CN107977626A - The group technology of a kind of electronic equipment operational data - Google Patents
The group technology of a kind of electronic equipment operational data Download PDFInfo
- Publication number
- CN107977626A CN107977626A CN201711244439.9A CN201711244439A CN107977626A CN 107977626 A CN107977626 A CN 107977626A CN 201711244439 A CN201711244439 A CN 201711244439A CN 107977626 A CN107977626 A CN 107977626A
- Authority
- CN
- China
- Prior art keywords
- group
- grounding
- data
- sequence number
- grounding data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Debugging And Monitoring (AREA)
Abstract
The group technology of a kind of electronic equipment operational data, its method first gathers the sample magnitude of different operating index in electronic equipment with fixed frequency, feature selecting is carried out to the sample magnitude of different operating index, obtain with electronic equipment operating status most close to working index, by training data based on the sample magnitude of the working index, its packet cycle is calculated, is grouped in chronological order;Then pass through the cycle where the characteristic value failure judgement of each grounding data, grounding data are grouped with the cycle of guilty culprit, it is grouped calculating two classes by nonlinear state assessment algorithm, obtains fault threshold, realizes the general fault detection to different type electronic equipment.
Description
Technical field
The present invention relates to malfunction monitoring technical field, more particularly to the group technology of a kind of electronic equipment operational data.
Background technology
So far, the maintenance mode of equipment has three kinds, first, periodic maintenance, such a maintaining method it is of high cost, it is necessary to from
Line overhauls;Second, being safeguarded after failure, this mode is to cause damage or the loss of other biggers in equipment, belongs to correction maintenance;
Third, when equipment is run monitoring device some characteristic quantities, to determine equipment state (good, failure).
Obviously operation when monitoring device there is greater advantage, also effectively reduced while maintenance cost is relatively low maintenance when
Between and the equipment damage caused by failure;But existing online monitoring alarm algorithm poor universality now, is not suitable for difference
The equipment of type, simulated failure test is of high cost, and fault sample is difficult to obtain, and cannot meet diversified demand, for example, training
When fault sample be difficult obtain, there are imbalance problem for sample;Failure mode is very much, it is difficult to limit;Data volume is big, fault location
It is difficult.
The content of the invention
For above-mentioned technical problem, the shortcomings of the prior art is overcome, the present invention provides a kind of electronic equipment work
The group technology of data, realizes the good classification to failure group and normal data group.
Specifically, the method that a kind of electronic equipment standard working index is set, comprises the following steps,
Obtain the grounding data for being used for training generation Electronics Standards working index;Wherein, the grounding
Data include the sample magnitude of the electronic equipment operation work sampled with fixed frequency;Work belonging to the sample magnitude
It is identical with the working index belonging to the standard working index to make index;
Packet cycle is calculated according to the grounding data, and with the packet cycle to the grounding data
It is grouped, and each group of sequence number is determined according to time sequencing;
By calculating the temporal signatures value of each grounding data group, judge whether the grounding data group belongs to event
Barrier group, and record the group sequence number of failure group;
According to the group sequence number of failure group, from the grounding extracting data training sample group after packet;The trained sample
The group sequence number for the grounding data group that this group includes is continuous, and is not admitted to failure group;
According to the training sample group, calculate generation and be used to judge standard of the electronic equipment operation with the presence or absence of failure
Working index.
As a further improvement, described calculate packet cycle according to the grounding data, and with the packet week
Phase is grouped the grounding data, and determines that each group of sequence number comprises the following steps according to time sequencing,
The sequence temporal signatures value of the grounding data is subjected to Fourier transformation, obtains the grounding data
Corresponding intensity spectrum;
The frequency component of amplitude maximum is screened from the intensity spectrum, by the inverse of the frequency component of the amplitude maximum
As packet cycle.
As a further improvement, the temporal signatures value by calculating each grounding data group, judges the basis
Whether training data group belongs to failure group, and the group sequence number for recording failure group comprises the following steps, and calculates each group of basis instruction
Practice the variance and average of data, as the physical characteristic values of each group of grounding data, it is inclined that record falls into physical characteristic values
If the group sequence number of the sequence number record of the grounding group of poor scope is recorded under standard, the corresponding basis of this group of sequence number is judged
Training data group belongs to guilty culprit group.
