CN109492689A - A kind of electric car method for diagnosing faults - Google Patents
A kind of electric car method for diagnosing faults Download PDFInfo
- Publication number
- CN109492689A CN109492689A CN201811312879.8A CN201811312879A CN109492689A CN 109492689 A CN109492689 A CN 109492689A CN 201811312879 A CN201811312879 A CN 201811312879A CN 109492689 A CN109492689 A CN 109492689A
- Authority
- CN
- China
- Prior art keywords
- attribute
- training
- decision
- neural network
- fault
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of electric car method for diagnosing faults, comprising: the signal for acquiring accelerator pedal position, speed, motor speed, generator speed and engine speed establishes Decision Table for Fault, and pre-process to the data of acquisition；The Decision Table for Fault of foundation is subjected to binary system granulation, attribute reduction is carried out to the Decision Table for Fault with Granule Computing, obtains minimal attribute set；Using the conditional attribute in the minimal attribute set as input, using its decision attribute as output, carry out the training of BP neural network and RBF neural, determine the hidden layer number of plies and hidden layer neuron number of nodes of BP neural network and RBF neural, training parameter is set, and the training ginseng includes at least: frequency of training, training function；Real-time fault diagnosis is carried out using trained BP neural network and RBF neural, obtains the fault diagnosis result of two networks；BP neural network is merged using evidence theory to obtain last diagnostic result with the diagnostic result of RBF neural.
Description
Technical field
The present invention relates to the fault diagnosis technology fields of electric car, more particularly to a kind of electric car fault diagnosis side
Method.
Background technique
For electric car as a product that structure is complicated, components are numerous, reliability and safety are that measure its good
Bad important indicator.Fault diagnosis technology is by judging electric car operating status and exception, to reinforce electronic vapour
The security performance of vehicle guarantees traffic safety.Therefore, electric car fault diagnosis technology is furtherd investigate, is had important
Theory significance and application value.Electric car is due to possessing multiple subsystems and interrelated, close-coupled so as to cause its event
Hinder between phenomenon and failure cause with extremely strong uncertain and non-linear.Thus, neural network is by its zmodem, non-
The features such as linear approximation ability and strong adaptive ability, has obtained in-depth study in electric car fault diagnosis.
Granule Computing does not need priori knowledge, is that research handles incomplete, inaccurate, fuzzy message new method, can be from amount
Potential, indispensable knowledge is excavated in many and diverse data, goes to effectively remove redundancy.Evidence theory fusion technology is because of its energy
The advantages of improving diagnostic accuracy and robustness is widely used in fault diagnosis system.
In document [Kong H, Zhang X, Bao W, et al.The Application of Granular
Computing in Electric Vehicle Fault Diagnosis[J].Australian Journal of
Electrical&Electronics Engineering, 2014,11 (3): 327-337.] in, author by BP neural network with
Granule Computing algorithm combines, and the sample dimension of pump is effectively reduced first with Granule Computing theory, the sample conduct after recycling reduction
The input of BP neural network is trained test to neural network, accelerates neural network failure diagnosis speed.But this method
Shortcomings:
(1) although Granule Computing theory can reduce sample dimension, removal redundant attributes, in actual operation due to every
A attribute more or less has certain influence on fault diagnosis result, and the attribute that removal influences very little can accelerate neural network event
Barrier diagnosis speed, but can also reduce fault diagnosis accuracy.
" one kind is based on multi-sensor information to Chinese invention patent (CN104330255A) disclosed on February 04th, 2015
The Gear Fault Diagnosis of fusion ", it combines DS evidence theory with SOM neural network, utilizes the fusion rule of evidence theory
Fusion diagnosis is carried out to evidence, error bring uncertainty is reduced, improves fault diagnosis accuracy.But this method exists not
Foot:
(1) when fault sample substantial amounts, dimension are high, SOM neural metwork training speed can decline, and real-time can obtain not
To guarantee；
(2) intuition phase can be resulted from when there is height conflict between evidence using traditional DS evidence theory
It is being contrary to as a result, even generate mistake conclusion.
