CN113323699B - Method for accurately identifying fault source of hydraulic support system based on data driving - Google Patents
Method for accurately identifying fault source of hydraulic support system based on data driving Download PDFInfo
- Publication number
- CN113323699B CN113323699B CN202110636227.5A CN202110636227A CN113323699B CN 113323699 B CN113323699 B CN 113323699B CN 202110636227 A CN202110636227 A CN 202110636227A CN 113323699 B CN113323699 B CN 113323699B
- Authority
- CN
- China
- Prior art keywords
- hydraulic support
- support system
- working parameters
- fault
- statistic
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000002159 abnormal effect Effects 0.000 claims abstract description 42
- 238000003745 diagnosis Methods 0.000 claims abstract description 30
- 208000024891 symptom Diseases 0.000 claims abstract description 13
- 238000007418 data mining Methods 0.000 claims abstract description 8
- 239000000839 emulsion Substances 0.000 claims description 33
- 238000011897 real-time detection Methods 0.000 claims description 22
- 239000013598 vector Substances 0.000 claims description 21
- 239000011159 matrix material Substances 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 15
- 230000036541 health Effects 0.000 claims description 14
- 238000001514 detection method Methods 0.000 claims description 13
- 230000009471 action Effects 0.000 claims description 10
- 238000005065 mining Methods 0.000 claims description 7
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 4
- 239000007788 liquid Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 2
- 230000008878 coupling Effects 0.000 abstract description 5
- 238000010168 coupling process Methods 0.000 abstract description 5
- 238000005859 coupling reaction Methods 0.000 abstract description 5
- 238000012423 maintenance Methods 0.000 abstract description 3
- 239000003245 coal Substances 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000001364 causal effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
- E21D15/00—Props; Chocks, e.g. made of flexible containers filled with backfilling material
- E21D15/14—Telescopic props
- E21D15/44—Hydraulic, pneumatic, or hydraulic-pneumatic props
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
- E21D15/00—Props; Chocks, e.g. made of flexible containers filled with backfilling material
- E21D15/14—Telescopic props
- E21D15/46—Telescopic props with load-measuring devices; with alarm devices
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21F—SAFETY DEVICES, TRANSPORT, FILLING-UP, RESCUE, VENTILATION, OR DRAINING IN OR OF MINES OR TUNNELS
- E21F17/00—Methods or devices for use in mines or tunnels, not covered elsewhere
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21F—SAFETY DEVICES, TRANSPORT, FILLING-UP, RESCUE, VENTILATION, OR DRAINING IN OR OF MINES OR TUNNELS
- E21F17/00—Methods or devices for use in mines or tunnels, not covered elsewhere
- E21F17/18—Special adaptations of signalling or alarm devices
Landscapes
- Engineering & Computer Science (AREA)
- Mining & Mineral Resources (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Geology (AREA)
- Mechanical Engineering (AREA)
- Structural Engineering (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a method for accurately identifying a fault source of a hydraulic support system based on data driving, which comprises the steps of firstly adopting an abnormal data mining method when the hydraulic support system has a fault, accurately separating abnormal working parameters, then inputting the abnormal working parameters into a Bayesian network by using the abnormal working parameters as symptom nodes, accurately searching a real fault source causing the fault through decoupling operation of the Bayesian network, effectively decoupling the coupling relation of different faults of the hydraulic support system on the abnormal symptoms, obtaining more accurate diagnosis results and providing technical guidance for maintenance and management of the hydraulic support system.
Description
Technical Field
The invention belongs to the technical field of hydraulic support system fault identification, and particularly relates to a method for accurately identifying a hydraulic support system fault source based on data driving.
Background
In coal mining, a hydraulic support system is used for supporting a top plate, pushing a scraper conveyor and the like, and is core equipment for comprehensive mechanized mining of mines. One coal mining working face usually has tens or hundreds of hydraulic supports, and any hydraulic support breaks down to cause the working face to stop production and shut down, and economic loss is huge. Meanwhile, when the hydraulic support breaks down due to insufficient supporting force, rocks on the top plate of the coal seam can fall off, casualties and equipment damage are caused, especially, in many mines in China, the falling rocks fall off along the inclined working face, and the damage degree can be further increased. Reliable operation of the hydraulic support system is a necessary prerequisite for safe coal mining.
The hydraulic support system fault diagnosis and identification technology is an effective means for achieving the aim. The technology can accurately search the real fault reason causing the abnormity according to the abnormal symptoms shown by the hydraulic support system, thereby providing a basis for the maintenance of the hydraulic support system. In view of this problem, researchers have intensively studied. However, most of the existing hydraulic support system fault diagnosis and identification technologies are developed for a single fault type, and only the influence of one fault on the behavior state of the hydraulic support is considered. The hydraulic support system has a complex structure and a plurality of potential failure modes, and different failure types have the same expression on certain abnormal signs of the hydraulic support. This single failure diagnosis method cannot deal with this problem, and is liable to cause misdiagnosis or missed diagnosis of the failure. And researchers also put forward a method for analyzing the coupling relation between various faults and abnormal symptoms of the hydraulic support system based on a fault tree method, so that the decoupling diagnosis of the multiple faults is realized. Although the fault tree method can intuitively express the coupling relation among multiple faults, the method cannot realize the diagnosis reasoning from top to bottom, and therefore, the method cannot be independently applied to the diagnosis of the faults of the hydraulic support system.
