CN104677629A - Fault detection method for vehicle transmission - Google Patents
Fault detection method for vehicle transmission Download PDFInfo
- Publication number
- CN104677629A CN104677629A CN201410586460.7A CN201410586460A CN104677629A CN 104677629 A CN104677629 A CN 104677629A CN 201410586460 A CN201410586460 A CN 201410586460A CN 104677629 A CN104677629 A CN 104677629A
- Authority
- CN
- China
- Prior art keywords
- neural network
- output
- layer
- fault diagnosis
- input
- 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
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to a fault detection method for a vehicle transmission, and belongs to the field of vehicle fault detection. The method comprises the following steps: Step 1, acquiring initial sample data by an acquisition unit; Step 2, constructing a wavelet nerve network diagnosis model, and inputting the data sample acquired by the acquisition unit to perform sample training; Step 3, after the sample is trained, inputting real-time data acquired by the acquisition unit to perform fault diagnosis analysis. Compared with a conventional nerve network, the wavelet nerve network adopted by the fault diagnosis method of the transmission provided by the invention has the advantages that the fault diagnosis detection accuracy of the transmission with the wavelet nerve network is obviously improved, and the convergence speed is high.
Description
Technical field
The invention belongs to vehicle fault detection field, be specifically related to a kind of vehicle gear box fault detection method.
Background technology
At present, integrate and turn to the wet clutch fixed axle gearbox with straight drive function to be used widely at numerous endless-track vehicle, this wheel box is the organic synthesis of the technology such as mechanical, electrical, liquid, there is hydrodynamic drive, hydraulic stepless turn to, the technical characterstic such as power shifting and fluid drive, such wheel box has the handling characteristics of high power density, high intense and high reliability, and its function is complete, technical merit is advanced.While wet clutch wheel box significantly improves endless-track vehicle performance, also many urgent problems are brought, wherein comparatively outstanding is exactly how test fast wheel box running status and control, to ensure its riding quality good when normally working, timely maintenance during the sign that breaks down and fault diagnosis and maintenance quickly and accurately when breaking down, these problems are for the overall performance of wheel box, life-span, the importants such as reliability, running status can be grasped in time and be directly connected to wheel box integrity in use and the performance of the various function of vehicle.Therefore, research tracked vehicle gearbox fault detection method, exploitation vehicle mounted portable detection system is significant.
Summary of the invention
In order to overcome the deficiencies in the prior art, the invention provides a kind of vehicle gear box fault detection method.
Technical scheme of the present invention is: a kind of vehicle gear box method for diagnosing faults, comprises the steps: step one: the initial sample data of collecting unit collection; Step 2: set up wavelet neural network diagnostic model, the data sample that input collecting unit gathers, carries out sample training; Step 3: after sample training completes, the real time data that input collecting unit collects, carries out Analysis on Fault Diagnosis.The sample data of described collecting unit collection comprises: torque converter oil temperature, clutch pressure, hub rotation speed and torque converter pump wheel speed.Described wavelet neural network fault diagnosis model comprises input layer, hidden layer and output layer, and the neuron excitation function that hidden layer is chosen is Morlet small echo:
.The target error function of described input layer is:
In formula:
for the desired output of output layer n-th node;
for the actual output of network, P is input and output number of samples.The output of described hidden layer is:
in formula:
for input layer input;
hidden layer exports; M is input layer node; K is hidden layer node; w
kmfor the weights between hidden layer node and input layer; H() be Morlet wavelet function.The output of described output layer is:
in formula:
for output layer input; K is hidden layer node; N is output layer node; W
nkfor the weights between hidden layer node and output layer node; Sig() be Sigmod function.
The present invention has following good effect: Fault Diagnosis of Gearbox method of the present invention, adopt wavelet neural network, compared with traditional neural network, the Fault Diagnosis of Gearbox Detection accuracy of wavelet neural network has obvious lifting, but increase in high-speed cruising state lower deviation, analyzing reason should be the 8 layers of wavelet basis function locally optimal solution that made wavelet neural network be absorbed in set up based on experience; The fault diagnosis accuracy rate of algorithm of the present invention is the highest, improves at most 15%, 10% and 3% respectively than traditional neural network, wavelet neural network and genetic algorithm.Convergence of algorithm is herein fastest, and the speed of convergence of traditional neural network is the slowest, and the speed of convergence of genetic algorithm and wavelet neural network is suitable.
Accompanying drawing explanation
Fig. 1 is specific embodiment of the invention wavelet neural network structural drawing.