Summary of the invention
In view of the foregoing deficiencies of prior art, the purpose of the present invention is to provide a kind of electric car fault diagnosis sides
Method, it is intended to solve the problems, such as because neural metwork training caused by electric car fault sample is huge, dimension is high is slow-footed, and have
It is improved electric car fault diagnosis accuracy and robustness.
In order to achieve the above objects and other related objects, the present invention provides a kind of a kind of electricity of electric car method for diagnosing faults
Electrical automobile, the electric car method for diagnosing faults include at least:
The signal of accelerator pedal position, speed, motor speed, generator speed and engine speed is acquired, event is established
Hinder decision table, and the data of acquisition are pre-processed；
The Decision Table for Fault of foundation is subjected to binary system granulation, attribute is carried out about to the Decision Table for Fault with Granule Computing
Letter obtains minimal attribute set；
Using the conditional attribute in the minimal attribute set as input, using its decision attribute as output, BP nerve is carried out
The training of network and RBF neural determines the hidden layer number of plies and hidden layer neuron of BP neural network and RBF neural
Training parameter is arranged in number of nodes, and the training ginseng includes at least: frequency of training, training objective, learning rate and training function；
Real-time fault diagnosis is carried out using trained BP neural network and RBF neural, respectively obtains two networks
Fault diagnosis result；
BP neural network is merged using evidence theory to obtain last diagnostic knot with the diagnostic result of RBF neural
Fruit.
In a kind of implementation of the invention, the data of described pair of acquisition carry out pretreated step, comprising:
Discretization is carried out to data collected, the frequency-distributeds method such as uses to carry out Data Discretization processing.
In a kind of implementation of the invention, the Decision Table for Fault by foundation carries out binary system granulation, uses Granule Computing
The step of is carried out by attribute reduction, obtains minimal attribute set for the Decision Table for Fault, comprising:
Step 2.1, the phase that dependency degree k of the conditional attribute to decision attribute is for judging original decision information system is utilized
Capacitive, if when k=1, compatible, the then continuation execution downwards of decision information system；
Otherwise, it is broken down into compatible decision information system, while deleting identical decision rule；
Wherein, the conditional attribute includes but is not limited to: accelerator pedal position, speed, motor speed, generator speed
And engine speed, the decision attribute includes but is not limited to: normal, accelerator pedal failure, motor fault；
Step 2.2, core attributes C of the conditional attribute C relative to decision attribute D is sought0；
Step 2.3, by core attributes C0It is stored in minimal attribute set R, obtains property set B, wherein the property set B is item
Part attribute C removes core attributes C0Attribute afterwards, then goes to step 2.6；
Step 2.4, to all properties b ∈ B, dependency degree is calculated, the dependency degree of each attribute is acquired；
Step 2.5, minimal attribute set R=R ∪ { b is enabledi, if dependency degree k=1, goes to step 2.7；
Step 2.6, if R property set is equal to the right C property set relative to D property set relative to the dependency degree of D property set
Dependency degree, i.e. γR(D)=γC(D) then stop operation, otherwise go to step 2.4；
Step 2.7, a minimal attribute set R of current system is obtained.
As described above, a kind of electric car method for diagnosing faults of the invention, has the advantages that
1, when carrying out electric car fault diagnosis for neural network, fault sample Pang can effectively improve using Granule Computing
Greatly, the slow problem of the high bring neural network learning training speed of dimension.
2, the present invention is when using evidence theory fusion, it is contemplated that collision problem between evidence, using the conjunction of distance between evidence
Evidences conflict bring fusion mistake is avoided at method.
3, Decision fusion is carried out to multiple evidences using evidence theory, reduces the uncertain of electric car fault diagnosis
Property, improve the accuracy of electric car fault diagnosis and robustness.
Detailed description of the invention
Fig. 1 is the structural representation of the electric car method for diagnosing faults of the present invention based on GrC-NN and evidence theory
Figure；
Fig. 2 is Granule Computing attribute reduction flow diagram described in the method for the present invention；
Fig. 3 is neural network flow diagram of the present invention；
Fig. 4 is evidence theory of the present invention diagnosis fusion flow diagram.