Through the literature retrieval of the prior art, a public document 'hydraulic support fault diagnosis and prediction research' (major paper of Chinese mining university, 2020.05) provides a method for accurately identifying a fault source of a hydraulic support system based on data driving, and the public document comprises the following steps: "study on failure diagnosis of hydraulic mount. Firstly, determining monitoring indexes of hydraulic support equipment, acquiring running data when four types of faults occur, reducing dimensions, arranging the running data into a training set and a test set, and constructing a classification model of a support vector machine in MATLAB based on a libsvm-3.23 toolkit; then, aiming at the complexity and relevance among fault reasons, organizing the history records of the fault reasons of the hydraulic support into an Excel table, carrying out structure learning and parameter learning of a Bayesian network, and expressing the uncertainty relation among the fault reasons by using probability values; secondly, the classification result based on the support vector machine is used as a known evidence and is input into a Bayesian network for Bayesian network reasoning; finally, by way of example, the feasibility of the fault diagnosis model is verified. The disadvantages are as follows: the established Bayesian network diagnosis model of the hydraulic support system only contains fault events and does not contain abnormal working parameter information of the hydraulic support system, so that the fault reason obtained by each inference of the model is fixed, and the fault source causing the abnormality cannot be really searched according to the abnormal symptom actually shown by the hydraulic support system.
Disclosure of Invention
In order to solve the problems, the invention provides a method for accurately identifying a fault source of a hydraulic support system based on data driving. In order to achieve the technical purpose, the invention adopts the following technical scheme:
a method for accurately identifying a fault source of a hydraulic support system based on data driving further comprises the following steps of carrying out abnormal reason data mining on working parameters when the hydraulic support system is in fault, and determining the fault source by combining a Bayesian network diagnosis model, wherein the method comprises the following steps:
step 1: selecting working parameters of various hydraulic support systems as sample data;
step 2: the hydraulic support system fault definition comprises the following steps:
an off-line training stage:
step 2-11, collecting sample data of working parameters of the hydraulic support system in a healthy state;
step 2-12, preprocessing sample data of working parameters of the hydraulic support system in the healthy state, and recording the sample data as the standardized working parameters in the healthy state;
step 2-13, calculating Hotelling's T of working parameters in the normalized health state2The threshold values of the statistic and the Q statistic are respectively recorded as
And QUCL；
And (3) an online detection stage:
step 2-21: detecting data of working parameters of the hydraulic support system in real time;
step 2-22: preprocessing the data of the working parameters of the real-time detection hydraulic support system, and recording the data as standardized real-time detection working parameters;
step 2-23: building Hotelling's T in principal component subspace2And constructing Q statistic by statistic and residual subspace, and respectively recording the Q statistic as T2And Q;
step 2-24: judging whether the hydraulic support system has faults or not, and adopting an satisfied formula
Or (Q > Q)UCL) When the hydraulic support system fails, detecting working parameters in real time after corresponding standardization, and performing abnormal reason data mining;
and 3, step 3: the method for mining the data of the reasons of the abnormal working parameters comprises the following steps:
step 3-1: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The threshold value of the statistic contribution degree and the threshold value of the contribution degree to the Q statistic are respectively recorded as
And
step 3-2: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The statistic contribution degree and the contribution degree to the Q statistic are respectively recorded as
And
step 3-3: separating abnormal working parameters of the hydraulic support system; when in use
Or
Judging abnormal working parameters of the hydraulic support system;
and 4, step 4: establishing a Bayesian network diagnosis model for hydraulic support system fault diagnosis, wherein the Bayesian network diagnosis model consists of two types of nodes, namely a fault and a symptom, and the two types of nodes are connected through directed edges according to a fault mechanism and characteristics of a hydraulic support system to form a topological structure; each fault node is a fault type which causes the hydraulic support system to work abnormally; each symptom node is a plurality of types of working parameters of the hydraulic support system; and (3) inputting the separated abnormal working parameters of the hydraulic support system serving as observation evidences into the Bayesian network diagnosis model, outputting fault types causing the abnormal working of the hydraulic support system, and confirming fault sources.
Further, in step 2-12, sample data of working parameters of the hydraulic support system in the healthy state is preprocessed and recorded as working parameters in the standardized healthy state
The formula used is as follows:
wherein, Xi' is the working parameter of the hydraulic support system in a healthy state;
is Xi' average value; sXiIs Xi' standard deviation; different values of i represent different kinds of operating parameters, i is 1,2,3,4,5,6,7 … n; n is the total number of categories of operating parameters.
