置との通信を行う下位通信部と、
前記管理サーバ装置又は前記上位の中継サーバ装置との通信を行う上位通信部と、
前記機器管理装置から前記下位通信部を介して受信した前記機器のIDと、前記管理サーバ装置又は前記上位の中継サーバ装置から前記上位通信部を介して配信されたグループ数に基づいて自身の装置に接続する各機器が属するグループを決定する第1のグループ割当部と、
前記機器管理装置又は前記下位の中継サーバ装置から前記下位通信部を介して受信した機器の分布情報をグループ毎に統合するとともに、前記管理サーバ装置又は前記上位の中継サーバ装置から前記上位通信部を介して配信されたグループ毎の統合分布情報を前記下位通信部を介して前記機器管理装置又は前記下位の中継サーバ装置に配信する第2の分布統合部と、
前記上位通信部を介して管理サーバ装置又は上位の中継サーバ装置から配信された前記統合分布情報もしくは前記統合分布情報のパラメータを、前記下位通信部を介して前記機器管理装置又は前記下位の中継サーバ装置に配信する分布配信部と、
前記複数の機器管理装置から送信された分布情報間の差異又は当該分布情報間の差異のロバスト平均の統計量を計算する統計情報収集部と、
前記管理サーバ装置又は上位のサーバ装置から前記上位通信部を介して配信されたしきい値を、前記下位通信部を介して前記機器管理装置又は下位の中継サーバ装置へさらに配信するしきい値配信部とを備えたことを特徴とする。
実質的に同一の種類の複数の機器の物理量を監視し、前記各機器の異常を検出するための分散型機器異常検出システムの機器管理装置であり、前記複数の機器にそれぞれ接続され、前記各機器を管理する複数の機器管理装置であって、
前記複数の機器を分割して構成された少なくとも3つのグループの各機器の物理量についての分布情報を各グループ毎に統合してなる統合分布情報と、前記各機器の分布情報との差異とをそれぞれ計算する分布比較部と、
前記各グループの分布情報に基づいて、分布情報間の差異のロバスト平均を計算するロバスト平均計算部と、
前記計算されたロバスト平均に基づいて前記機器が異常か否かを判定する異常判定部とを備えたことを特徴とする。
Claims (10)
- 実質的に同一の種類の複数の機器の物理量を監視し、前記各機器の異常を検出するための分散型機器異常検出システムであって、
前記複数の機器にそれぞれ接続され、前記各機器を管理する複数の機器管理装置と、
前記複数の機器管理装置と通信可能な管理サーバ装置とを備え、
前記各機器管理装置は、
前記管理サーバ装置との通信を行う第1の通信部と、
前記機器の物理量を繰り返し計測する計測部と、
前記計測された機器の物理量から、前記機器の分布情報を計算する分布情報生成部と、
前記分布情報生成部により生成された機器の分布情報と、前記第1の通信部を介して前記管理サーバ装置から配信された機器全体の統合分布情報との間の差異を計算する分布比較部と、
前記計算された分布情報間の差異に基づいて機器が異常か否かを判定する異常判定部とを備え、
前記管理サーバ装置は、
前記複数の機器管理装置との通信を行う第2の通信部と、
前記機器管理装置から前記第1の通信部を介して送信された機器毎の分布情報に基づいて、前記機器毎の分布情報を統合した統合分布情報を計算する第1の分布統合部とを備え、
前記機器管理装置は、前記生成された機器毎の分布情報を、前記第1の通信部を介して前記管理サーバ装置へ送信し、
前記管理サーバ装置は、前記計算された前記統合分布情報を、前記第2の通信部を介して前記機器管理装置に配信することを特徴とする分散型機器異常検出システム。 - 前記機器管理装置は、
前記機器に固有のIDを保持する機器IDメモリと、
前記複数の機器を分割して構成された複数のグループについて配信された各グループの分布情報に基づいて、分布情報間の差異のロバスト平均を計算するロバスト平均計算部をさらに備え、
前記機器IDメモリは前記機器のIDを、前記第1の通信部を介して前記管理サーバ装置へ送信し、
前記分布比較部は、前記第1の通信部を介して前記管理サーバ装置から配信された、少なくとも3つのグループの統合分布情報と、前記生成された機器の分布情報の差異をそれぞれ計算し、
前記異常判定部は、前記計算されたロバスト平均に基づいて前記機器が異常か否かを判定し、
前記管理サーバ装置は、
前記複数の機器の総数に基づいてグループ数を決定し、前記機器管理装置から前記第2の通信部を介して受信した機器のIDに基づいて自身の装置に接続する各機器が属するグループを決定する第2のグループ割当部をさらに備え、
前記第2の通信部は前記グループ数を前記機器管理装置に配信し、
前記第1の分布統合部は、前記機器管理装置から前記第2の通信部を介して受信した機器の分布情報をグループ毎に統合し、前記第2の通信部を介して前記機器管理装置へ配信することを特徴とする請求項1記載の分散型機器異常検出システム。 - 前記分布比較部は前記計算された分布情報間の差異を前記第1の通信部を介して前記管理サーバ装置に送信し、もしくは、前記ロバスト平均計算部は前記計算された分布情報間の差異に基づいてロバスト平均を計算して前記第1の通信部を介して前記管理サーバ装置に送信し、
前記異常判定部は、前記計算されたロバスト平均と、前記管理サーバ装置から前記第1の通信部を介して配信されたしきい値に基づいて、当該機器が異常状態か否かを判定し、
