CN110457979A - The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker - Google Patents

The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker Download PDF

Info

Publication number
CN110457979A
CN110457979A CN201810433376.XA CN201810433376A CN110457979A CN 110457979 A CN110457979 A CN 110457979A CN 201810433376 A CN201810433376 A CN 201810433376A CN 110457979 A CN110457979 A CN 110457979A
Authority
CN
China
Prior art keywords
tensor
matrix
tucker
fuzzy
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.)
Pending
Application number
CN201810433376.XA
Other languages
Chinese (zh)
Inventor
王威
周文起
李岩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwest A&F University
Original Assignee
Northwest A&F University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Northwest A&F University filed Critical Northwest A&F University
Priority to CN201810433376.XA priority Critical patent/CN110457979A/en
Publication of CN110457979A publication Critical patent/CN110457979A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M15/00—Testing of engines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G06N5/048—Fuzzy inferencing
    • 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
    • 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

Abstract

The present invention provides a kind of Fault Diagnosis of Engine that fuzzy control is decomposed based on tensor tucker, data in existing diesel engine ECU are integrated, form three rank tensors, pass through the thought decomposed using tensor tucker, it is decomposed with the Tucker of the tensor of HOSVD-ALS simultaneous and carries out feature extraction, obtain core tensor and its dropping cut slice matrix Bm, to core tensor dropping cut slice matrix BmData carry out data normalization processing, using fuzzy method, to treated, two tensor reasoning operations obtain output moment matrix, it obtains accurately exporting moment matrix after output moment matrix anti fuzzy method, and then realizes and fault diagnosis is carried out to diesel engine short circuit sensor, open circuit and failure.