Further, in steps 2-13, Hotelling's T of the normalized health state operating parameter is calculated2The threshold values of the statistic and the Q statistic are respectively recorded as
And QUCL(ii) a The method comprises the following steps:
step 2-131: and (3) calculating a covariance matrix S of the standardized working parameters, wherein the calculation method comprises the following steps:
wherein m is the number of sampling groups of the working parameters in the standardized health state; x*A sample data matrix of the working parameters in a standardized health state;
step 2-132: calculating the eigenvalue and the eigenvector of the covariance matrix S, and arranging the eigenvalue and the eigenvector in a descending order according to the magnitude of the eigenvalue; wherein the characteristic value is recorded as lambdai(ii) a The feature vector is denoted as pi；i＝1,2,3,4,5,6,7…n；
Step 2-133: calculating the eigenvalue lambdaiVariance contribution rate CONTiAnd the cumulative variance contribution rate CPV (l) of the first characteristic value, wherein the calculation method comprises the following steps:
wherein,
is the sum of n eigenvalues, j ═ 1,2,3,4,5,6,7 … n;l≤n；
step 2-134: setting a threshold value CPV (cumulative variance contribution rate) according to a detection requirement standardUCLThrough the formula CPV (k) > CPVUCLDetermining a boundary value k;
step 2-135: determining Hotelling's T2Threshold value of statistic and threshold value of Q statistic
And QUCLThe calculation formula is as follows:
wherein, Fα(k, m-k) represents the critical value of the F distribution with two degrees of freedom of k and m-k at the significance level α; c. CαIs a critical value of standard normal distribution under the significance level alpha; the value of the significance level alpha is selected according to the detection requirement;
wherein, thetaiAnd h0The calculation method comprises the following steps:
wherein, thetaiWherein i is 1,2,3,
is the j-th eigenvalue to the power i, i 1,2,3, j 1,2,3,4,5,6,7 … n.
Further, in steps 2-23, Hotelling's T is constructed in the principal component subspace2And constructing Q statistic by statistic and residual subspace, and respectively recording the Q statistic as T2And Q, comprising the steps of:
step 2-231: according to the k values obtained in the steps 2-134, the feature vector p is processediIs divided into
And
step 2-232: wherein use is made of
Projecting the standardized real-time detection working parameters to a principal component subspace for the basis vectors, and calculating T2(ii) a By using
Projecting the standardized real-time detection working parameters to a residual subspace for the basis vectors, and calculating Q; the calculation formula is as follows:
wherein Λ is a diagonal matrix formed by the first k eigenvalues of the covariance matrix S;
detecting the column vectors of the working parameters in real time after a group of standardization;
the standardized real-time detection working parameters are expressed by the following formula:
wherein,
detecting working parameters in real time after standardization; x2iWorking parameters of the hydraulic support system are detected in real time; different values of i represent different kinds of operating parameters, i is 1,2,3,4,5,6,7 … n; n is the total number of categories of operating parameters.
Further, in step 3-1, the Hotelling's T pair is detected in real time after the standardization of the hydraulic support system when the hydraulic support system has faults is calculated2The threshold value of the statistic contribution degree and the threshold value of the contribution degree to the Q statistic are respectively recorded as
And
the formula used is as follows:
wherein,
a critical value of chi-square distribution with a degree of freedom of 1 under a significance level alpha; the value of the significance level alpha is selected according to the detection requirement; xiiIs a column vector with the ith element being 1 and the remaining elements being 0.
Further, in step 3-2, the Hotelling's T pair is detected in real time after the standardization of the hydraulic support system when the hydraulic support system has a fault is calculated2The statistic contribution degree and the contribution degree to the Q statistic are respectively marked as
And
the formula used is as follows:
wherein xi isiIs a column vector with the ith element being 1 and the remaining elements being 0.
Further, in step 1, working parameters of various types of hydraulic support systems are selected as sample data, wherein the working parameters comprise pump outlet emulsion pressure, pump outlet emulsion temperature, emulsion pressure after a filter, front support emulsion pressure, support moving displacement deviation, bottom plate specific pressure, top beam inclination angle and electromagnetic valve driving end current.
Further, in step 4, the fault types causing the abnormal operation of the hydraulic support system include: the method comprises the following steps of emulsion pump failure, filter blockage, no action/slow action of a stand column, no action/slow action of a jack, pipeline leakage, safety valve failure, electromagnetic valve failure, insufficient emulsion and emulsion internal liquid mixing.