As a further improvement, it is described according to the training sample group, calculate generation and be used to judge the electronic equipment fortune
Row comprises the following steps with the presence or absence of the standard working index of failure, according to the group sequence number of guilty culprit group, in chronological order will
Grounding data group after packet is divided into training sample group and test sample group;Wherein, the training sample group includes
Each grounding data group is not admitted to guilty culprit group;The test sample group includes at least one grounding data group
It is guilty culprit group;
According to nonlinear state assessment algorithm, calculating acquisition is carried out to each grounding data group in training sample group
Fault threshold;
According to the fault threshold, judge to be determined to have the grounding data group of failure in the test sample group
Group sequence number it is whether consistent with the group sequence number of record;
If so, then refer to using the fault threshold as the judgement electronic equipment operation with the presence or absence of the standard work of failure
Mark.
As a further improvement, it is described according to nonlinear state assessment algorithm, each basis in training sample group is instructed
Practice data group to carry out calculating acquisition fault threshold, specifically include:
The data of any instant in the test sample group are observation vector;
Extract the history observation vector in several described training sample groups;
Will several described history observation vector constructive memory matrixes;
The observation vector is inputted to the memory square and exports to obtain predicted vector;
Calculate each observation vector and corresponding pre- direction finding in addition to the observation vector at the guilty culprit group moment
The difference of amount, determines that difference maximum in residing difference is the fault threshold.
As a further improvement, the observation vector and the relational expression of the predicted vector are
Wherein, yestFor the predicted vector, yestFor the observation vector, D is the dot-blur pattern.
Brief description of the drawings
In order to illustrate more clearly of technical scheme, attached drawing needed in embodiment will be made below
Simply introduce, it should be apparent that, drawings in the following description are only some embodiments of the present invention, general for this area
For logical technical staff, without creative efforts, other attached drawings can also be obtained according to these attached drawings.
Fig. 1 is first embodiment of the invention overall flow schematic diagram;
Fig. 2 is the C phase output voltage intensity spectrum schematic diagrames of UPS three phase mains in first embodiment of the invention;
Fig. 3 is the fault threshold schematic diagram of UPS three phase mains in first embodiment of the invention.
Fig. 4 is second embodiment of the invention overall structure diagram.
Embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work
Embodiment, belongs to the scope of protection of the invention.
The present invention provides various embodiments, specifically please refers to Fig.1 (S10-S50 process steps), and Fig. 1 is the of the present invention
A kind of embodiment, it comprises the following steps,
S10, with the sample magnitude of each working index of the operation work of fixed frequency sampling electronic equipment, and to every
All sample magnitudes that one working index samples form the corresponding time series of the working index;
UPS three phase mains makes solution as electronic equipment to working index, sample magnitude, time series in the present embodiment
Release;UPS three phase mains includes some working indexs, including cell voltage, incoming frequency, A phase input voltages, B phase input voltages,
C phase input voltages, sensor number, A phase output currents, B phase output currents, C phase output currents, output frequency, A phases export negative
Carry, B phase output loadings, C phase output loadings, output state (normal 0/ abnormal 1), A phase output voltages, B phase output voltages, C phases
Output voltage, city's electricity condition (normal 0/ failure 1)) and sample state mark (normal, alert);Respectively by this several work
Index is acquired with the fixed frequency of sampling in every 10 minutes once;By the sample magnitude of each working index, in chronological order
Just arrangement form time series data.
S20, carries out feature selecting to the corresponding time series of each working index, therefrom determines and the electronic equipment
The working index of operating status degree of correlation maximum and the degree of correlation maximum functional index series temporal signatures value, and to determine
The corresponding sample magnitude of the maximum working index be basic training data;
The above-mentioned feature selecting mode referred to, can use the conventional data algorithms such as sequence backward selection algorithm;
Specifically, it is described to the corresponding time series of each working index carry out feature selecting, therefrom determine with it is described
The working index of electronic equipment operating status degree of correlation maximum, including:Extract the corresponding time series of each working index
Whole sequence temporal signatures values, is merged into the feature complete or collected works of time series by characteristic value;Using sequence backward selection algorithm pair
The feature complete or collected works of time series carry out feature selecting;Bring the sequence temporal signatures value extracted after feature selecting into evaluation
Function, proposes that the sequence temporal signatures of classifying quality difference are worth to optimal time series characteristic value;By the optimal feature
It is worth corresponding working index and is determined as working index with the electronic equipment operating status degree of correlation maximum.In UPS three-phase electricities
In the example in source, " C phases input voltage " is optimal characteristics, i.e., the degree of correlation between C phases input voltage and ups power state is most
Greatly;In other words, when ups power breaks down, similarly there is exception in C phase input voltages;Ups power normal operation, C phases input
Voltage similarly normal operation, therefore the work of whole ups power can be reacted by detecting the working status of C phase input voltages
State whether failure.