前記管理サーバ装置は、
前記複数の機器管理装置から送信された分布情報間の差異又は当該分布情報間の差異のロバスト平均の統計量を計算する統計情報収集部と、
前記機器管理装置から前記第2の通信部を介して受信した前記分布情報間の差異又はそのロバスト平均の統計量に基づいてしきい値を決定するしきい値決定部をさらに備え、
前記第2の通信部は、前記しきい値決定部によって決定されたしきい値を前記機器管理装置へ配信することを特徴とする請求項2記載の分散型機器異常検出システム。 - 前記計測部は、前記機器への入力値と前記機器からの出力値を繰り返し計測し、
前記分布情報生成部は前記計測された機器への入力値の分布と、前記機器への入力値と前記機器からの出力値の結合分布を生成し、
前記分布比較部は、前記生成された機器への入力値と前記機器からの出力値の結合分布から、ある入力に関する第1の条件付き分布を計算し、前記管理サーバ装置から前記第1の通信部を介して配信された統合分布情報から、前記入力に関する第2の条件付き分布を計算し、前記第1の条件付き分布と前記第2の条件付き分布との間の差異を計算し、前記第1の条件付き分布と前記第2の条件付き分布との間の差異の、前記分布情報生成部によって計算された機器への入力値の分布に関する期待値を、分布情報間の差異として計算することを特徴とする請求項2記載の分散型機器異常検出システム。 - 前記計測部は、前記機器の物理量又は前記機器の入出力値として整数値、実数値又はベクトル値を繰り返し計測し、
前記分布情報生成部は前記繰り返し計測された機器の物理量又は入出力の計測値の所定の統計値を計算して、前記計算された統計値に基づいて所定の理論分布のパラメータを推定し、
前記分布比較部は、前記推定された理論分布のパラメータと、前記第1の通信部を介して管理サーバ装置から配信された統合分布のパラメータから、当該分布情報間の差異を計算し、
前記第1の分布統合部は、前記機器管理装置から受信した前記機器の物理量又は入出力の計測値の所定の統計値を統合し、前記統合された統計値に基づいて所定の統合分布のパラメータを推定することを特徴とする請求項2~4のうちのいずれか1つに記載の分散型機器異常検出システム。 - 前記所定の統計値は、平均、分散、モーメント、最大、最小、中央値、あるいはそれらの組み合わせであることを特徴とする請求項5記載の分散型機器異常検出システム。
- 前記機器の物理量又は前記機器の入出力の計測値は、前記機器を計測して得られる整数値、実数値あるいはベクトル値に対し、線形あるいは非線形な変換を行った結果として得られる整数値、実数値あるいはベクトル値であることを特徴とする請求項5又は6記載の分散型機器異常検出システム。
- 前記複数の機器管理装置又は下位の中継サーバ装置とそれぞれ通信可能であってかつ、前記管理サーバ装置又は上位の中継サーバ装置と通信可能な少なくとも1つの中継サーバ装置をさらに備え、
前記中継サーバ装置は、
前記複数の機器管理装置又は前記下位の中継サーバ装置との通信を行う下位通信部と、
前記管理サーバ装置又は前記上位の中継サーバ装置との通信を行う上位通信部と、
前記機器管理装置から前記下位通信部を介して受信した前記機器のIDと、前記管理サーバ装置又は前記上位の中継サーバ装置から前記上位通信部を介して配信されたグループ数に基づいて自身の装置に接続する各機器が属するグループを決定する第1のグループ割当部と、
前記機器管理装置又は前記下位の中継サーバ装置から前記下位通信部を介して受信した機器の分布情報をグループ毎に統合するとともに、前記管理サーバ装置又は前記上位の中継サーバ装置から前記上位通信部を介して配信されたグループ毎の統合分布情報を前記下位通信部を介して前記機器管理装置又は前記下位の中継サーバ装置に配信する第2の分布統合部と、
前記上位通信部を介して管理サーバ装置又は上位の中継サーバ装置から配信された前記統合分布情報もしくは前記統合分布情報のパラメータを、前記下位通信部を介して前記機器管理装置又は前記下位の中継サーバ装置に配信する分布配信部と、
前記複数の機器管理装置から送信された分布情報間の差異又は当該分布情報間の差異のロバスト平均の統計量を計算する統計情報収集部と、
前記管理サーバ装置又は上位のサーバ装置から前記上位通信部を介して配信されたしきい値を、前記下位通信部を介して前記機器管理装置又は下位の中継サーバ装置へさらに配信するしきい値配信部とを備えたことを特徴とする請求項1記載の分散型機器異常検出システム。 - 前記各機器は、電力変換装置、蓄電池、ブレーキ装置、又はモータであることを特徴とする請求項1~8のうちのいずれか1つに記載の分散型機器異常検出システム。
- 実質的に同一の種類の複数の機器の物理量を監視し、前記各機器の異常を検出するための分散型機器異常検出システムの機器管理装置であり、前記複数の機器にそれぞれ接続され、前記各機器を管理する複数の機器管理装置であって、
前記複数の機器を分割して構成された少なくとも3つのグループの各機器の物理量についての分布情報を各グループ毎に統合してなる統合分布情報と、前記各機器の分布情報との差異とをそれぞれ計算する分布比較部と、
前記各グループの分布情報に基づいて、分布情報間の差異のロバスト平均を計算するロバスト平均計算部と、