The invention has the following beneficial effects:
through the implementation of the scheme, the Bayesian network is an effective intelligent pattern recognition algorithm; the algorithm can effectively express qualitative and quantitative coupling relations among the multiple variables by utilizing the directed acyclic graph and probability information, and then realizes decoupling operation on the multiple variables by defining a message transmission process on the directed acyclic graph; the method for accurately identifying the fault source of the hydraulic support system based on data driving combines abnormal data mining with a Bayesian network, adopts an abnormal data mining method when the hydraulic support system has a fault, accurately separates abnormal working parameters, inputs the abnormal working parameters into a Bayesian network diagnosis model, accurately searches a real fault source causing the fault, effectively decouples the coupling relation of different faults of the hydraulic support system on abnormal symptoms, has more accurate diagnosis results, and provides technical guidance for maintenance and management of the hydraulic support system.
Drawings
FIG. 1 is a flow chart of a method for accurately identifying a fault source of a hydraulic support system based on data driving according to the present invention;
fig. 2 is a topological structure of a bayesian network diagnosis model for hydraulic support system fault diagnosis in the present invention.
Detailed Description
Specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but it should be understood that the scope of the present invention is not limited to the specific embodiments.
As shown in fig. 1, the method for accurately identifying the fault source of the hydraulic support system based on data driving is used for performing abnormal cause data mining on working parameters when the hydraulic support system has a fault and determining the fault source by combining a bayesian network diagnosis model, and comprises the following steps;
step 1: selecting working parameters of various hydraulic support systems as sample data; preferably, the following 8 working parameters of the hydraulic support system are selected as the collection objects, namely, the pump outlet emulsion pressure X1Emulsion temperature X of pump outlet2Pressure X of emulsion after filter3Front strut emulsion pressure X4Displacement deviation X of moving frame5Specific pressure of the bottom plate X6Top beam inclination angle X7Current X at drive end of electromagnetic valve8。
Step 2: the hydraulic support system fault definition comprises the following steps:
an off-line training stage:
step 2-11, collecting sample data of working parameters of the hydraulic support system in a healthy state, and recording the sample data as Xi', different values of i represent different kinds of operating parameters, i ═ 1,2,3,4,5,6,7 … n; n is the total number of the kinds of the working parameters, namely n is 8, and is respectively the pump outlet emulsion pressure X1' and pump outlet emulsion temperature X2', emulsion pressure after filter X3' front pillar emulsion pressure X4' shift frame displacement deviation X5' bottom plateSpecific pressure X6' Top beam inclination angle X7', current X at drive end of electromagnetic valve8′。
Step 2-12: preprocessing the sample data of the working parameters to enable the sample data to fall into a small specific interval so as to eliminate the influence on the fault detection result due to different working parameter dimensions; removing unit limit of the sample data, converting the sample data into a dimensionless pure numerical value, and recording the dimensionless pure numerical value as a working parameter in a standardized health state; the formula used is as follows:
wherein,
working parameters under the standardized health state;
is Xi' average values of
SXiIs Xi' standard deviations of { S } respectivelyX1,SX2,SX3,SX4,SX5,SX6,SX7,SX8}。
Step 2-13: hotelling's T for calculating normalized working parameters2The threshold value of the statistic and the threshold value of the Q statistic are respectively recorded as
And QUCL(ii) a The method comprises the following steps:
step 2-131: and (3) calculating a covariance matrix S of the working parameters in the standardized health state, wherein the calculation method comprises the following steps:
wherein m is the number of sampling groups of the working parameters in the standardized health state; x*The row of the sample data matrix of the working parameters in the standardized health state represents the values of the working parameters in the standardized health state at the same moment, and the column of the sample data matrix represents the values of the working parameters at m different moments;
step 2-132: calculating the eigenvalue and the eigenvector of the covariance matrix S, and arranging the eigenvalue and the eigenvector in a descending order according to the magnitude of the eigenvalue; wherein the characteristic value is recorded as lambdai(ii) a The feature vector is denoted as pi；i＝1,2,3,4,5,6,7,8。
Step 2-133: calculating the eigenvalue lambdaiVariance contribution rate CONTiAnd the cumulative variance contribution rate CPV (l) of the first characteristic value, wherein the calculation method comprises the following steps:
wherein,
is the sum of 8 eigenvalues; j is 1,2,3,4,5,6,7, 8; l is less than or equal to 8;
step 2-134: setting a threshold value CPV (cumulative variance contribution rate) according to a detection requirement standardUCLThrough the formula CPV (k) > CPVUCLDetermining a boundary value k;
step 2-135: determining Hotelling's T2Threshold value of statistic and threshold value of Q statistic
And QUCLThe calculation formula is as follows:
wherein, Fα(k, m-k) represents the critical value of the F distribution with two degrees of freedom of k and m-k at the significance level α, and the numerical value can be obtained by looking up a table; c. CαThe standard normal distribution is a critical value under the significance level alpha, and the numerical value can be obtained by looking up a table; the value of the significance level alpha is selected according to the detection requirement;
wherein, thetaiAnd h0The calculation method comprises the following steps:
wherein, thetaiWherein i is 1,2,3,
is the j-th eigenvalue to the power i, i 1,2,3, j 1,2,3,4,5,6,7, 8.