Specific state calculates packet cycle according to the grounding data, including:By the grounding data into
Row Fourier transformation, obtains the corresponding intensity spectrum of the grounding data;Amplitude maximum is screened from the intensity spectrum
Frequency component, using the inverse of the frequency component of the amplitude maximum as packet cycle.It is incorporated in showing for UPS three phase mains
Example, as shown in Fig. 2, after the time series of the C phases input voltage is carried out Fourier transformation, its maximum frequency component exists
At f=1.16e-0.5Hz, the inverse of its maximum frequency component for 23.9532 it is small when, be approximately equal to 24 it is small when, so obtaining UPS tri-
When the packet cycle of phase power supply is 24 small, i.e., it is equal to a packet cycle within one day.
S30, packet cycle is calculated according to the temporal aspect gauge of the grounding data, and with the packet cycle pair
The grounding data are grouped, and each group of sequence number is determined according to time sequencing;By calculating each grounding
The sequence temporal signatures value of data group, judges whether the grounding data group belongs to guilty culprit group, and records guilty culprit
The group sequence number of group;
The variance and average of each group of grounding data are calculated, the physics as each group of grounding data is special
Value indicative, record fall into the sequence number of the grounding group of physical characteristic values deviation range;Physical characteristic values deviation model is fallen into by described
The grounding data judging enclosed belongs to guilty culprit group for the corresponding grounding data group of this group of sequence number.
With reference to the example of UPS three phase mains, to a cycle group, i.e., the time series of intraday C phases output voltage into
The calculating of row average and variance, determines the cycle of guilty culprit;Assuming that the C phases output voltage in UPS three phase mains is carried out
The monitoring of 81 days, broke down at the 13rd day;In other words it is exactly within the 13rd day guilty culprit group, required by physical characteristic values here
The data taken are the specific period in the cycle, for example, failure be occur the 13rd inaction interval 8 points 20 minutes, but in the cycle
The specific period time interval it is too short, cause packet excessive;Therefore temporally level shows, such as the history of tracking failure occurs
Time, is first shown failure day (inaction interval), then the exact failure period being shown in inaction interval.
S40, according to the group sequence number of guilty culprit group, is divided into instruction by the grounding data group after packet in chronological order
Practice sample group and test sample group;Wherein, each grounding data group that the training sample group includes is not admitted to failure
Place group;It is guilty culprit group that the test sample group, which includes at least one grounding data group,;
With reference to the example of UPS three phase mains, it is assumed that the 13rd day is failure day (the 13rd group of cycle is guilty culprit group), in base
Plinth training data component group is with the 30th day to demarcate, and first 30 days are test sample group, and latter 51 days are training sample group;Here it is worth
It is noted that the class interval the 30th day mentioned in embodiment is only that example is not intended to limit those skilled in the art's partition testing
The selection of sample group, grounding data group.
S50, according to nonlinear state assessment algorithm, calculates each grounding data group in training sample group
Obtain fault threshold;
According to the fault threshold, judge to be determined to have the grounding data group of failure in the test sample group
Group sequence number it is whether consistent with the group sequence number of record;
If so, then refer to using the fault threshold as the judgement electronic equipment operation with the presence or absence of the standard work of failure
Mark.
" according to nonlinear state assessment algorithm, calculating is carried out to each grounding data group in training sample group and is obtained
Obtain fault threshold ", specifically include, the data of any instant in the test sample group are observation vector;Extract several institutes
State the history observation vector in training sample group;Will several described history observation vector building process dot-blur patterns;By described in
Observation vector inputs to the memory square and exports to obtain predicted vector;Calculate in addition to the observation vector at the guilty culprit group moment
The difference of each observation vector and corresponding predicted vector, determines that difference maximum in residing difference is the failure threshold
Value;Here it is worth noting that remove guilty culprit group moment observation vector each observation vector be UPS three phase mains just
Situation about often working, in a normal operating situation, selects its least normal difference, will be least normal under normal operative condition
Difference is fault threshold as failure criterion.