Specific embodiment
Illustrate embodiments of the present invention below by way of specific specific example, those skilled in the art can be by this specification
Other advantages and efficacy of the present invention can be easily understood for disclosed content.The present invention can also pass through in addition different specific realities
The mode of applying is embodied or practiced, the various details in this specification can also based on different viewpoints and application, without departing from
Various modifications or alterations are carried out under spirit of the invention.
Please refer to Fig. 1-Fig. 4.It should be noted that diagram provided in the present embodiment only illustrates this hair in a schematic way
Bright basic conception, only shown in schema then with related component in the present invention rather than component count when according to actual implementation,
Shape and size are drawn, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its component
Being laid out kenel may also be increasingly complex.
Fig. 1 is the structural representation of the electric car method for diagnosing faults of the present invention based on GrC-NN and evidence theory
Figure, it may be seen that a kind of electric car method for diagnosing faults based on GrC-NN and evidence theory provided by the invention, including
Following steps:
Step 1, the fault data for acquiring electric car constructs original decision information system and establishes decision table, and to adopting
The data of collection are pre-processed；Specific pretreatment is primarily referred to as carrying out discretization to connection attribute data, use etc. frequencies from
Arching pushing takes equal number of attribute value to be spaced as one every time that is, since initial position, if the attribute value sum of the attribute
For m, discrete is k class, then the number of samples in each interval is m/k；
Fig. 2 is Granule Computing attribute reduction flow diagram described in the method for the present invention, from Figure 2 it can be seen that the step of the present embodiment
Rapid 2.
The process of attribute reduction based on Granule Computing are as follows:
Step 2.1, the compatibility of original decision information system is judged using the dependency degree k of attribute, if when k=1, decision
Compatible, the then continuation execution downwards of information system；Otherwise it is broken down into compatible decision information system, while deleting complete phase
Same decision rule；
Step 2.2, core attributes C of the conditional attribute C relative to decision attribute D is sought0, i.e., by calculating each attribute
Dependency degree (importance) deletes the unessential redundancy condition attribute in conditional attribute collection C, if
Then a ∈ C0, wherein NE (i) represents ciThe number of middle nonzero element, ciFor the element of the i-th row in C matrix, U is domain and is non-
Empty finite aggregate；
Step 2.3, if property set R is core attributes C0, property set B is that conditional attribute C removes core attributes C0Attribute afterwards, so
After go to step 2.6；
Step 2.4, to all properties b ∈ B, k=γ is calculatedR∪{b}(D), the dependency degree (importance) of each attribute is acquired；
Step 2.5, R=R ∪ { b is enabledi, if k=1, go to step 2.7；
Step 2.6, if R property set is equal to dependence of the C property set relative to D property set relative to the dependency degree of D property set
Spend γR(D)=γC(D) then stop operation, otherwise go to step 2.4；
Step 2.7, a minimal attribute set R of current system is obtained；
In the present embodiment, binary system granulation refers to the equivalence class that the property set in U determines to construct grain, for every in U
One equivalence class is all indicated with a string of binary characters；Given knowledge base S=(U, R), wherein U is the set of entire objects,
Referred to as domain, R are the equivalence relation on domain U, it is the set of an attribute or a variety of attributes.Random subsetAll will
Domain U is divided into mutually disjoint equivalence class.Equivalence relation IND (r) is referred to as the grain with knowledge r.The binary system for being l with length
Number definition is each with the grain of knowledge r.If ui∈[X]IND(r), otherwise it is 0 that the i-th bit of corresponding binary number, which is 1, wherein
L is the gesture of U.In the present embodiment, dependency degree are as follows:
K=| POSC(D) |/U=card (POSC(D))/card(U) (1)
In formula (1), POSC(D) it is known as the positive domain D of C, card (POSC(D)) POS is indicatedC(D) element number, card (U)
Indicate the number of entire object set；
Step 3, BP neural network and RBF neural network model are established, as seen from Figure 3, with the condition in minimal attribute set
Attribute carries out the training of BP neural network and RBF neural using its decision attribute as output as input, determines BP mind