Embodiment
Contrast accompanying drawing below, by the description to embodiment, the specific embodiment of the present invention is as the effect of the mutual alignment between the shape of involved each component, structure, each several part and annexation, each several part and principle of work, manufacturing process and operation using method etc., be described in further detail, have more complete, accurate and deep understanding to help those skilled in the art to inventive concept of the present invention, technical scheme.
Hardware system of the present invention comprises torque converter oil temperature sensor, clutch pressure sensor, hub rotation speed sensor, torque converter pump wheel speed sensor and control system, sensor gathers the oil temperature of torque converter, clutch pressure, hub rotation speed and torque converter pump wheel speed respectively, the data collected are uploaded to control system by sensor after treatment, control system carries out analyzing and processing to data, judges whether to there is failure problems.The present invention adopts wavelet neural network algorithm to the process of input data analysis.Make full use of wavelet transformation and there is the advantage that good Time-Frequency Localization character and neural network have self-learning function, for fault diagnosis.Algorithm of the present invention, before search wavelet neural network hidden layer link weights, first uses genetic algorithm to calculate, is optimized wavelet neural network structure.
The wavelet-neural network model that the present invention adopts comprises input layer, hidden layer and output layer, and output layer adopts linear convergent rate, and input layer has M(m=1, and 2 ..., N) and individual neuron, hidden layer has K(k=1, and 2 ... K) individual neuron, as shown in Figure 1.
The neuron excitation function that hidden layer is chosen is Morlet small echo
(1)
Vibrating in order to avoid causing in weights and threshold correction when sample training one by one, adopting groups training method.To the output also not weighted sum simply of network, but first to the output weighted sum of network hidden layer small echo node, then after Sigmoid functional transformation, obtain final network and export, be conducive to treatment classification problem like this, reduce the possibility of dispersing in training process simultaneously.
Given P(p=1,2 ..., P) and organize input and output sample, learning rate is η (η >0), and factor of momentum is λ (0< λ <1), and target error function is
（2）
In formula:
for the desired output of output layer n-th node;
for the actual output of network.
The target of algorithm is constantly adjustment network parameters, makes error function reach minimum value.
Hidden layer exports
（3）
In formula:
for input layer input;
hidden layer exports; M is input layer node; K is hidden layer node; w
kmfor the weights between hidden layer node and input layer; H() be Morlet wavelet function.
Output layer exports
(4)
In formula:
for output layer input; K is hidden layer node; N is output layer node; W
nkfor the weights between hidden layer node and output layer node; Sig() be Sigmod function.
By each weight w of neural network
km, w
nkweave into the solution of a character string as problem in order, adopt real coding as follows
w
01w
02……w
1mw
o1……w
kmw
nk
Evaluation function is
f=1/（1+E）
In formula: the expression formula of E is shown in formula (2).
Concrete operations are as follows:
(1) initialization colony: in order to produce as much as possible may solution, the individuality in colony can be divided into groups;
(2) calculate the fitness of each individuality and sort, genetic operator being acted on circulation of future generation and perform, until satisfy condition.
Theoretical model after wavelet neural network increases momentum term inherits the advantage of BP neural network and wavelet neural network, has outstanding approximation of function and pattern recurrence performance simultaneously, avoids local minimum, have better practicality.In order to avoid when the complicated network structure, wavelet neural network is difficult to the problem finding optimum solution, and algorithm is before search wavelet neural network hidden layer link weights herein, first uses genetic algorithm to calculate, is optimized wavelet neural network structure.
The data of the present invention to four class sensor collections learn, and com-parison and analysis is carried out to data, compared with traditional neural network, the Fault Diagnosis of Gearbox Detection accuracy of wavelet neural network has obvious lifting, but increase in high-speed cruising state lower deviation, analyzing reason should be the 8 layers of wavelet basis function locally optimal solution that made wavelet neural network be absorbed in set up based on experience; The fault detect accuracy rate of algorithm of the present invention is the highest, improves at most 15%, 10% and 3% respectively than traditional neural network, wavelet neural network and genetic algorithm.Convergence of algorithm is herein fastest, and the speed of convergence of traditional neural network is the slowest, and the speed of convergence of genetic algorithm and wavelet neural network is suitable.Compared with algorithm of the present invention, genetic algorithm needs a large amount of training sample, and the accuracy rate of diagnosis is lower when training sample is less, and genetic algorithm has randomness, there is larger difference in the result of evolving for each time, the therefore poor reliability of result, can not stably be separated.In addition, Gearbox Fault be random, the signal collected is non-linear stochastic signal mostly, and genetic algorithm process non-linear constrain problem time need add penalty factor, this will make computing velocity significantly slow down; When processing Gearbox Fault problem, if the feature quantity extracted is more, the dimension of proper vector is just larger, and this will make genetic algorithm be difficult to process and optimize.