前記計算されたロバスト平均に基づいて前記機器が異常か否かを判定する異常判定部とを備えたことを特徴とする機器管理装置。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2017526301A JP6371478B2 (ja) | 2015-07-01 | 2016-06-21 | 分散型機器異常検出システム |
US15/572,170 US10783219B2 (en) | 2015-07-01 | 2016-06-21 | Distributed equipment abnormality detection system for monitoring physical amounts of equipments and detecting abnormality of each equipment |
DE112016002999.3T DE112016002999T5 (de) | 2015-07-01 | 2016-06-21 | Anomalie-detektionssystem für verteilte ausrüstungen |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2015-132636 | 2015-07-01 | ||
JP2015132636 | 2015-07-01 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2017002673A1 true WO2017002673A1 (ja) | 2017-01-05 |
Family
ID=57608070
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2016/068411 WO2017002673A1 (ja) | 2015-07-01 | 2016-06-21 | 分散型機器異常検出システム |
Country Status (4)
Country | Link |
---|---|
US (1) | US10783219B2 (ja) |
JP (2) | JP6371478B2 (ja) |
DE (1) | DE112016002999T5 (ja) |
WO (1) | WO2017002673A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018205960A (ja) * | 2017-06-01 | 2018-12-27 | 株式会社東芝 | 監視制御システム、分散サーバ及び監視制御方法 |
JP2020046964A (ja) * | 2018-09-19 | 2020-03-26 | ファナック株式会社 | 特性判定装置、特性判定方法及び特性判定プログラム |
CN112262353A (zh) * | 2018-06-11 | 2021-01-22 | 松下知识产权经营株式会社 | 异常解析装置、制造***、异常解析方法以及程序 |
KR102654240B1 (ko) * | 2022-10-26 | 2024-04-04 | 주식회사 에이아이네이션 | 딥러닝, 머신러닝 및 통계 모델을 이용한 산업설비 이상 감지 장치 및 시스템 |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108829978A (zh) * | 2018-06-20 | 2018-11-16 | 哈尔滨工业大学 | 基于统计特征评估与高斯过程回归的直流电源在线异常检测方法 |
JP7309335B2 (ja) | 2018-09-19 | 2023-07-18 | シャープ株式会社 | 端末装置、基地局装置、および、通信方法 |
JP7226103B2 (ja) * | 2019-05-29 | 2023-02-21 | オムロン株式会社 | 制御システム、サポート装置およびサポートプログラム |
DE102020108326A1 (de) | 2020-03-26 | 2021-09-30 | Audi Aktiengesellschaft | Verfahren zur Erkennung einer Fremdzelle und/oder eines Fremdmoduls in einer Batterievorrichtung eines Fahrzeugs sowie Batterievorrichtung und Kraftfahrzeug mit einer Batterievorrichtung |
CN113884889B (zh) * | 2021-10-29 | 2024-04-26 | 章鱼博士智能技术(上海)有限公司 | 一种电池安全预警的方法、装置、存储介质及电子设备 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04168506A (ja) * | 1990-11-01 | 1992-06-16 | Toshiba Corp | 設備監視診断システム |
JPH04336606A (ja) * | 1991-05-14 | 1992-11-24 | Toshiba Corp | プロセス制御装置 |
JPH09128034A (ja) * | 1995-11-06 | 1997-05-16 | Fujitsu Ltd | 監視システムと方法 |