Training parameter is arranged in the hidden layer number of plies and hidden layer neuron number of nodes through network and RBF neural, i.e. setting training time
Number, training objective, learning rate and training function；
Fig. 3 is neural network flow diagram of the present invention, is needed in the present embodiment respectively to BP neural network
It is trained with RBF neural with test, firstly, being trained to BP neural network, according in any closed interval
Any one continuous function can be approached with the BP network of a hidden layer, the network therefore BP neural network haves three layers, respectively
Input layer, hidden layer, output layer.Wherein hidden layer plays very important effect to the performance of BP neural network, its node
Number t uses empirical equation:
In formula (2), x is input neuron number；Y is output neuron number；Constant of the s between 1-10.In the present embodiment
In, take x=5, y=3 to calculate to obtain t=[3,12], reuse the ascending change t of trial-and-error method, and comprehensively consider output error and
The number of iterations, finally selected node in hidden layer is 10.Hidden layer and output layer neuron transmission function be respectively tansig with
Logsig, training function select trainlm, and maximum frequency of training is 2000, learning rate 0.01, and target mean square error is
0.005；Training BP neural network, if training network error does not reach requirement, failure to train needs to reset training
Parameter carries out the training of next round.If the network of training reaches error requirements, BP neural network training is completed.Then, right
RBF neural is trained, and in the present embodiment, the expansion rate of RBF neural is 2, and target mean square error is
0.005；Training RBF Neural Network, if training network error does not reach requirement, failure to train needs to reset instruction
Practice parameter, carries out the training of next round.If the network of training reaches error requirements, RBF neural training is completed.
Step 4, real-time fault diagnosis is carried out using trained BP neural network and RBF neural, respectively obtains two
The fault diagnosis result of a network；
Step 5, BP neural network is merged using evidence theory to obtain with the diagnostic result of RBF neural final
Diagnostic result；
Fig. 4 is evidence theory of the present invention diagnosis fusion flow diagram, from fig. 4, it can be seen that demonstrate,proving described in this example
According to the process of theory fusion are as follows:
Step 5.1, it determines the basic probability assignment of each proposition in identification framework, that is, utilizes BP neural network and RBF nerve
The output of network determines its Basic probability assignment function are as follows:
In formula (3), n is fault category number；yiThe actual value of node layer is exported for i-th of neural network；tiFor nerve net
The desired value of i-th of network output node layer；BiFor i-th kind of fault category；M (θ) is the uncertainty of evidence；
Step 5.2, it diagnoses according to synthesizing BP neural network and RBF neural based on the evidence theory of distance between evidence
As a result:
Distance between two evidences are as follows:
Similarity between two Basic probability assignment functions are as follows:
Sim(m1,m2)=1-d (m1,m2) (5)
Evidence miSupport are as follows:
Evidence miConfidence level degree are as follows:
Conflict between any two evidence are as follows:
The global conflict of n evidence are as follows:
In formula (9), weight is defined as
Effective property coefficient of conflicting evidence are as follows:
ε=e-k (10)
The Basic probability assignment function of each proposition is obtained by formula (3), then the synthesis of evidence theory are as follows:
In formula (11), k is the conflict for measuring n evidence；ε is effective property coefficient of conflicting evidence；
Step 5.3, judgment rule is formulated, obtains diagnostic result；
In this example, the judgment rule of formulation refer to by belief function Bel and verisimilitude function Pl obtain confidence interval [Bel,
Pl], function Bel (A) is known as the belief function of A, indicates that evidence is genuine trusting degree for A；Function Pl (A) is known as A seemingly
True function indicates that evidence is the trusting degree of non-vacation for A.It determines in conjunction with the uncertainty m (θ) of evidence, and by following fusion
Plan rule obtains last diagnostic result B:
Rule
Regular 2Bel (B)-Bel (Bj) > ε1, Bel (B)-m (θ) > ε2,ε1,ε2∈ R and ε1,ε2> 0
Regular 3m (θ) < ε3,ε3∈ R and ε3> 0
Rule 1 shows the maximum confidence that belief function should have；Rule 2 shows the confidence level and other events of diagnosis
The confidence level and uncertainty for hindering type are greater than ε1、ε2；Rule 3 shows that the uncertainty of evidence is less than ε3, meeting above-mentioned 3
Rule just can determine that final diagnostic result B.