As shown in figure 3, with reference to the example of UPS three phase mains, by the spy in the training sample group of latter 51 days per continuous three days
Value indicative does average treatment and forms 17 groups of characteristic values for calculating similarity, by the use of first 30 days as test sample group, carries out non-linear
Status assessment, obtains fault threshold as 300, and determine guilty culprit periodic groups (my god) be the 13rd day.
The observation vector and the relational expression of the predicted vector areIts
In, yestFor the predicted vector, yobsFor the observation vector, D is the process dot-blur pattern;For convenience of the relation of understanding
Formula, the present embodiment are further described the formula reasoning,
It is mutually related variables assuming that a certain process or equipment share n, if at a time i, n variable observing
Observation vector is denoted as, i.e.,
X (i)=[x1,x2,...,xn]T
The construction of procedure dot-blur pattern is the first of Nonlinear State Estimate Technology modelings
A step.M history observation vector is gathered, anabolic process process dot-blur pattern is
Each row observation vector in procedure dot-blur pattern represents a normal operating conditions of equipment.By reasonable
The subspace (being represented with D) that m history observation vector in the procedure dot-blur pattern of selection is turned into can represent process
Or the whole dynamic process of equipment normal operation.Therefore, the construction essence of procedure dot-blur pattern is exactly to process or equipment
The learning process of normal operation characteristic.
The input of NSET is a certain etching process or the observation vector y of equipment0bs, the output of model is to the pre- of the input
Direction finding amount yest.The residual error for outputting and inputting predicted vector for constructing the model is
R=yobs-yest
Minimization is carried out to residual error, i.e.,
Observation vector y can be then inputted to obtain to any one0bsGenerating the weight vector that a m is tieed up is
W=(DTD)-1DTyobs
So that
yest=D (DTD)-1DTyobs
Practical problem often has " non-linear ", in order to characterize " similarity degree " between vector by DTD and DTyobsIn multiplication fortune
It is changed to Accorded with for nonlinear operation, for substituting the multiplying in ordinary channel computing.Here Euler's distance is often taken:
I.e. final result is:
As shown in figure 4, Fig. 4 is second embodiment of the invention, there is provided a kind of general machine learning based on data mining
Device, including sampling unit, the sampling unit are referred to each work of the operation work of fixed frequency sampling electronic equipment
Target sample magnitude, and the corresponding sequential of the working index is formed to all sample magnitudes that each working index samples
Sequence;
Feature selection unit, the feature selection unit are used to carry out the corresponding time series of each working index special
Sign selection, therefrom determines and the working index of the electronic equipment operating status degree of correlation maximum and the degree of correlation maximum work
Make the sequence temporal signatures value of index, and the training number based on the definite corresponding sample magnitude of the maximum working index
According to;
Packet cycle unit, the packet cycle unit calculate point according to the temporal aspect gauge of the grounding data
The group cycle, and the grounding data are grouped with the packet cycle, and determine each group according to time sequencing
Sequence number;
Guilty culprit group judging unit, the guilty culprit group judging unit is by calculating each grounding data group
Physical characteristic values, judge whether the grounding data group belongs to guilty culprit group, and record the group sequence number of guilty culprit group;
Test training grouped element, group sequence number of the grouped element according to guilty culprit group is trained in the test, temporally suitable
Grounding data group after packet is divided into training sample group and test sample group by sequence;Wherein, the training sample group bag
The each grounding data group included is not admitted to guilty culprit group;The test sample group includes at least one grounding number
It is guilty culprit group according to group;
Fault threshold computing unit, the fault threshold computing unit is used for according to nonlinear state assessment algorithm, to instruction
Practice each grounding data group in sample group to carry out calculating acquisition fault threshold;
According to the fault threshold, judge to be determined to have the grounding data group of failure in the test sample group
Group sequence number it is whether consistent with the group sequence number of record;
If so, then refer to using the fault threshold as the judgement electronic equipment operation with the presence or absence of the standard work of failure
Mark.
The sampling unit is further used for, and extracts the corresponding sequence temporal signatures value of each working index, will be all
Sequence temporal signatures value be merged into the feature complete or collected works of time series;Feature using sequence backward selection algorithm to time series
Complete or collected works carry out feature selecting;Bring the sequence temporal signatures value extracted after feature selecting into evaluation function, obtain optimal
Sequence temporal signatures value;The optimal corresponding working index of temporal signatures value is determined as running with the electronic equipment
The working index of state degree of correlation maximum.