JP2004030491A (ja) * | 2002-06-28 | 2004-01-29 | Hitachi Industries Co Ltd | 機器メンテナンス診断システム |
US20140343895A1 (en) * | 2012-02-01 | 2014-11-20 | Abb Research Ltd | Monitoring of primary devices in a power system |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3930171B2 (ja) | 1998-12-03 | 2007-06-13 | 株式会社日本自動車部品総合研究所 | 組電池の監視装置 |
JP3649092B2 (ja) | 2000-06-21 | 2005-05-18 | 日産自動車株式会社 | 組電池異常検出装置 |
JP2003264906A (ja) | 2002-03-11 | 2003-09-19 | Sanyo Electric Co Ltd | 自動車用電池管理システム |
JP2009076346A (ja) | 2007-09-21 | 2009-04-09 | Panasonic Corp | 車両用蓄電池の状態表示システム |
JP2009223416A (ja) * | 2008-03-13 | 2009-10-01 | Omron Corp | リモート装置及び監視システム |
JP5343168B2 (ja) | 2010-06-24 | 2013-11-13 | パナソニック株式会社 | 電池の劣化度を取得するための方法及びそのシステム |
JP2013064649A (ja) | 2011-09-16 | 2013-04-11 | Toshiba Corp | 電池システムの故障検出システム |
JP5992186B2 (ja) | 2012-03-16 | 2016-09-14 | 株式会社東芝 | 二次電池装置および二次電池装置の異常検出方法 |
JP5914669B2 (ja) * | 2012-08-31 | 2016-05-11 | 株式会社日立製作所 | サービス性能監視方法 |
KR102233812B1 (ko) * | 2014-07-30 | 2021-03-31 | 삼성전자주식회사 | 제조 공정에 있어서 설비로부터의 데이터를 처리하는 방법 및 시스템 |
JP6488197B2 (ja) * | 2015-05-29 | 2019-03-20 | 株式会社日立製作所 | 異常検出方法、異常検出装置、及びネットワークシステム |
-
2016
- 2016-06-21 JP JP2017526301A patent/JP6371478B2/ja active Active
- 2016-06-21 WO PCT/JP2016/068411 patent/WO2017002673A1/ja active Application Filing
- 2016-06-21 DE DE112016002999.3T patent/DE112016002999T5/de active Pending
- 2016-06-21 US US15/572,170 patent/US10783219B2/en active Active
-
2018
- 2018-07-12 JP JP2018132421A patent/JP6584598B2/ja active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04168506A (ja) * | 1990-11-01 | 1992-06-16 | Toshiba Corp | 設備監視診断システム |
JPH04336606A (ja) * | 1991-05-14 | 1992-11-24 | Toshiba Corp | プロセス制御装置 |
JPH09128034A (ja) * | 1995-11-06 | 1997-05-16 | Fujitsu Ltd | 監視システムと方法 |
JP2004030491A (ja) * | 2002-06-28 | 2004-01-29 | Hitachi Industries Co Ltd | 機器メンテナンス診断システム |
US20140343895A1 (en) * | 2012-02-01 | 2014-11-20 | Abb Research Ltd | Monitoring of primary devices in a power system |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018205960A (ja) * | 2017-06-01 | 2018-12-27 | 株式会社東芝 | 監視制御システム、分散サーバ及び監視制御方法 |
CN112262353A (zh) * | 2018-06-11 | 2021-01-22 | 松下知识产权经营株式会社 | 异常解析装置、制造***、异常解析方法以及程序 |
JP2020046964A (ja) * | 2018-09-19 | 2020-03-26 | ファナック株式会社 | 特性判定装置、特性判定方法及び特性判定プログラム |