The above-described embodiments merely illustrate the principles and effects of the present invention, and is not intended to limit the present invention.It is any ripe
The personage for knowing this technology all without departing from the spirit and scope of the present invention, carries out modifications and changes to above-described embodiment.Cause
This, institute is complete without departing from the spirit and technical ideas disclosed in the present invention by those of ordinary skill in the art such as
At all equivalent modifications or change, should be covered by the claims of the present invention.
Claims (3)
1. a kind of electric car method for diagnosing faults, which is characterized in that the electric car method for diagnosing faults includes at least:
The signal for acquiring accelerator pedal position, speed, motor speed, generator speed and engine speed, establishes failure and determines
Plan table, and the data of acquisition are pre-processed；
The Decision Table for Fault of foundation is subjected to binary system granulation, attribute reduction is carried out to the Decision Table for Fault with Granule Computing, is obtained
To minimal attribute set；
Using the conditional attribute in the minimal attribute set as input, using its decision attribute as output, BP neural network is carried out
With the training of RBF neural, the hidden layer number of plies and hidden layer neuron node of BP neural network and RBF neural are determined
Training parameter is arranged in number, and the training ginseng includes at least: frequency of training, training objective, learning rate and training function；
Real-time fault diagnosis is carried out using trained BP neural network and RBF neural, respectively obtains the event of two networks
Hinder diagnostic result；
BP neural network is merged using evidence theory to obtain last diagnostic result with the diagnostic result of RBF neural.
2. a kind of electric car method for diagnosing faults according to claim 1, which is characterized in that the data of described pair of acquisition
Carry out pretreated step, comprising:
Discretization is carried out to data collected, the frequency-distributeds method such as uses to carry out Data Discretization processing.
3. a kind of electric car method for diagnosing faults according to claim 1, which is characterized in that the failure by foundation
Decision table carries out binary system granulation, carries out attribute reduction to the Decision Table for Fault with Granule Computing, obtains the step of minimal attribute set
Suddenly, comprising:
It step 2.1, is for judging the compatible of original decision information system using dependency degree k of the conditional attribute to decision attribute
Property, if when k=1, compatible, the then continuation execution downwards of decision information system；
Otherwise, it is broken down into compatible decision information system, while deleting identical decision rule；
Wherein, the conditional attribute includes but is not limited to: accelerator pedal position, speed, motor speed, generator speed and hair
Motivation revolving speed, the decision attribute includes but is not limited to: normal, accelerator pedal failure, motor fault；
Step 2.2, core attributes C of the conditional attribute C relative to decision attribute D is sought0；
Step 2.3, by core attributes C0It is stored in minimal attribute set R, obtains property set B, wherein the property set B is conditional attribute
C removes core attributes C0Attribute afterwards, then goes to step 2.6；
Step 2.4, to all properties b ∈ B, dependency degree is calculated, the dependency degree of each attribute is acquired；
Step 2.5, minimal attribute set R=R ∪ { b is enabledi, if dependency degree k=1, goes to step 2.7；
Step 2.6, if R property set is equal to dependence of C property set in the right relative to D property set relative to the dependency degree of D property set
Degree then stops operation, otherwise goes to step 2.4；
Step 2.7, a minimal attribute set R of current system is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811312879.8A CN109492689A (en) | 2018-11-06 | 2018-11-06 | A kind of electric car method for diagnosing faults |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811312879.8A CN109492689A (en) | 2018-11-06 | 2018-11-06 | A kind of electric car method for diagnosing faults |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109492689A true CN109492689A (en) | 2019-03-19 |