Description

The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker
Technical field
The invention belongs to Diagnosis of Diesel Motor fields, and in particular to a kind of Diagnosis Method of Diesel Fault.
Background technique
The normal operation of diesel engine is the important guarantee of equipment safety operating, but, diesel engine fault type is more, draws The reason of playing failure is also very much, i.e., Diesel Engine Trouble has the characteristics of " fruit mostly because " and " one because of more fruits ", once engine Some section failure, the working condition that frequently can lead to engine is bad, and then influences the normal operation of equipment, may It will cause the injures and deaths on economic loss or even personnel.
Diagnosis Method of Diesel Fault is gradually evolved by traditional empirical method and observation in order to technique of dynamic measurement Based on, using signal processing technology as the conventional diagnostic techniques method of core, in recent years, but gradually development to be calculated with artificial intelligence Method is the Intelligence Diagnosis method of core, still, in existing signal characteristic extracting methods, the signal handled often table It is now the one-dimensional signal in time domain or frequency domain, i.e. vector, and is not independent from each other between each signal of engine actually, Numerous signals interfere with each other feature larger and with strong non-linear and complicated coupling, in addition, due to diesel engine event Barrier has the characteristics of " fruit mostly because " and " one because of more fruits ", causes the diagnostic result of conventional fault diagnosis system inaccurate.
Summary of the invention
Deficiency present in view of the above technology, the present invention provides a kind of engines based on tensor resolution fuzzy control Method for diagnosing faults integrates the data in existing diesel engine ECU by the thought using tensor resolution, forms Three rank tensors, and then feature extraction is carried out to data using tensor resolution, and then realize that the blurring to diesel engine controls, The feature extracting method decomposed based on tensor Tucker can inner link between deeper mining data, can be more preferable Adaptation Diesel Engine Trouble " fruit mostly because " and the characteristics of " one because of more fruits ".
The scheme of the invention is: step 1 obtains sensor signal received in Engine ECU；
Data are constituted three rank tensors being made of signal classification, revolving speed, time using the model of tensor by step 2；
Step 3 is decomposed using the Tucker of the tensor of HOSVD-ALS simultaneous and carries out feature extraction, tensor A is resolved into core TensorAnd obtain the m dropping cut slice matrix B comprising signal classification informationm, dropping cut slice matrix is just It is that three rank tensors are subjected to horizontal resection by each signal classification, obtained matrix is dropping cut slice matrix；
Step 4, to the core tensor dropping cut slice B after decompositionmData normalization processing is carried out, so that the later period carries out based on mould Paste the fault diagnosis of control；
Step 5 carries out fuzzy control fault diagnosis to the tensor after decomposition, according to input quantity BmAnd fuzzy rule, by fuzzy Fuzzy filtering calculates control amount matrix U, and by the precisely controlled moment matrix u of control amount U anti fuzzy method；
Step 6 the characteristics of according to obtained control moment matrix u, determines the failure of diesel engine.
The present invention aiming at the shortcomings in the prior art, provides a kind of diesel oil that fuzzy control is decomposed based on tensor Tucker Machine method for diagnosing faults, the present invention acquire the signal of each sensor received in ECU first, these data are then constituted one Three rank tensors being made of signal classification, revolving speed, time, decomposed with the tensor Tucker of HOSVD-ALS simultaneous The method that feature extraction is decomposed obtains core tensor B, and carries out dropping cut slice to B and obtain m signal classification dropping cut slice matrix Bm, by core tensor dropping cut slice matrix BmIn data be normalized, then use fuzzy rule and input quantity reasoning Composite calulation goes out control amount matrix U, then the precisely controlled moment matrix u of anti fuzzy method, the characteristics of according to u, and then just really The failure of diesel engine is determined.
The advantage of the invention is that can not be needed a large amount of with the inner link between each data of excavation of tensor depth Empirical data pass through with fuzzy method, improve by the method for tensor resolution fuzzy diagnosis come Diagnosis of Diesel failure To the accuracy of diesel engine failure diagnosis, passes through and decomposed using the Tucker of the tensor of HOSVD-ALS simultaneous and Fuzzy Control Theory processed is reduced and is interfered with each other between external interference and signal, solves diesel engine fault " fruit mostly because " and " one because of more fruits " The problem of.
Detailed description of the invention
Fig. 1 is detailed process of the present invention to accident analysis.
Fig. 2 is the tensor tucker decomposition algorithm original tensor with HOSVD-ALS simultaneousCore after decomposition Tensor.
Fig. 3 is the output quantity U indicated with fuzzy set C obtained with fuzzy control.
Fig. 4 is the Tucker characteristics of decomposition extraction process with the tensor of HOSVD-ALS simultaneous.
Fig. 5 is the Diagnosis Method of Diesel Fault flow chart that fuzzy control is decomposed based on tensor Tucker.
Specific embodiment
A kind of Diagnosis Method of Diesel Fault decomposing fuzzy control based on tensor Tucker, specific embodiment It is as follows.
The data-signal of sensor includes: air flow meter, coolant temperature sensing in the step one acquisition ECU Device, intake manifold pressure sensor, regulating piston motion sensor, engine speed sensor etc..
The step two constructs three ranks being made of signal classification, revolving speed, crank angle according to the above sensor signal Tensor, signal classification is the signal in addition to crankshaft position sensor, vehicle speed sensor, and revolving speed is engine sensing Device signal, crank angle are CRANK SENSOR.
The Tucker decomposition is a kind of principal component analysis of high-order, it is by a tensor representation at a core (core) tensor is multiplied by a matrix along each mode, it is assumed that each column vector of the factor matrix decomposed is orthogonal, and leads to The main left singular vector for crossing mould n expansion matrix is calculated, and the Tucker of step three tensor decomposes specific as follows:
The purpose that tensor Tucker is decomposed is to find tensorApproximate tensor, and original can be retained to the greatest extent Important information in tensor, it may be assumed that
Wherein,,,And l≤i, m≤j, n≤k decomposition can be obtained optimal Solution:
The tensor reconstructed by calculatingIt follows that
Therefore minimum problems can be equivalent to following max problem:
By tensorCarry out mode-1, mode-2, mode-3 matrixing:
。
Then carrying out Higher-order Singular value decomposition, i.e. the HOSVD tensor Tucker is decomposed,
To A(1)A(2)A(3)Matrix singular value decomposition is carried out respectively
Carry out low order processing: by B(n)Preceding R(n)A column vector is assigned to matrix b(n)
Calculate core tensor
Tensor after calculating reconstruct。
HOSVD tensor tucker decomposition algorithm cannot be guaranteed the three rank tensorsApproximation, but can obtain The initial value that good ALS tensor tucker is decomposed, therefore with iteration alternating least-squares, i.e. the ALS tensor Tucker is decomposed.
It is decomposed with the ALS tensor Tucker
It repeats
Calculate core tensor
Tensor after calculating reconstruct
Until core tensor B reaches convergence
Obtain core tensor。
Input of the output of HOSVD algorithm as ALS algorithm, and then obtain good ALS algorithm output and be used as tucker The final result of decomposition, with the tensor tucker decomposition algorithm of HOSVD-ALS simultaneous by former tensorAfter decomposition
Core tensor is
Core tensor is subjected to dropping cut slice processing, obtains the matrix B under each signal classificationm。
Core tensor dropping cut slice matrix B after being decomposed in the step four to tensor tuckermCarry out data normalizing Change processing, and as the input of fuzzy controller.
Fuzzy rule base is made of following n rule in the step five:
Ri: if B1 is D1i and B2 is D2i and … and Bm is Dmi then U is Ci, i=1,2,…,n
In formula, B1、B2、…、BmFor the core tensor dropping cut slice matrix inputted after normalized, U is the control amount square of output Battle array, CiRespectively B1、B2、…、Bm, fuzzy set of the U in its domain.
Output fuzzy quantity matrix U is indicated with fuzzy set C are as follows:
" ∧ " is and operation in formula, and " ° " is synthesis operation, and " → " is implication operation.
Anti fuzzy method is carried out to fuzzy output amount, obtains 4 × 3 matrix u of accurate output quantity, every a line difference of matrix u Air flow meter, cooling-water temperature transmitter, intake manifold pressure sensor, regulating piston motion sensor are represent, it is each Column representing fault type is short circuit, open circuit or sensor failure, one and only one data of the matrix u of output are 1, remainder data It is zero, the row and column that data are 1 forms the existing failure of engine, and then has accurately been determined that the failure of engine, this method can To judge the Diesel Engine Trouble as caused by various short circuit sensors, open circuit or sensor failure, keep worker rapid Investigation maintenance failure.