The packet cycle unit is further used for,
The sequence temporal signatures value of the grounding data is subjected to Fourier transformation, obtains the grounding data
Corresponding intensity spectrum;
The frequency component of amplitude maximum is screened from the intensity spectrum, by the inverse of the frequency component of the amplitude maximum
As packet cycle.
The guilty culprit group judging unit is further used for,
The variance and average of each group of grounding data are calculated, the physics as each group of grounding data is special
Value indicative, if record fall into physical characteristic values deviation range grounding group sequence number record group sequence number be recorded in standard it
Under, then judge that the corresponding grounding data group of this group of sequence number belongs to guilty culprit group.
The fault threshold computing unit is further used for calculating every in addition to the observation vector at the guilty culprit group moment
The difference of one observation vector and corresponding predicted vector, determines that difference maximum in residing difference is the failure threshold
Value;
The data of any instant in the test sample group are observation vector;
Extract the history observation vector in several described training sample groups;
Will several described history observation vector building process dot-blur patterns;
The observation vector is inputted to the memory square and exports to obtain predicted vector;
The observation vector and the relational expression of the predicted vector areIts
In, yestFor the predicted vector, yestFor the observation vector, D is the process dot-blur pattern.
Third embodiment of the invention additionally provides a kind of general machine learning system based on data mining, the embodiment
Learning system include, processor, memory and be stored in the memory and be configured as being performed by the processor
Computer program, the processor performs the computer program, such as realizes the program of Multi-screen display system;
Exemplary, the computer program can be divided into one or more modules, one or more of moulds
Block is stored in the memory, and is performed by the processor, to complete the present embodiment.One or more of modules can
To be the series of computation machine programmed instruction section that can complete specific function, which is used to describing the computer program existing
Implementation procedure in the control method terminal device of Multi-screen display system.
The learning system can be the computing devices such as desktop PC, notebook, palm PC and cloud server.
The learning system may include, but be not limited only to, processor, memory, display.Those skilled in the art can be with
Understand, the schematic diagram is only the example of learning system, does not form the restriction to learning system, can be included than illustrating more
More or less components, either combines some components or different components, such as learning system can also include input and output
Equipment, network access equipment, bus etc..
Alleged processor can be central processing unit (Central Processing Unit, CPU), can also be it
His general processor, digital signal processor (Digital Signal Processor, DSP), application-specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor can also be any conventional processor
Deng the processor is the control centre of learning system, utilizes various interfaces and each portion of the whole learning system of connection
Point.
The memory can be used for storing the computer program and/or module, and the processor is by running or performing
The computer program and/or module being stored in the memory, and the data being stored in memory are called, realize study
The various functions of system.The memory can mainly include storing program area and storage data field, wherein, storing program area can deposit
Application program (such as sound-playing function, text conversion function etc.) needed for storage operating system, at least one function etc.;Storage
Data field can be stored uses created data (such as voice data, text message data etc.) etc. according to mobile phone.In addition, deposit
Reservoir can include high-speed random access memory, can also include nonvolatile memory, such as hard disk, memory, plug-in type
Hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card
(Flash Card), at least one disk memory, flush memory device or other volatile solid-state parts.
Wherein, if the module that learning system integrates is realized in the form of SFU software functional unit and is used as independent product pin
Sell or in use, can be stored in a computer read/write memory medium.Based on such understanding, the present invention realizes above-mentioned
All or part of flow in embodiment method, can also instruct relevant hardware to complete by computer program, described
Computer program can be stored in a computer-readable recording medium, which, can when being executed by processor
The step of realizing above-mentioned each embodiment of the method.Wherein, the computer program includes computer program code, the computer
Program code can be source code form, object identification code form, executable file or some intermediate forms etc..The computer can
Reading medium can include:Any entity or device of the computer program code, recording medium, USB flash disk, mobile hard can be carried
Disk, magnetic disc, CD, computer storage, read-only storage (ROM, Read-Only Memory), random access memory
(RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..Need what is illustrated
It is that the content that the computer-readable medium includes can be fitted according to legislation in jurisdiction and the requirement of patent practice
When increase and decrease, such as in some jurisdictions, according to legislation and patent practice, computer-readable medium, which does not include electric carrier wave, to be believed
Number and telecommunication signal.