Family
ID=65693867
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811312879.8A Pending CN109492689A (en) | 2018-11-06 | 2018-11-06 | A kind of electric car method for diagnosing faults |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492689A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110084366A (en) * | 2019-03-25 | 2019-08-02 | 合肥工业大学智能制造技术研究院 | Electric car fault diagnosis neural network method based on Granule Computing |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103714383A (en) * | 2014-01-09 | 2014-04-09 | 北京泰乐德信息技术有限公司 | Rail transit fault diagnosis method and system based on rough set |
| CN106202886A (en) * | 2016-06-29 | 2016-12-07 | 中国铁路总公司 | Track circuit red band Fault Locating Method based on fuzzy coarse central Yu decision tree |
-
2018
- 2018-11-06 CN CN201811312879.8A patent/CN109492689A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103714383A (en) * | 2014-01-09 | 2014-04-09 | 北京泰乐德信息技术有限公司 | Rail transit fault diagnosis method and system based on rough set |
| CN106202886A (en) * | 2016-06-29 | 2016-12-07 | 中国铁路总公司 | Track circuit red band Fault Locating Method based on fuzzy coarse central Yu decision tree |
Non-Patent Citations (3)
| Title |
|---|
| 严刚等: "《Grc-SVM在电动汽车故障诊断中的应用》", 《机械工程与自动化》 * |
| 孔慧芳等: "《基于神经网络和证据理论的DCT电控系统故障诊断》", 《合肥工业大学学报（自然科学版）》 * |
| 陈泽华: "《粒计算及人工选择算法理论研究》", 《万方数据知识服务平台》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110084366A (en) * | 2019-03-25 | 2019-08-02 | 合肥工业大学智能制造技术研究院 | Electric car fault diagnosis neural network method based on Granule Computing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960303B (en) | Unmanned aerial vehicle flight data anomaly detection method based on LSTM | |
| Xia et al. | Automatic generation method of test scenario for ADAS based on complexity | |
| JP6855535B2 (en) | Simulation data optimization methods, devices, storage media and programs | |
| CN110533070A (en) | A kind of mechanical breakdown intelligent diagnosing method based on migration original shape network under small sample | |
| Tageldin et al. | Examining pedestrian evasive actions as a potential indicator for traffic conflicts | |
| CN114048468A (en) | Intrusion detection method, intrusion detection model training method, device and medium | |
| Liu et al. | A concurrent fault diagnosis method of transformer based on graph convolutional network and knowledge graph | |
| CN110851654A (en) | Industrial equipment fault detection and classification method based on tensor data dimension reduction | |
| CN112433927A (en) | Cloud server aging prediction method based on time series clustering and LSTM | |
| Du et al. | Finding diverse failure scenarios in autonomous systems using adaptive stress testing | |
| Jiang et al. | Efficient and unbiased safety test for autonomous driving systems | |
| CN109492689A (en) | A kind of electric car method for diagnosing faults | |
| Shi et al. | Automatic clustering for unsupervised risk diagnosis of vehicle driving for smart road | |
| Kuwajima et al. | Open problems in engineering and quality assurance of safety critical machine learning systems | |
| CN102521202B (en) | Automatic discovery method of complex system oriented MAXQ task graph structure | |
| CN103888460A (en) | Controller local area network protocol verification method based on state space search | |
| CN116403090A (en) | Small-size target detection method based on dynamic anchor frame and transducer | |
| Guo et al. | Automated scenario generation and evaluation strategy for automatic driving system | |
| WO2023193196A1 (en) | Autonomous driving test case generation method and apparatus, and electronic device and storage medium | |
| CN116166642A (en) | Spatio-temporal data filling method, system, equipment and medium based on guide information | |
| CN115310079A (en) | Display method based on intelligent network connection automobile attack matrix | |
| CN115248993A (en) | Method and device for detecting reality of simulation scene model and storage medium | |
| CN113239034A (en) | Big data resource integration method and system based on artificial intelligence and cloud platform | |
| Gao et al. | WDBM: Weighted Deep Forest Model Based Bearing Fault Diagnosis Method | |
| Zhang et al. | Elimination mechanism of glue variables for solving SAT problems in linguistics |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190319 |
|
| RJ01 | Rejection of invention patent application after publication |