Above by reference to the accompanying drawings to invention has been exemplary description; obvious specific implementation of the present invention is not subject to the restrictions described above; as long as have employed the improvement of the various unsubstantialities that method of the present invention is conceived and technical scheme is carried out; or design of the present invention and technical scheme directly applied to other occasion, all within protection scope of the present invention without to improve.
Claims (6)
1. based on a Fault Diagnosis of Engine for wavelet neural network, it is characterized in that: comprise the steps:
Step one: the initial sample data gathering vehicle exhaust;
Step 2: set up wavelet neural network diagnostic model, the data sample that input gathers, carries out sample training;
Step 3: after sample training completes, inputs the real time data collected, carries out Analysis on Fault Diagnosis, exports fault type.
2. the Fault Diagnosis of Engine based on wavelet neural network according to claim 1, is characterized in that: in described step one, the initial sample data of vehicle exhaust comprises CO
2, HC, CO
1and O
2content percentage.
3. the Fault Diagnosis of Engine based on wavelet neural network according to claim 1, it is characterized in that: described wavelet neural network fault diagnosis model comprises input layer, hidden layer and output layer, the neuron excitation function that hidden layer is chosen is Morlet small echo:
。
4. the Fault Diagnosis of Engine based on wavelet neural network according to claim 1, is characterized in that: the target error function of described input layer is:
In formula:
for the desired output of output layer n-th node;
for the actual output of network, P is input and output number of samples.
5. the Fault Diagnosis of Engine based on wavelet neural network according to claim 1, is characterized in that: the output of described hidden layer is:
In formula:
for input layer input;
hidden layer exports; M is input layer node; K is hidden layer node; w
kmfor the weights between hidden layer node and input layer; H() be Morlet wavelet function.
6. the Fault Diagnosis of Engine based on wavelet neural network according to claim 1, is characterized in that: the output of described output layer is:
In formula:
for output layer input; K is hidden layer node; N is output layer node; W
nkfor the weights between hidden layer node and output layer node; Sig() be Sigmod function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410586460.7A CN104677629A (en) | 2014-10-28 | 2014-10-28 | Fault detection method for vehicle transmission |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410586460.7A CN104677629A (en) | 2014-10-28 | 2014-10-28 | Fault detection method for vehicle transmission |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104677629A true CN104677629A (en) | 2015-06-03 |
Family
ID=53312981
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410586460.7A Pending CN104677629A (en) | 2014-10-28 | 2014-10-28 | Fault detection method for vehicle transmission |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104677629A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104915715A (en) * | 2015-06-24 | 2015-09-16 | 南京航空航天大学 | Multi-method combination avionics system fault diagnosis method |
| CN106338264A (en) * | 2016-08-19 | 2017-01-18 | 江苏大学 | Fault diagnosis method for switch reluctance BSG position sensor used for hybrid power vehicle |
| CN108729494A (en) * | 2018-06-22 | 2018-11-02 | 山东大学 | Wear fault diagnosis method in speed-varying box of bulldozer service phase based on oil liquid monitoring |
| CN109855878A (en) * | 2018-12-29 | 2019-06-07 | 青岛海洋科学与技术国家实验室发展中心 | Computer-readable medium, engine failure detection device and ship |
| CN112414446A (en) * | 2020-11-02 | 2021-02-26 | 南昌智能新能源汽车研究院 | Data-driven transmission sensor fault diagnosis method |
| CN113109669A (en) * | 2021-04-12 | 2021-07-13 | 国网陕西省电力公司西安供电公司 | Power distribution network series-parallel line fault positioning method based on traveling wave characteristic frequency |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001304954A (en) * | 2000-04-20 | 2001-10-31 | Rion Co Ltd | Fault diagnosis method and device |
| CN101839805A (en) * | 2010-03-19 | 2010-09-22 | 同济大学 | Method for quality inspection of active fault and diagnosis of intelligent fault of engine |
| CN103235206A (en) * | 2012-11-05 | 2013-08-07 | 王少夫 | Transformer fault diagnosis method |
-
2014
- 2014-10-28 CN CN201410586460.7A patent/CN104677629A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001304954A (en) * | 2000-04-20 | 2001-10-31 | Rion Co Ltd | Fault diagnosis method and device |