It should be noted that device embodiment described above is only schematical, wherein described be used as separating component
The unit of explanation may or may not be physically separate, can be as the component that unit is shown or can also
It is not physical location, you can with positioned at a place, or can also be distributed in multiple network unit.Can be according to reality
Need to select some or all of module therein to realize the purpose of this embodiment scheme.In addition, device provided by the invention
In embodiment attached drawing, the connection relation between module represents there is communication connection between them, specifically can be implemented as one or
A plurality of communication bus or signal wire.Those of ordinary skill in the art are without creative efforts, you can to understand
And implement.
The above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also considered as
Protection scope of the present invention.
Claims (6)
1. the group technology of a kind of electronic equipment operational data, it is characterised in that comprise the following steps:
Obtain the grounding data for being used for training generation Electronics Standards working index;Wherein, the grounding data
Sample magnitude including the electronic equipment operation work sampled with fixed frequency;Work belonging to the sample magnitude refers to
Mark is identical with the working index belonging to the standard working index;
Packet cycle is calculated according to the grounding data, and the grounding data are carried out with the packet cycle
It is grouped, and each group of sequence number is determined according to time sequencing;
By calculating the temporal signatures value of each grounding data group, judge whether the grounding data group belongs to failure
Group, and record the group sequence number of failure group;
According to the group sequence number of failure group, from the grounding extracting data training sample group after packet;The training sample group
Including the group sequence number of grounding data group be continuous, and be not admitted to failure group;
According to the training sample group, calculate generation and be used to judge that the electronic equipment operation works with the presence or absence of the standard of failure
Index.
2. according to the method described in claim 1, it is characterized in that, described calculate packet week according to the grounding data
Phase, and the grounding data are grouped with the packet cycle, and each group of sequence number is determined according to time sequencing
Comprise the following steps,
The sequence temporal signatures value of the grounding data is subjected to Fourier transformation, the grounding data is obtained and corresponds to
Intensity spectrum;
From the intensity spectrum screen amplitude maximum frequency component, using the frequency component of the amplitude maximum it is reciprocal as
Packet cycle.
3. the according to the method described in claim 1, it is characterized in that, time domain by calculating each grounding data group
Characteristic value, judges whether the grounding data group belongs to failure group, and the group sequence number for recording failure group comprises the following steps, meter
The variance and average of each group of grounding data are calculated, as the physical characteristic values of each group of grounding data, record
If the group sequence number for falling into the sequence number record of the grounding group of physical characteristic values deviation range is recorded under standard, judging should
The corresponding grounding data group of group sequence number belongs to guilty culprit group.
4. according to the method described in claim 1, it is characterized in that, described be used for according to the training sample group, calculating generation
Judge that the electronic equipment operation comprises the following steps with the presence or absence of the standard working index of failure, according to the group of guilty culprit group
Sequence number, is divided into training sample group and test sample group by the grounding data group after packet in chronological order;Wherein, it is described
Each grounding data group that training sample group includes is not admitted to guilty culprit group;The test sample group includes at least one
A grounding data group is guilty culprit group;
According to nonlinear state assessment algorithm, each grounding data group in training sample group is carried out to calculate acquisition failure
Threshold value;
According to the fault threshold, judge to be determined to have the group of the grounding data group of failure in the test sample group
Whether whether sequence number consistent with the group sequence number of record;
If so, it then whether there is the standard working index of failure using the fault threshold as the judgement electronic equipment operation.
5. according to the method described in claim 4, it is characterized in that, described according to nonlinear state assessment algorithm, to training sample
Each grounding data group in this group, which calculate, obtains fault threshold, specifically includes:
The data of any instant in the test sample group are observation vector;
Extract the history observation vector in several described training sample groups;
Will several described history observation vector constructive memory matrixes;
The observation vector is inputted to the memory square and exports to obtain predicted vector;
Calculate each observation vector and corresponding predicted vector in addition to the observation vector at the guilty culprit group moment
Difference, determines that difference maximum in residing difference is the fault threshold.