And (3) an online detection stage:
step 2-21: detecting the working parameter data of the hydraulic support system in the step 1 in real time, and recording the working parameter of the hydraulic support system detected in real time as X2iDifferent values of i represent different kinds of working parameters, i is 1,2,3,4,5,6,7 … n; n is the total number of the kinds of the working parameters, i.e. n is 8, and is respectively the pump outlet emulsion pressure X21Emulsion temperature X of pump outlet22Pressure X of emulsion after filter23Front strut emulsion pressure X24Displacement deviation X of moving frame25Specific pressure of the bottom plate X26Top beam inclination angle X27Current X at drive end of electromagnetic valve28。
Step 2-22: preprocessing the data of the working parameters of the real-time detection hydraulic support system, recording the data as standardized real-time detection working parameters, and using a formula as follows:
wherein,
detecting working parameters in real time after standardization; x2iWorking parameters of the hydraulic support system are detected in real time; i is 1,2,3,4,5,6,7, 8.
Step 2-23: constructing Hotelling's T in principal component subspace2And constructing Q statistic by statistic and residual subspace, and respectively recording the Q statistic as T2And Q, comprising the steps of:
step 2-231: according to the k values obtained in the steps 2-134, the feature vector p is processediIs divided into
And
step 2-232: wherein use is made of
Projecting the standardized real-time detection working parameters to a principal component subspace for the basis vectors, and calculating T2(ii) a By using
Projecting the standardized real-time detection working parameters to a residual subspace for the basis vectors, and calculating Q; the correlation is used for quantitatively describing the working parameters of the hydraulic support system; t is2And Q is calculated as:
wherein Λ is a diagonal matrix formed by the first k eigenvalues of the covariance matrix S, and the calculation formula is:
Step 2-24: judging whether the hydraulic support system has faults or not, and adopting an satisfied formula
Or (Q > Q)UCL) And then, judging that the hydraulic support system has a fault, and detecting working parameters in real time after corresponding standardization at the moment and carrying out abnormal reason data mining.
And step 3: the method for mining the data of the reasons of the abnormal working parameters comprises the following steps:
step 3-1: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The threshold value of the statistic contribution degree and the threshold value of the contribution degree to the Q statistic are respectively recorded as
And
the formula used is as follows:
wherein,
a critical value of chi-square distribution with a degree of freedom of 1 under the significance level alpha; the value of the significance level alpha is selected according to the detection requirement; xiiIs a column vector with the ith element being 1 and the remaining elements being 0.
Step 3-2: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The statistic contribution degree and the contribution degree to the Q statistic are respectively recorded as
And
the formula used is as follows:
wherein ξiIs a column vector with the ith element being 1 and the remaining elements being 0.
Step 3-3: separating abnormal working parameters of the hydraulic support system; when in use
Or
And judging the abnormal working parameters of the hydraulic support system.
And 4, step 4: constructing a Bayesian network diagnosis model for hydraulic support system fault diagnosis as shown in FIG. 2; the Bayesian network is a directed acyclic graph, wherein nodes represent random variables, directed edges among the nodes represent causal relationships among the random variables, the directed edges point to effects by 'causes', and the causal association strength is characterized by conditional probability.
In the invention, the Bayesian network diagnosis model consists of two types of nodes of faults and symptoms, wherein each fault node is marked as f and is a fault type causing abnormal working of a hydraulic support system; the types of faults include: emulsion pump failure f1Filter clogging f2No motion/slow motion of the column f3No action/slow action of jack f4Leakage f in the pipeline5Safety valve failure f6Failure of solenoid valve f7And a deficiency of emulsion f8And the internal liquid of the emulsion f9(ii) a Each symptom node is marked as X and is the various working parameters of the hydraulic support system selected in the step 1; the working parameters comprise: emulsion pressure X at pump outlet1Emulsion temperature X of pump outlet2Pressure X of emulsion after filter3Front strut emulsion pressure X4Displacement deviation X of moving frame5Specific pressure of the bottom plate X6Top beam inclination angle X7Current X at drive end of electromagnetic valve8。
The directed edge points to a symptom node X from a fault node f, and the directed edge indicates that when the fault f occurs, the working parameter X of the hydraulic support system is abnormal with a certain probability; the probability of the abnormality of the working parameter X is described by a conditional probability P (X | f), wherein the formula of the conditional probability P (X | f) is as follows:
in step 3-3, after the abnormal working parameters of the hydraulic support system are separated, the abnormal working parameters of the hydraulic support system are input into a Bayesian network, and the probability of each fault is calculated through Bayes' theorem; taking a binary state (normal and abnormal, or whether or not) as an example, when detecting that the working parameter X is abnormal, inputting the working parameter X as an observation evidence in a manner that X is 1, calculating the occurrence probability of each fault by using a formula of conditional probability P (X | f) based on bayesian theorem after inputting the observation evidence, and outputting the fault with the maximum occurrence probability as a real fault source after calculating.