| CN101839805A (en) * | 2010-03-19 | 2010-09-22 | 同济大学 | Method for quality inspection of active fault and diagnosis of intelligent fault of engine |
| CN103235206A (en) * | 2012-11-05 | 2013-08-07 | 王少夫 | Transformer fault diagnosis method |
Non-Patent Citations (2)
| Title |
|---|
| 王伟杰等: "基于小波神经网络的汽车发动机故障诊断", 《控制与决策》 * |
| 郑令: "基于小波神经网络的机械故障诊断方法的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104915715A (en) * | 2015-06-24 | 2015-09-16 | 南京航空航天大学 | Multi-method combination avionics system fault diagnosis method |
| CN106338264A (en) * | 2016-08-19 | 2017-01-18 | 江苏大学 | Fault diagnosis method for switch reluctance BSG position sensor used for hybrid power vehicle |
| CN106338264B (en) * | 2016-08-19 | 2018-08-21 | 江苏大学 | The method for diagnosing faults of hybrid vehicle switching magnetic-resistance BSG position sensors |
| CN108729494A (en) * | 2018-06-22 | 2018-11-02 | 山东大学 | Wear fault diagnosis method in speed-varying box of bulldozer service phase based on oil liquid monitoring |
| CN109855878A (en) * | 2018-12-29 | 2019-06-07 | 青岛海洋科学与技术国家实验室发展中心 | Computer-readable medium, engine failure detection device and ship |
| CN112414446A (en) * | 2020-11-02 | 2021-02-26 | 南昌智能新能源汽车研究院 | Data-driven transmission sensor fault diagnosis method |
| CN113109669A (en) * | 2021-04-12 | 2021-07-13 | 国网陕西省电力公司西安供电公司 | Power distribution network series-parallel line fault positioning method based on traveling wave characteristic frequency |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tang et al. | An improved convolutional neural network with an adaptable learning rate towards multi-signal fault diagnosis of hydraulic piston pump | |
| CN104677629A (en) | Fault detection method for vehicle transmission | |
| Tang et al. | Intelligent fault diagnosis of hydraulic piston pump based on deep learning and Bayesian optimization | |
| Haidong et al. | Intelligent fault diagnosis of rolling bearing using deep wavelet auto-encoder with extreme learning machine | |
| He et al. | Fault diagnosis of flywheel bearing based on parameter optimization variational mode decomposition energy entropy and deep learning | |
| Yao et al. | A lightweight neural network with strong robustness for bearing fault diagnosis | |
| CN112161784B (en) | Mechanical fault diagnosis method based on multi-sensor information fusion migration network | |
| Sun et al. | Sparse deep stacking network for fault diagnosis of motor | |
| Zhu et al. | Acoustic signal-based fault detection of hydraulic piston pump using a particle swarm optimization enhancement CNN | |
| Zhang et al. | Instance-based ensemble deep transfer learning network: A new intelligent degradation recognition method and its application on ball screw | |
| CN110334764B (en) | Rotary machine intelligent fault diagnosis method based on integrated depth self-encoder | |
| Tang et al. | An adaptive deep learning model towards fault diagnosis of hydraulic piston pump using pressure signal | |
| Rao et al. | A deep bi-directional long short-term memory model for automatic rotating speed extraction from raw vibration signals | |
| CN105760839A (en) | Bearing fault diagnosis method based on multi-feature manifold learning and support vector machine | |
| Guo et al. | Coupling fault diagnosis of wind turbine gearbox based on multitask parallel convolutional neural networks with overall information | |
| CN106874957A (en) | A kind of Fault Diagnosis of Roller Bearings | |
| CN101872165A (en) | Method for fault diagnosis of wind turbines on basis of genetic neural network | |
| Tang et al. | Towards a fault diagnosis method for rolling bearing with Bi-directional deep belief network | |
| CN109029974A (en) | A kind of epicyclic gearbox fault detection method based on one-dimensional convolutional neural networks | |
| CN109932174A (en) | A kind of Fault Diagnosis of Gear Case method based on multitask deep learning | |
| CN104680233A (en) | Wavelet neural network-based engine failure diagnosing method | |
| CN113188794A (en) | Gearbox fault diagnosis method and device based on improved PSO-BP neural network | |
| CN115859077A (en) | Multi-feature fusion motor small sample fault diagnosis method under variable working conditions | |
| CN110837851A (en) | Fault diagnosis method for hydraulic pump of electro-hydrostatic actuator | |
| Meng et al. | Intelligent fault diagnosis of gearbox based on differential continuous wavelet transform-parallel multi-block fusion residual network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150603 |
|
| WD01 | Invention patent application deemed withdrawn after publication |