KR102654240B1 (ko) * | 2022-10-26 | 2024-04-04 | 주식회사 에이아이네이션 | 딥러닝, 머신러닝 및 통계 모델을 이용한 산업설비 이상 감지 장치 및 시스템 |
Also Published As
Publication number | Publication date |
---|---|
JP6584598B2 (ja) | 2019-10-02 |
JP2018174702A (ja) | 2018-11-08 |
DE112016002999T5 (de) | 2018-03-15 |
JP6371478B2 (ja) | 2018-08-08 |
US20180143942A1 (en) | 2018-05-24 |
US10783219B2 (en) | 2020-09-22 |
JPWO2017002673A1 (ja) | 2018-02-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6584598B2 (ja) | 機器管理装置、管理サーバ、及び分散型機器異常検出方法 | |
JP6313502B2 (ja) | 蓄電池評価装置、蓄電システム、蓄電池評価方法およびコンピュータプログラム | |
US10243923B2 (en) | Method and system for allocating communication ID of battery management module | |
US11221367B2 (en) | Evaluation device, energy storage system, evaluation method and non-transitory computer readable medium | |
CN110740061B (zh) | 故障预警方法、装置及计算机存储介质 | |
CN105234097B (zh) | 基于大数据的用于电池梯次利用的电动汽车动力电池管理***及方法 | |
CN102801552A (zh) | 基于网络建模的用于错误隔离和错误减轻的***和方法 | |
JP6431644B2 (ja) | 蓄電池評価装置、蓄電システムおよび蓄電池評価方法 | |
CN105872061B (zh) | 一种服务器集群管理方法、装置及*** | |
WO2018172216A1 (en) | Battery energy store | |
CN107707612B (zh) | 一种负载均衡集群的资源利用率的评估方法及装置 | |
CN105897599A (zh) | 通信装置、通信***以及通信方法 | |
WO2022142013A1 (zh) | 基于人工智能的ab测试方法、装置、计算机设备及介质 | |
WO2018012151A1 (ja) | 情報処理装置、制御方法、及びプログラム | |
CN109905268A (zh) | 网络运维的方法及装置 | |
CN108183814A (zh) | 用电信息采集***的通信信道的故障排查方法和装置 | |
Camci et al. | Sampling based State of Health estimation methodology for Li-ion batteries | |
CN108248641A (zh) | 一种城市轨道交通数据处理方法及装置 | |
CN105678456B (zh) | 一种电能计量装置运行状态自动评估方法及其*** | |
CN116014901A (zh) | 基于数字孪生技术实现输变电设备工作状态监控方法 | |
CN114764600A (zh) | 新能源车电池包的寿命预测方法、装置和电子设备 | |
US9229898B2 (en) | Causation isolation using a configuration item metric identified based on event classification | |
CN115001147B (zh) | 一种光伏发电数据采集方法、***及云平台 | |
EP4400852A1 (en) | Battery condition monitoring | |
KR102646874B1 (ko) | 배터리 모니터링 데이터 관리 장치 및 그 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16817781 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2017526301 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15572170 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 112016002999 Country of ref document: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 16817781 Country of ref document: EP Kind code of ref document: A1 |