Claims (2)

1. a kind of diesel engine failure diagnosis system for decomposing fuzzy control based on tensor Tucker, it is characterised in that:
(1) sensor signal received in Engine ECU is obtained；
(2) data are constituted to three rank tensors being made of signal classification, revolving speed, time using the model of tensor；
(3) it is decomposed using the Tucker of the tensor of HOSVD-ALS simultaneous and carries out feature extraction, tensor A is resolved into core tensor, and dropping cut slice is carried out to B, obtain the m dropping cut slice matrix Bs comprising signal classification informationm；
(4) data normalization processing is carried out to the tensor after decomposing in (3), i.e., by core tensor dropping cut slice obtained in (3) Matrix BmIn data conversion at the decimal between (0,1), the fault diagnosis based on fuzzy control is carried out with the later period；
(5) to treated the core tensor dropping cut slice matrix B of data normalization in (4)mCarry out fuzzy control fault diagnosis, root According to input quantity BmAnd fuzzy rule, go out control amount matrix U by fuzzy reasoning composite calulation, and by control amount matrix U Anti-fuzzy Change obtains accurate 4 × 3 control moment matrix u；
(6) moment matrix u is controlled according to obtained in (5), every a line of matrix u respectively represents air flow meter, coolant liquid temperature Sensor, intake manifold pressure sensor, regulating piston motion sensor are spent, each column representing fault type is short circuit, open circuit Or sensor failure, one and only one data of the matrix u of output are 1, remainder data zero, the row and column that data are 1 forms The existing failure of engine, and then determine the failure cause of diesel engine.
2. a kind of diesel engine failure diagnosis system that fuzzy control is decomposed based on tensor Tucker as described in claim 1, Be characterized in that: this method can be diagnosed by various short circuit sensors, open circuit or Diesel Engine Trouble caused by failing, and makes work People can rapidly check maintenance failure.
CN201810433376.XA 2018-05-08 2018-05-08 The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker Pending CN110457979A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810433376.XA CN110457979A (en) 2018-05-08 2018-05-08 The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810433376.XA CN110457979A (en) 2018-05-08 2018-05-08 The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker

Publications (1)

Publication Number Publication Date
CN110457979A true CN110457979A (en) 2019-11-15

Family

ID=68480407

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810433376.XA Pending CN110457979A (en) 2018-05-08 2018-05-08 The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker

Country Status (1)

Country Link
CN (1) CN110457979A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110765659A (en) * 2019-11-20 2020-02-07 大连理工大学 Modeling method for supporting third-order tensor machine for aircraft engine fault diagnosis
CN112431753A (en) * 2021-01-25 2021-03-02 赛腾机电科技（常州）有限公司 Multiple quantitative diagnosis method for shoe loosening fault of axial plunger pump
CN112488312A (en) * 2020-12-07 2021-03-12 江苏自动化研究所 Tensor-based automatic coding machine construction method
CN113295413A (en) * 2021-06-24 2021-08-24 北京交通大学 Traction motor bearing fault diagnosis method based on indirect signals
CN114235413A (en) * 2021-12-28 2022-03-25 频率探索智能科技江苏有限公司 Method for constructing three-order tensor model of multi-path signal

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110765659A (en) * 2019-11-20 2020-02-07 大连理工大学 Modeling method for supporting third-order tensor machine for aircraft engine fault diagnosis
CN112488312A (en) * 2020-12-07 2021-03-12 江苏自动化研究所 Tensor-based automatic coding machine construction method
CN112431753A (en) * 2021-01-25 2021-03-02 赛腾机电科技（常州）有限公司 Multiple quantitative diagnosis method for shoe loosening fault of axial plunger pump
CN113295413A (en) * 2021-06-24 2021-08-24 北京交通大学 Traction motor bearing fault diagnosis method based on indirect signals
CN114235413A (en) * 2021-12-28 2022-03-25 频率探索智能科技江苏有限公司 Method for constructing three-order tensor model of multi-path signal
CN114235413B (en) * 2021-12-28 2023-06-30 频率探索智能科技江苏有限公司 Method for constructing multi-channel signal third-order tensor model

Similar Documents

Publication Publication Date Title
CN110457979A (en) The Diagnosis Method of Diesel Fault of fuzzy control is decomposed based on tensor Tucker
CN104712542B (en) A kind of reciprocating compressor sensitive features based on Internet of Things are extracted and method for diagnosing faults
CN103471841B (en) A kind of rotating machinery vibrating failure diagnosis method
Wang et al. Research on the fault monitoring method of marine diesel engines based on the manifold learning and isolation forest
CN108492399B (en) Bulldozer diesel engine fault diagnosis expert system and method based on big data analysis
CN111412978B (en) Reciprocating machinery abnormity detection method based on fault-free vibration signal
CN108488638B (en) Line leakage system and method based on sound wave suction wave hybrid monitoring
CN111830408A (en) Motor fault diagnosis system and method based on edge calculation and deep learning
CN103900824B (en) Diagnosis Method of Diesel Fault based on transient speed cluster analysis
CN105760839A (en) Bearing fault diagnosis method based on multi-feature manifold learning and support vector machine
CN105834835A (en) Method for monitoring tool wear on line based on multiscale principal component analysis
CN108804740B (en) Long-distance pipeline pressure monitoring method based on integrated improved ICA-KRR algorithm
CN110263649B (en) Variable-load piston engine impact characteristic extraction method based on EMD and Teager operators
CN105626502B (en) Plunger pump health evaluating method based on wavelet packet and laplacian eigenmaps
CN105487530B (en) The low failure prediction system of diesel engine row's temperature and method
CN107145675A (en) Diagnosing fault of power transformer device and method based on BP neural network algorithm
CN111678699B (en) Early fault monitoring and diagnosing method and system for rolling bearing
CN106873568A (en) Sensor fault diagnosis method based on H infinity robust Unknown Input Observer
CN104949840A (en) Diesel engine fault diagnosis method based on vibration analysis
CN108153987A (en) A kind of hydraulic pump Multiple faults diagnosis approach based on the learning machine that transfinites
CN106339720A (en) Automobile engine failure detection method
CN104142680A (en) Multiple-sensor fault diagnosis system and method based on robust input training neural network
Li et al. Intelligent fault diagnosis of aeroengine sensors using improved pattern gradient spectrum entropy
CN115809505A (en) Aerospace power fault diagnosis and health assessment method and system
CN114742094A (en) Vibration signal defect feature extraction method based on deep morphological convolution network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20191115