The above-mentioned embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention, and any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention; accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (3)
1. The method for accurately identifying the fault source of the hydraulic support system based on data driving is characterized in that abnormal reason data mining is carried out on working parameters when the hydraulic support system is in fault, and the fault source is determined by combining a Bayesian network diagnosis model, and comprises the following steps:
step 1: selecting working parameters of various hydraulic support systems as sample data;
step 2: the hydraulic support system fault definition comprises the following steps:
an off-line training stage:
step 2-11, collecting sample data of working parameters of the hydraulic support system in a healthy state;
step 2-12, preprocessing sample data of working parameters of the hydraulic support system in the healthy state, and recording the sample data as the standardized working parameters in the healthy state;
step 2-13, calculating Hotelling's T of working parameters in the normalized health state2The threshold values of the statistic and the Q statistic are respectively recorded as
And QUCL；
And (3) an online detection stage:
step 2-21: detecting data of working parameters of the hydraulic support system in real time;
step 2-22: preprocessing the data of the working parameters of the real-time detection hydraulic support system, and recording the data as standardized real-time detection working parameters;
step 2-23: building Hotelling's T in principal component subspace2And constructing Q statistic by statistic and residual subspace, and respectively recording the Q statistic as T2And Q;
step 2-24: judging whether the hydraulic support system has faults or not, and performing the following steps
Or (Q)>QUCL) When the hydraulic support system fails, detecting working parameters in real time after corresponding standardization, and excavating abnormal reason data;
and step 3: the method for mining the data of the reasons of the abnormal working parameters comprises the following steps:
step 3-1: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The threshold value of the statistic contribution degree and the threshold value of the contribution degree to the Q statistic are respectively recorded as
And
step 3-2: calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The statistic contribution degree and the contribution degree to the Q statistic are respectively recorded as
And
step 3-3: separating abnormal working parameters of the hydraulic support system; when in use
Or
Judging abnormal working parameters of the hydraulic support system;
and 4, step 4: establishing a Bayesian network diagnosis model for hydraulic support system fault diagnosis, wherein the Bayesian network diagnosis model consists of two types of nodes, namely a fault and a symptom, and the two types of nodes are connected through directed edges according to a fault mechanism and characteristics of a hydraulic support system to form a topological structure; each fault node is a fault type which causes the hydraulic support system to work abnormally; each symptom node is a plurality of types of working parameters of the hydraulic support system; inputting the separated abnormal working parameters of the hydraulic support system into a Bayesian network diagnosis model as observation evidence, outputting fault types causing the abnormal working of the hydraulic support system, and confirming fault sources;
in the step 2-12, sample data of working parameters of the hydraulic support system in the healthy state is preprocessed and recorded as working parameters in the standardized healthy state
The formula used is as follows:
wherein, Xi' is the working parameter of the hydraulic support system in a healthy state;
is Xi' average value; sXiIs Xi' standard deviation; different values of i represent different kinds of operating parameters, i is 1,2,3,4,5,6,7 … n; n is the total number of the kinds of the working parameters;
in steps 2-13, Hotelling's T of the normalized working parameter in health state is calculated2The threshold values of the statistic and the Q statistic are respectively recorded as
And QUCL(ii) a The method comprises the following steps:
step 2-131: and (3) calculating a covariance matrix S of the standardized working parameters, wherein the calculation method comprises the following steps:
wherein m is the number of sampling groups of the working parameters in the standardized health state; x*A sample data matrix of working parameters in a health state after standardization;
step 2-132: calculating the eigenvalue and the eigenvector of the covariance matrix S, and arranging the eigenvalue and the eigenvector in a descending order according to the magnitude of the eigenvalue; wherein the characteristic value is recorded as lambdai(ii) a The feature vector is denoted as pi；i＝1,2,3,4,5,6,7…n；
Step 2-133: calculating the eigenvalue lambdaiVariance contribution rate CONTiAnd the cumulative variance contribution rate CPV (l) of the first characteristic value, wherein the calculation method comprises the following steps:
wherein,
is the sum of n eigenvalues, j ═ 1,2,3,4,5,6,7 … n; l is less than or equal to n;
step 2-134: setting a threshold value CPV (cumulative variance contribution rate) according to a detection requirement standardUCLStraight-through type CPV (k)>CPVUCLDetermining a boundary value k;
step 2-135: determining Hotelling's T2Threshold value of statistic and Q statisticThreshold value
And QUCLThe calculation formula is as follows:
wherein, Fα(k, m-k) represents the critical value of the F distribution with two degrees of freedom of k and m-k at the significance level α; c. CαIs a critical value of standard normal distribution under the significance level alpha; the value of the significance level alpha is selected according to the detection requirement;
wherein, thetaiAnd h0The calculation method comprises the following steps:
wherein, thetaiWherein i is 1,2,3,
is the i power of the j-th eigenvalue, i 1,2,3, j 1,2,3,4,5,6,7 … n;
in steps 2-23, Hotelling's T is constructed in the principal component subspace2And constructing Q statistic by using statistic and residual subspace, and respectively marking as T2And Q, comprising the steps of:
step 2-231: according to the k values obtained in the steps 2-134, the feature vector p is processediIs divided into
And
step 2-232: wherein use is made of
Projecting the standardized real-time detection working parameters to a principal component subspace for the basis vectors, and calculating T2(ii) a By using
Projecting the standardized real-time detection working parameters to a residual subspace for the basis vectors, and calculating Q; the calculation formula is as follows:
wherein Λ is a diagonal matrix formed by the first k eigenvalues of the covariance matrix S;
detecting the column vectors of the working parameters in real time after a group of standardization;
the standardized real-time detection working parameters are expressed by the following formula:
wherein,
detecting working parameters in real time after standardization; x2iWorking parameters of the hydraulic support system are detected in real time; different values of i represent different kinds of operating parameters, i is 1,2,3,4,5,6,7 … n; n is the total number of the kinds of the working parameters;
step 3-1, calculating the standardized real-time detection working parameter pair Hotelling's T when the hydraulic support system has a fault2The threshold value of the statistic contribution degree and the threshold value of the contribution degree to the Q statistic are respectively recorded as
And
the formula used is as follows:
wherein,
a critical value of chi-square distribution with a degree of freedom of 1 under the significance level alpha; the value of the significance level alpha is selected according to the detection requirement; xiiThe column vector is that the ith element is 1, and the other elements are 0;
in step 3-2, the Hotelling's T pair is detected in real time after the standardization of the hydraulic support system when the hydraulic support system has a fault is calculated2The statistic contribution degree and the contribution degree to the Q statistic are respectively recorded as
And
the formula used is as follows:
wherein ξiIs a column vector with the ith element being 1 and the remaining elements being 0.
2. The method for accurately identifying the fault source of the hydraulic support system based on the data driving as claimed in claim 1, wherein in step 1, working parameters of various types of hydraulic support systems are selected as sample data, including pump outlet emulsion pressure, pump outlet emulsion temperature, filter rear emulsion pressure, front support emulsion pressure, support moving displacement deviation, bottom plate specific pressure, top beam inclination angle and electromagnetic valve driving end current.
3. The method for accurately identifying the fault source of the hydraulic support system based on the data driving according to claim 1, wherein in the step 4, the fault type causing the abnormal operation of the hydraulic support system comprises the following steps: the method comprises the following steps of emulsion pump failure, filter blockage, no action/slow action of a stand column, no action/slow action of a jack, pipeline leakage, safety valve failure, electromagnetic valve failure, insufficient emulsion and emulsion internal liquid mixing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110636227.5A CN113323699B (en) | 2021-06-08 | 2021-06-08 | Method for accurately identifying fault source of hydraulic support system based on data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110636227.5A CN113323699B (en) | 2021-06-08 | 2021-06-08 | Method for accurately identifying fault source of hydraulic support system based on data driving |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113323699A CN113323699A (en) | 2021-08-31 |
| CN113323699B true CN113323699B (en) | 2022-06-07 |
Family
ID=77420070
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110636227.5A Active CN113323699B (en) | 2021-06-08 | 2021-06-08 | Method for accurately identifying fault source of hydraulic support system based on data driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113323699B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116028849B (en) * | 2022-12-30 | 2024-05-14 | 西安重装智慧矿山工程技术有限公司 | Emulsion pump fault diagnosis method based on depth self-coding network |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1110831A (en) * | 1964-12-22 | 1968-04-24 | Coal Industry Patents Ltd | Flow meter monitoring of a system of mine roof supports |
| CN103400231A (en) * | 2013-08-12 | 2013-11-20 | 中国矿业大学 | Equipment health management system and database modeling method thereof |
| CN103527194A (en) * | 2013-10-15 | 2014-01-22 | 淮北矿业（集团）有限责任公司 | Real-time monitoring and intelligent evaluation system for health degree of electric traction coal mining machine and method thereof |
| CN106774185A (en) * | 2015-11-23 | 2017-05-31 | 璧典凯 | A kind of fully-mechanized mining working surface hydraulic support computer distribution type monitoring system |
| CN108861426A (en) * | 2018-07-10 | 2018-11-23 | 太原理工大学 | A kind of drag conveyor chain rupture failure multi-parameter fusion identification device |
| CN111472827A (en) * | 2020-04-10 | 2020-07-31 | 太原理工大学 | Intelligent decision-making method for hydraulic support group following propulsion behavior |
| CN112417766A (en) * | 2020-12-03 | 2021-02-26 | 深制科技(苏州)有限公司 | Fault diagnosis method mainly based on fault-free data |
| CN112576312A (en) * | 2020-12-31 | 2021-03-30 | 中国矿业大学 | Data collection and processing method for electric-hydraulic control support of intelligent fully-mechanized coal mining face |
| CN112798042A (en) * | 2020-12-30 | 2021-05-14 | 中国矿业大学 | Intelligent diagnosis method for working state and supporting quality of hydraulic support |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA3128957A1 (en) * | 2019-03-04 | 2020-03-03 | Bhaskar Bhattacharyya | Near real-time detection and classification of machine anomalies using machine learning and artificial intelligence |
-
2021
- 2021-06-08 CN CN202110636227.5A patent/CN113323699B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1110831A (en) * | 1964-12-22 | 1968-04-24 | Coal Industry Patents Ltd | Flow meter monitoring of a system of mine roof supports |
| CN103400231A (en) * | 2013-08-12 | 2013-11-20 | 中国矿业大学 | Equipment health management system and database modeling method thereof |
| CN103527194A (en) * | 2013-10-15 | 2014-01-22 | 淮北矿业（集团）有限责任公司 | Real-time monitoring and intelligent evaluation system for health degree of electric traction coal mining machine and method thereof |
| CN106774185A (en) * | 2015-11-23 | 2017-05-31 | 璧典凯 | A kind of fully-mechanized mining working surface hydraulic support computer distribution type monitoring system |
| CN108861426A (en) * | 2018-07-10 | 2018-11-23 | 太原理工大学 | A kind of drag conveyor chain rupture failure multi-parameter fusion identification device |
| CN111472827A (en) * | 2020-04-10 | 2020-07-31 | 太原理工大学 | Intelligent decision-making method for hydraulic support group following propulsion behavior |
| CN112417766A (en) * | 2020-12-03 | 2021-02-26 | 深制科技(苏州)有限公司 | Fault diagnosis method mainly based on fault-free data |
| CN112798042A (en) * | 2020-12-30 | 2021-05-14 | 中国矿业大学 | Intelligent diagnosis method for working state and supporting quality of hydraulic support |
| CN112576312A (en) * | 2020-12-31 | 2021-03-30 | 中国矿业大学 | Data collection and processing method for electric-hydraulic control support of intelligent fully-mechanized coal mining face |
Non-Patent Citations (3)
| Title |
|---|
| 基于MATLABGUI煤矿液压支架故障诊断专家***设计;胡特特 等;《煤矿机械》;20151231(第12期);第285-287页 * |
| 智慧矿山背景下我国煤矿机械故障诊断研究现状与展望;樊红卫 等;《振动与冲击》;20201231;第39卷(第24期);第194-200页 * |
| 液压支架故障诊断与预测研究;张振;《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅰ辑》;20210115(第01期);第25-39页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113323699A (en) | 2021-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113723632B (en) | Industrial equipment fault diagnosis method based on knowledge graph | |
| WO2021093140A1 (en) | Cross-project software defect prediction method and system thereof | |
| US9280517B2 (en) | System and method for failure detection for artificial lift systems | |
| CN111259947A (en) | Power system fault early warning method and system based on multi-mode learning | |
| CN101957889B (en) | Selective wear-based equipment optimal maintenance time prediction method | |
| CN108921230A (en) | Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network | |
| CN111817880A (en) | Oil and gas field production equipment health management system and implementation method | |
| Deng et al. | Integrating feature optimization using a dynamic convolutional neural network for chemical process supervised fault classification | |
| CN116522088B (en) | Nuclear power plant operation data analysis method and system based on machine learning | |
| CN107065834A (en) | The method for diagnosing faults of concentrator in hydrometallurgy process | |
| CN110009126B (en) | Online alarm analysis method based on fusion of PLS model and PCA contribution degree | |
| CN113323699B (en) | Method for accurately identifying fault source of hydraulic support system based on data driving | |
| Gao et al. | Mechanical equipment health management method based on improved intuitionistic fuzzy entropy and case reasoning technology | |
| Antonello et al. | Association rules extraction for the identification of functional dependencies in complex technical infrastructures | |
| Xu et al. | Wear particle classification using genetic programming evolved features | |
| CN114721336A (en) | Information security event early warning method for technological parameters of instrument control system | |
| CN117674128A (en) | Automatic fault removal method based on power dispatching system | |
| Tripathy et al. | Explaining Anomalies in Industrial Multivariate Time-series Data with the help of eXplainable AI | |
| CA3189344A1 (en) | Explaining machine learning output in industrial applications | |
| Xiao et al. | Fault diagnosis of unseen modes in chemical processes based on labeling and class progressive adversarial learning | |
| CN117312972A (en) | Method for identifying health state of scraper conveyor speed reducer | |
| Bai et al. | Data-driven approaches: Use of digitized operational data in process safety | |
| Corrêa et al. | Data-driven approach for labelling process plant event data | |
| Alinezhad et al. | A modified bag-of-words representation for industrial alarm floods | |
| US20230022100A1 (en) | Prognostic and health management system for system management and method thereof |
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 |