6. according to the method described in claim 5, it is characterized in that, the observation vector and the relationship expression of the predicted vector
Formula isWherein, yestFor the predicted vector, yestFor the observation vector, D is described
Dot-blur pattern.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711244439.9A CN107977626B (en) | 2017-11-30 | 2017-11-30 | Grouping method for electronic equipment working data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711244439.9A CN107977626B (en) | 2017-11-30 | 2017-11-30 | Grouping method for electronic equipment working data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107977626A true CN107977626A (en) | 2018-05-01 |
CN107977626B CN107977626B (en) | 2020-09-15 |
Family
ID=62008841
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711244439.9A Active CN107977626B (en) | 2017-11-30 | 2017-11-30 | Grouping method for electronic equipment working data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107977626B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635873A (en) * | 2018-12-19 | 2019-04-16 | 佛山科学技术学院 | A kind of UPS failure prediction method |
CN113945788A (en) * | 2021-10-14 | 2022-01-18 | 深圳市杰普特光电股份有限公司 | Detection method, detection device, detection equipment, electronic equipment and readable storage medium |
CN116774109A (en) * | 2023-06-26 | 2023-09-19 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103529825A (en) * | 2013-10-23 | 2014-01-22 | 上海白丁电子科技有限公司 | Automatic equipment failure analysis and diagnosis method and device thereof |
US20150177030A1 (en) * | 2013-12-19 | 2015-06-25 | Uchicago Argonne, Llc | Transient multivariable sensor evaluation |
-
2017
- 2017-11-30 CN CN201711244439.9A patent/CN107977626B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103529825A (en) * | 2013-10-23 | 2014-01-22 | 上海白丁电子科技有限公司 | Automatic equipment failure analysis and diagnosis method and device thereof |
US20150177030A1 (en) * | 2013-12-19 | 2015-06-25 | Uchicago Argonne, Llc | Transient multivariable sensor evaluation |
Non-Patent Citations (2)
Title |
---|
常澍平等: "非线性状态估计(NSET)建模方法在故障预警***中的应用", 《软件》 * |
李新丽: "大型风力发电机组的故障诊断研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635873A (en) * | 2018-12-19 | 2019-04-16 | 佛山科学技术学院 | A kind of UPS failure prediction method |
CN113945788A (en) * | 2021-10-14 | 2022-01-18 | 深圳市杰普特光电股份有限公司 | Detection method, detection device, detection equipment, electronic equipment and readable storage medium |
CN113945788B (en) * | 2021-10-14 | 2024-01-30 | 深圳市杰普特光电股份有限公司 | Detection method, detection device, detection apparatus, electronic apparatus, and readable storage medium |
CN116774109A (en) * | 2023-06-26 | 2023-09-19 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
CN116774109B (en) * | 2023-06-26 | 2024-01-30 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
Also Published As
Publication number | Publication date |
---|---|
CN107977626B (en) | 2020-09-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107944721A (en) | A kind of general machine learning method based on data mining, device and system | |
CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
Brahma et al. | Real-time identification of dynamic events in power systems using PMU data, and potential applications—models, promises, and challenges | |
CN108009063A (en) | The method of a kind of electronic equipment fault threshold detection | |
US11656263B2 (en) | Effective feature set-based high impedance fault detection | |
CN108008176A (en) | A kind of photovoltaic array real-time state monitoring and fault location system | |
CN107977626A (en) | The group technology of a kind of electronic equipment operational data | |
Ezzat et al. | Microgrids islanding detection using Fourier transform and machine learning algorithm | |
CN108009582A (en) | The method that a kind of electronic equipment standard working index is set | |
CN112461289A (en) | Ring main unit fault monitoring method, system, terminal and storage medium | |
CN111738348B (en) | Power data anomaly detection method and device | |
CN109755937A (en) | A kind of regional power grid inertia calculation method and apparatus based on measurement | |
CN110824297B (en) | Single-phase earth fault discrimination method and device based on SVM (support vector machine) | |
CN109613324A (en) | A kind of detection method and device of Harmonics amplification | |
CN111612149A (en) | Main network line state detection method, system and medium based on decision tree | |
CN114184870A (en) | Non-invasive load identification method and equipment | |
CN112398226A (en) | Power supply system electricity stealing prevention method, system, terminal and storage medium | |
CN115420988B (en) | Method, device, equipment and storage medium for identifying abnormal electricity consumption user | |
CN114389241B (en) | Relay protection setting value setting method and device for active power distribution network | |
CN115146715A (en) | Power utilization potential safety hazard diagnosis method, device, equipment and storage medium | |
CN108233379A (en) | Test method, device, equipment and storage medium for safety and stability control device | |
CN111915451A (en) | Method for calculating daily power curve of transformer area | |
CN113484573B (en) | Abnormal electricity utilization monitoring method based on energy data analysis | |
CN118091520B (en) | Intelligent regulation and control method and system for 10us square wave surge testing equipment | |
CN117200449B (en) | Multi-dimensional algorithm analysis-based power grid monitoring management method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |