CN100595548C - Automotive engine fault diagnosis system and method based on sparse expression - Google Patents
Automotive engine fault diagnosis system and method based on sparse expression Download PDFInfo
- Publication number
- CN100595548C CN100595548C CN200810198342A CN200810198342A CN100595548C CN 100595548 C CN100595548 C CN 100595548C CN 200810198342 A CN200810198342 A CN 200810198342A CN 200810198342 A CN200810198342 A CN 200810198342A CN 100595548 C CN100595548 C CN 100595548C
- Authority
- CN
- China
- Prior art keywords
- module
- signal
- fault
- checked
- submodule
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides an automobile engine fault diagnosing system based on sparse representation, comprising: a DSP module used for operating diagnosis algorithm; an IDE module used for storing a fault file database, a database of signals to be detected and a diagnosis result filing database; a network interface module used for receiving aural signals of the automobile engine coming from the internet, and sending the diagnosis result; an I/O module used for providing a man-machine interaction mechanism; and an FPGA module used for system control and the interface adaptation among all the modules. The automobile engine fault diagnosing method comprises the steps of: preprocessing, sparse decomposition, the feature extraction of the fault file and the signals to be detected, classificationalgorithm training, fault diagnosis, diagnosis result analysis, the updating of the fault file database and the database of the signals to be detected, stop judgment and diagnosis result output. Thesystem and the method can realize remote diagnosis to the aural signals of the automobile engine, the needed fault file database is small, the diagnosis accuracy is high, the cost is low, and the system is convenient for maintenance and upgrading.
Description
Technical field
The present invention relates to the automobile engine failure diagnosis technology, be specifically related to automobile engine failure diagnosis system and method based on rarefaction representation.
Background technology
The sound that sends in the motor car engine course of work has comprised a large amount of mechanical movement signals, and when engine broke down, variation and distortion just can appear in the sound and the frequency spectrum thereof of its vibration.Thereby decompose by the voice signal that engine is sent, make that the useful signal in the voice signal can separate with garbage signal, extract the feature of signal then, can detect the running status of engine, thereby it is carried out fault diagnosis.
At present, the system that utilizes the motor car engine voice signal to carry out fault diagnosis mainly is based on PC, also there is part be based on embedded architecture, but whole structure does not change, all be to obtain the engine sound signal, handle then, can't realize remote fault diagnosis from sensor, and cost is higher, upgrading and difficult in maintenance.
On the other hand, existing method for diagnosing faults is in order to obtain higher precision, must set up fault file database in large scale in advance to extract discriminating power good signal feature, not only expended a large amount of costs but also inconvenience in actual applications.And motor car engine forms of motion complexity is various, driving source is many, the harmonic wave of voice signal belongs to non-stationary signal, constitutes very complexity, and actual monitoring is often introduced the complicated background noise, cause stronger interference, traditional time domain or frequency domain method are difficult to hold effectively its characteristics.
Summary of the invention
The objective of the invention is to overcome the shortcoming and defect of above-mentioned prior art, a kind of automobile engine failure diagnosis system based on rarefaction representation is provided, this system can carry out remote fault diagnosis to the motor car engine voice signal that comes automatic network, and data-handling capacity is powerful, working stability, cost is low, is convenient to safeguard and upgrading.
The present invention also aims to provide the method that realizes automobile engine failure diagnosis based on the automobile engine failure diagnosis system of rarefaction representation by above-mentioned, this method will diagnose believable signal to be checked to add the fault file database, thereby reduce the cost of setting up fault data, improve detection efficiency; And select basis function rarefaction representation class differences signal adaptively, obtain the strong feature of recognition capability, to improve the accuracy of fault diagnosis.
The object of the invention is achieved through the following technical solutions: this comprises based on the automobile engine failure diagnosis system of rarefaction representation:
Be used to receive signal to be checked that the fault diagnosis terminal transmits and the Network Interface Module that postbacks diagnostic result;
Be used for control and administering digital signal Processing (DSP) module, Network Interface Module, data storage (IDE) module and input and output (I/O) module, the signal to be checked that reception is transmitted through Network Interface Module also generates Signals Data Base to be checked and deposits the IDE module in, the diagnostic result of DSP module output is deposited in field programmable logic array (FPLA) (FPGA) module of the diagnostic result filing database of IDE module;
Be used to store the IDE module of engine sound signal fault archive database, Signals Data Base to be checked and diagnostic result filing database;
From the IDE module, read signal to be checked and diagnose, export the DSP module of diagnostic result then;
Be used to provide the I/O module of man-machine interaction;
Simultaneously, described FPGA module is connected with DSP module, IDE module, Network Interface Module, I/O module simultaneously, and the DSP module is connected with the IDE module simultaneously, and described Network Interface Module is connected with the fault diagnosis terminal signaling.
Described DSP module is used to diagnose the motor car engine voice signal, comprises peripheral components such as dsp chip and corresponding D DR2 internal memory, Flash chip, and described Flash chip is used to solidify fault diagnosis algorithm, and the DDR2 internal memory is used to realize the high-speed computation of data.
Described Network Interface Module adopts GSM (GPRS) or wired ethernet different communication modes such as (10M/100Mbps) to be connected with the fault diagnosis terminal signaling, receive the motor car engine voice signal of automatic network, and the transmission diagnostic result is finished visit and management to long-range engine sound signal picker.
Described I/O module comprises keyboard input interface, liquid crystal display output interface, usb data introducting interface and JATG debugging interface, and man-machine interaction mechanism and system management, maintenance and upgrading are provided.
Described FPGA module adopts ARM9 kernel or PPC kernel, can transplant embedded real-time operating system (as μ COS, VxWorks etc.) in kernel.
Above-mentioned DSP module, IDE module, Network Interface Module and I/O module all realize that by FPGA interface is adaptive.
Described DSP module comprises:
Be used for the engine sound signal of engine sound signal fault archive database is carried out pre-service, and the signal to be checked in the Signals Data Base to be checked is carried out pretreated pretreatment module;
Be used for that pretreated fault file class differences signal is carried out adaptive sparse and decompose, to select the Sparse Decomposition module of optimal base function;
Be used to calculate the inner product of described optimal base function and fault file signal, to extract the fault signature extraction module of fault file feature;
Be used to utilize the sorting algorithm module of fault file features training sorting algorithm;
Be used to calculate the inner product of described optimal base function and signal to be checked, to generate the signal characteristic extraction module to be checked of signal characteristic to be checked;
Be used for sorting algorithm that signal characteristic substitution sorting algorithm module to be checked is trained, to carry out the fault diagnosis module of fault diagnosis;
Be used for according to signal characteristic to be checked fault diagnosis degree of confidence thresholding being set, judge whether believable diagnostic result analysis module of diagnostic result to the distance of the lineoid of classifying;
Be used for the diagnostic result analysis module is judged as the fault file database update module that believable signal to be checked adds the fault data file store;
Be used for the diagnostic result analysis module is judged as the Signals Data Base update module to be checked that incredible signal to be checked is retained in Signals Data Base to be checked;
Be used to verify the judging module of ending of Signals Data Base to be checked;
Be used for diagnostic result is sent to the diagnostic result output module of FPGA;
Simultaneously, described pretreatment module, the Sparse Decomposition module, the fault signature extraction module, the sorting algorithm module, fault diagnosis module, the diagnostic result analysis module, fault file database update module connects successively, described fault file database update module is connected with the Sparse Decomposition module simultaneously, described diagnostic result analysis module while and Signals Data Base update module to be checked, by judging module, the diagnostic result output module connects successively, described Sparse Decomposition module with after signal characteristic extraction module to be checked is connected respectively with fault diagnosis module, be connected by judging module.
Described sorting algorithm module is support vector machine module or other sorting algorithm modules.
Described Sparse Decomposition module comprises:
The class differences submodule that is used for calculation engine fault file database 2 class signal average differences;
Be used to make up the super complete basic submodule of super complete base;
Adopt coupling track algorithm and particle swarm optimization algorithm, select the basis function chooser module of minority basis function rarefaction representation fault file class differences signal;
Be used for the parameter of selected basis function is encoded, with the initialization submodule of initialization population;
Be used for decomposing one by one the submodule of decomposition one by one of residue signal;
Be used to calculate the ratio of the average of decomposing the engine failure archive database class differences signal that the each residue signal energy that decomposes of submodule and class differences submodule calculated one by one, when if this ratio is not less than default thresholding, then returns and decompose decomposition that submodule proceeds to decompose one by one to residue signal one by one by the judgement submodule;
Simultaneously, described class differences submodule, super complete basic submodule simultaneously and after basis function chooser module is connected successively with the initialization submodule, one by one decompose submodule, decompose by adjudicating submodule and be connected.
Utilize above-mentioned automobile engine failure diagnosis system based on rarefaction representation to realize the method for automobile engine failure diagnosis, comprise the steps:
(1) after the system start-up, the FPGA module is responsible for dispatching and finishing the self check of IDE module, DSP module, Network Interface Module and I/O module;
(2) the FPGA module receives the response message of fault diagnosis terminal by Network Interface Module broadcast system initiation message, the record network topology structure, and be stored in the IDE module;
(3) the fault diagnosis terminal sends motor car engine voice signal to be checked to system, and this signal reaches the FPGA module by network interface, generates Signals Data Base to be checked, and is stored in the IDE module;
(4) the DSP module is called the fault file database in the IDE module, by pretreatment module the engine sound signal in the fault file database is carried out pre-service earlier; Then pretreated fault file class differences signal is carried out adaptive sparse and decompose, select the optimal base function by the Sparse Decomposition module; Calculate the inner product of selected basis function and fault file signal then by the fault signature extraction module, extract the fault file feature; By the sorting algorithm module, utilize fault file features training sorting algorithm at last;
(5) the DSP module is taken out signal to be checked from the IDE module, by pretreatment module signal to be checked is carried out pre-service earlier; Then by signal characteristic extraction module to be checked, the inner product of selected basis function of calculation procedure (4) and signal to be checked generates signal characteristic to be checked; By fault diagnosis module,, carry out fault diagnosis then with the sorting algorithm that the feature substitution step (4) of signal to be checked is trained; At last by the diagnostic result analysis module, according to the distance of signal characteristic to be checked to the lineoid of classifying, fault diagnosis degree of confidence thresholding is set, judge whether diagnostic result is credible, if credible, by fault file database update module, believable signal to be checked is added the fault file database, otherwise,, incredible signal to be checked is retained in Signals Data Base to be checked by Signals Data Base update module to be checked;
(6) the DSP module is by verifying Signals Data Base to be checked by judging module, if it be empty or in the front and back two-wheeled is diagnosed no change, then diagnostic result is gathered and is sent to the FPGA module; Otherwise the diagnosis of a new round is carried out in repeating step (4), (5);
(7) the FPGA module diagnostic result that will gather deposits the diagnostic result filing database in, and by Network Interface Module diagnostic result is sent to the fault diagnosis terminal.
In the said method, pretreatment module is carried out sample frequency, the data length that pre-service comprises that selection is suitable to the engine sound signal in the fault file database, utilizes median filtering algorithm to remove noise.
In the said method, pretreatment module is carried out sample frequency, the data length that pre-service comprises that selection is suitable to signal to be checked, utilizes median filtering algorithm to remove noise.
In the said method, step (4), (5) described sorting algorithm are support vector machine.
In the said method, when described fault file database is 2 classification fault file databases, the described Sparse Decomposition module of step (4) is carried out the adaptive sparse decomposition to pretreated fault file class differences signal, selects the optimal base function, specifically may further comprise the steps:
A, class differences submodule calculate the difference D (t) of fault file database 2 class signal averages, that is:
Wherein x (t) is a fault file database engine sound signal, ω
1, ω
2The expression class formative, n
1, n
2Be respectively the quantity of two class signals;
Parameter b wherein
nThe normalization coefficient of expression basis function makes
And u
n, s
n, f
n, β
nThe peak value moment of representing basis function respectively, decay factor, frequency and initial phase;
C, basis function chooser module adopt coupling track algorithm and particle swarm optimization algorithm, select minority basis function rarefaction representation fault file class differences signal, that is:
Wherein
Be selected basis function, a
kBe rarefaction representation coefficient, a
kSimilarity degree between difference signal D (t) and the basis function has been described; Find the solution
a
kProcess as follows:
(a) establish residue signal r the 0th time
0(t)=and D (t), the initialization submodule is to the basis function parameters u
n, s
n, f
n, β
nEncode the initialization population;
(b) decompose submodule one by one and decompose residue signal one by one, establishing the p time residue signal is r
p(t), according to particle swarm optimization algorithm, select best particle { u
p, s
p, f
p, β
p, make basis function
With r
p(t) the most similar, then
Rarefaction representation coefficient a
pFor:
(c) decomposition is responsible for calculating the residue signal energy by the judgement submodule || r
p(t) || with || D (t) || ratio, when if this ratio is not less than default thresholding, then returns and decompose submodule one by one residue signal is proceeded to decompose one by one, if this ratio is during less than default thresholding, then stop to decompose, selected optimal base function is decomposed in output each time.
When described fault file database be 3 the classification more than the fault file database time, adopt differential method one by one, move above-mentioned method for diagnosing faults during to 2 classification fault file databases repeatedly, judge one by one whether signal to be checked belongs to each fault of fault file database.
The automobile engine failure diagnosis system that the present invention is based on rarefaction representation has the following advantages with respect to prior art:
(1) the present invention obtains remote automobile engine voice signal to be checked by network, and tracing trouble classification in time greatly facilitates the auto repair work of each site (especially remote districts);
(2) system of the present invention not only has powerful data-handling capacity, and diagnosis speed is fast, and is convenient to safeguard and upgrading that cost is lower;
(3) the present invention will diagnose believable signal to be checked to add the fault file database, and required fault file database is little, can reduce cost effectively, improves detection efficiency;
(4) the present invention introduces engine sound signal to be checked as feedback information, and basis function that can adaptive selection rarefaction representation obtains the strong feature of recognition capability, thereby improves the fault diagnosis accuracy.
Description of drawings
Fig. 1 is the hardware structure diagram of system of the present invention;
Fig. 2 is the structural drawing of the dsp chip in the system of the present invention;
Fig. 3 is the structural drawing of the Sparse Decomposition module in the system of the present invention;
Fig. 4 is the FB(flow block) that the present invention is based on the automobile engine failure diagnosis method of rarefaction representation;
Fig. 5 is that the Sparse Decomposition module is carried out Sparse Decomposition, selects the FB(flow block) of optimal base function when described fault file database is 2 classification fault file databases.
Embodiment
Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.
Embodiment
Figure 1 shows that the hardware structure diagram of system of the present invention, this is based on the automobile engine failure diagnosis system of rarefaction representation, comprise DSP module, IDE module, Network Interface Module, I/O module, FPGA module, described FPGA module is connected with DSP module, IDE module, Network Interface Module, I/O module simultaneously, the DSP module is connected with the IDE module simultaneously, and described Network Interface Module is connected with the fault diagnosis terminal signaling.
Described DSP module is used to diagnose the motor car engine voice signal;
Described IDE module stores has engine sound signal fault archive database, Signals Data Base to be checked and diagnostic result filing database.
Described Network Interface Module is used for receiving the motor car engine voice signal of automatic network, and sends diagnostic result;
Described I/O module is used to provide man-machine interaction mechanism, makes things convenient for system management, maintenance and upgrading.
Described FPGA module is used to realize the control and the management of DSP module, Network Interface Module, IDE module and I/O module;
Above-mentioned DSP module, IDE module, Network Interface Module and I/O module all realize that by FPGA interface is adaptive.
Described DSP module can be selected the floating type dsp chip (as TMS320C6727B) and the peripheral components such as corresponding D DR2, Flash of TI company for use.The Flash chip is used to solidify fault diagnosis algorithm, and the DDR2 internal memory is used to realize the high-speed computation of data.
Described IDE can select high capacity IDE for use, stores engine sound signal fault archive database, Signals Data Base to be checked and diagnostic result filing database.
Described Network Interface Module can adopt GSM (GPRS) or wired ethernet different communication modes such as (10M/100Mbps) to realize according to the actual requirements, finishes visit and management to long-range engine sound signal picker by network.
Described I/O module interface comprises keyboard input interface, liquid crystal display output interface, usb data introducting interface and JATG debugging interface.
Described FPGA module can be selected (as the XC3S400-PQ208) of Xilinx company for use, wherein comprises ARM9 kernel or PPC kernel, can transplant embedded real-time operating system (as μ COS, VxWorks etc.) in kernel, is used to realize the function of network and system management.
Figure 2 shows that the concrete structure of the DSP module in the system of the present invention, the DSP module comprises the pretreatment module that connects successively, the Sparse Decomposition module, the fault signature extraction module, the sorting algorithm module, fault diagnosis module, the diagnostic result analysis module, fault file database update module, described fault file database update module is connected with the Sparse Decomposition module simultaneously, described diagnostic result analysis module while and Signals Data Base update module to be checked, by judging module, the diagnostic result output module connects successively, described Sparse Decomposition module with after signal characteristic extraction module to be checked is connected respectively with fault diagnosis module, be connected by judging module.
The structure of described Sparse Decomposition module as shown in Figure 3, comprise class differences submodule, super complete basic submodule, basis function chooser module, initialization submodule, one by one decompose submodule, decompose by the judgement submodule, described class differences submodule, super complete basic submodule simultaneously and after basis function chooser module is connected successively with the initialization submodule, one by one decompose submodule, decompose by adjudicating submodule and be connected.
Figure 4 shows that the FB(flow block) of the automobile engine failure diagnosis method that the present invention is based on rarefaction representation, the course of work of native system may further comprise the steps:
(1) after the system start-up, the FPGA module is responsible for dispatching and finishing the self check of IDE module, DSP module, Network Interface Module and I/O module;
(2) the FPGA module receives the response message of fault diagnosis terminal by Network Interface Module broadcast system initiation message, the record network topology structure, and be stored in the IDE module;
(3) the fault diagnosis terminal sends motor car engine voice signal to be checked to system, and this signal reaches the FPGA module by network interface, generates Signals Data Base to be checked, and is stored in the IDE module;
(4) the DSP module is called the fault file database in the IDE module, earlier the engine sound signal in the fault file database is carried out pre-service by pretreatment module, comprise and select suitable sample frequency, data length, utilize median filtering algorithm to remove noise; Then pretreated fault file class differences signal is carried out adaptive sparse and decompose, select the optimal base function by the Sparse Decomposition module; Calculate the inner product of selected basis function and fault file signal then by the fault signature extraction module, extract the fault file feature; By the sorting algorithm module, utilize fault file features training support vector machine at last, support vector machine needs input fault archives training calibration just as a rod spear, just can use then, just calls canonical function as for the process of input training);
(5) the DSP module is taken out signal to be checked from the IDE module, by pretreatment module signal to be checked is carried out pre-service earlier, comprises and selects suitable sample frequency, data length, utilizes median filtering algorithm to remove noise; Then by signal characteristic extraction module to be checked, the inner product of selected basis function of calculation procedure (4) and signal to be checked generates signal characteristic to be checked; By fault diagnosis module,, carry out fault diagnosis then with the sorting algorithm that the feature substitution step (4) of signal to be checked is trained; At last by the diagnostic result analysis module, according to the distance of signal characteristic to be checked to the lineoid of classifying, fault diagnosis degree of confidence thresholding is set, judge whether diagnostic result is credible, if credible, by fault file database update module, believable signal to be checked is added the fault file database, otherwise,, incredible signal to be checked is retained in Signals Data Base to be checked by Signals Data Base update module to be checked;
(6) the DSP module is by verifying Signals Data Base to be checked by judging module, if it be empty or in the front and back two-wheeled is diagnosed no change, then diagnostic result is gathered and is sent to the FPGA module; Otherwise the diagnosis of a new round is carried out in repeating step (4), (5);
(7) the FPGA module diagnostic result that will gather deposits the diagnostic result filing database in, outputs to LCD screen from the I/O module simultaneously and shows, and by Network Interface Module diagnostic result is sent to the fault diagnosis terminal.
In the said method, when described fault file database was 2 classification fault file databases, the described Sparse Decomposition module of step (4) was carried out adaptive sparse to pretreated fault file class differences signal and is decomposed selection optimal base function, as shown in Figure 5, specifically may further comprise the steps:
A, class differences submodule calculate the difference D (t) of fault file database 2 class signal averages, that is:
Wherein x (t) is a fault file database engine sound signal, ω
1, ω
2The expression class formative, n
1, n
2Be respectively the quantity of two class signals;
Parameter b wherein
nThe normalization coefficient of expression basis function makes
And u
n, s
n, f
n, β
nThe peak value moment of representing basis function respectively, decay factor, frequency and initial phase;
C, basis function chooser module adopt coupling track algorithm and particle swarm optimization algorithm, select minority basis function rarefaction representation fault file class differences signal, that is:
Wherein
Be selected basis function, a
kBe rarefaction representation coefficient, a
kSimilarity degree between difference signal D (t) and the basis function has been described; Find the solution
a
kProcess as follows:
(a) establish residue signal r the 0th time
0(t)=and D (t), the initialization submodule is to the basis function parameters u
n, s
n, f
n, β
nEncode the initialization population;
(b) decompose submodule one by one and decompose residue signal one by one, establishing the p time residue signal is r
p(t), according to particle swarm optimization algorithm, select best particle { u
p, s
p, f
p, β
p, make basis function
With r
p(t) the most similar, then
Rarefaction representation coefficient a
pFor:
(c) decomposition is responsible for calculating the residue signal energy by the judgement submodule || r
p(t) || with || D (t) || ratio, when if this ratio is not less than default thresholding, then returns and decompose submodule one by one residue signal is proceeded to decompose one by one, if this ratio is during less than default thresholding, then stop to decompose, selected optimal base function is decomposed in output each time.
When described fault file database be 3 the classification more than the fault file database time, adopt differential method one by one, move above-mentioned method for diagnosing faults during to 2 classification fault file databases repeatedly, judge one by one whether signal to be checked belongs to each fault of fault file database.
Described differential method one by one, can explain as following Example: establishing the fault file database is the above fault file databases of 3 classification, and promptly fault has 3 classes: A, B, C, and new signal D; Judge that at first signal D belongs to A or do not belong to A; If do not belong to A, judge that so D belongs to B or do not belong to B; If D neither belongs to A and do not belong to B again, D belongs to C so.Method for diagnosing faults when each step 2 is classified the fault file database in fact need run the program of the method for diagnosing faults of above-mentioned 2 classification one time.One 3 classification problem just needs to run 2 times so, and the n classification problem just needs to run n-1 time.
The foregoing description is a preferred implementation of the present invention; but embodiments of the present invention are not restricted to the described embodiments; other any do not deviate from change, the modification done under spirit of the present invention and the principle, substitutes, combination, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.
Claims (4)
1, based on the automobile engine failure diagnosis system of rarefaction representation, it is characterized in that: comprising:
Be used to receive signal to be checked that the fault diagnosis terminal transmits and the Network Interface Module that postbacks diagnostic result;
Be used for control and management DSP module, Network Interface Module, IDE module and I/O module, the signal to be checked that reception is transmitted through Network Interface Module also generates Signals Data Base to be checked and deposits the IDE module in, the diagnostic result of DSP module output is deposited in the field programmable logic array (FPLA) FPGA module of the diagnostic result filing database of IDE module;
Be used to store the IDE module of engine sound signal fault archive database, Signals Data Base to be checked and diagnostic result filing database;
From the IDE module, read signal to be checked and diagnose, export the DSP module of diagnostic result then;
Be used to provide the I/O module of man-machine interaction;
Simultaneously, described FPGA module is connected with DSP module, IDE module, Network Interface Module, I/O module simultaneously, and the DSP module is connected with the IDE module simultaneously, and described Network Interface Module is connected with the fault diagnosis terminal signaling;
Described DSP module comprises:
Be used for the engine sound signal of engine sound signal fault archive database is carried out pre-service, and the signal to be checked in the Signals Data Base to be checked is carried out pretreated pretreatment module;
Be used for that pretreated fault file class differences signal is carried out adaptive sparse and decompose, to select the Sparse Decomposition module of optimal base function;
Be used to calculate the inner product of described optimal base function and fault file signal, to extract the fault signature extraction module of fault file feature;
Be used to utilize the sorting algorithm module of fault file features training sorting algorithm;
Be used to calculate the inner product of described optimal base function and signal to be checked, to generate the signal characteristic extraction module to be checked of signal characteristic to be checked;
Be used for sorting algorithm that signal characteristic substitution sorting algorithm module to be checked is trained, to carry out the fault diagnosis module of fault diagnosis;
Be used for according to signal characteristic to be checked fault diagnosis degree of confidence thresholding being set, judge whether believable diagnostic result analysis module of diagnostic result to the distance of the lineoid of classifying;
Be used for the diagnostic result analysis module is judged as the fault file database update module that believable signal to be checked adds the fault data file store;
Be used for the diagnostic result analysis module is judged as the Signals Data Base update module to be checked that incredible signal to be checked is retained in Signals Data Base to be checked;
Be used to verify the judging module of ending of Signals Data Base to be checked;
Be used for diagnostic result is sent to the diagnostic result output module of FPGA;
Simultaneously, described pretreatment module, the Sparse Decomposition module, the fault signature extraction module, the sorting algorithm module, fault diagnosis module, the diagnostic result analysis module, fault file database update module connects successively, described fault file database update module is connected with the Sparse Decomposition module simultaneously, described diagnostic result analysis module while and Signals Data Base update module to be checked, by judging module, the diagnostic result output module connects successively, described Sparse Decomposition module with after signal characteristic extraction module to be checked is connected respectively with fault diagnosis module, be connected by judging module.
2, according to the described automobile engine failure diagnosis system based on rarefaction representation of claim 1, it is characterized in that: described Sparse Decomposition module comprises:
The class differences submodule that is used for calculation engine fault file database 2 class signal average differences;
Be used to make up the super complete basic submodule of super complete base;
Adopt coupling track algorithm and particle swarm optimization algorithm, select the basis function chooser module of minority basis function rarefaction representation fault file class differences signal;
Be used for the parameter of selected basis function is encoded, with the initialization submodule of initialization population;
Be used for decomposing one by one the submodule of decomposition one by one of residue signal;
Be used to calculate the ratio of the average of decomposing the engine failure archive database class differences signal that the each residue signal energy that decomposes of submodule and class differences submodule calculated one by one, when if this ratio is not less than default thresholding, then returns and decompose decomposition that submodule proceeds to decompose one by one to residue signal one by one by the judgement submodule;
Simultaneously, described class differences submodule, super complete basic submodule simultaneously and after basis function chooser module is connected successively with the initialization submodule, one by one decompose submodule, decompose by adjudicating submodule and be connected.
3, a kind of method based on the automobile engine failure diagnosis of rarefaction representation of utilizing that each described system of claim 1~2 realizes is characterized in that, comprises the steps:
(1) after the system start-up, the FPGA module is responsible for dispatching and finishing the self check of IDE module, DSP module, Network Interface Module and I/O module;
(2) the FPGA module receives the response message of fault diagnosis terminal by Network Interface Module broadcast system initiation message, the record network topology structure, and be stored in the IDE module;
(3) the fault diagnosis terminal sends motor car engine voice signal to be checked to system, and this signal reaches the FPGA module by network interface, generates Signals Data Base to be checked, and is stored in the IDE module;
(4) the DSP module is called the fault file database in the IDE module, by pretreatment module the engine sound signal in the fault file database is carried out pre-service earlier; Then pretreated fault file class differences signal is carried out adaptive sparse and decompose, select the optimal base function by the Sparse Decomposition module; Calculate the inner product of selected basis function and fault file signal then by the fault signature extraction module, extract the fault file feature; By the sorting algorithm module, utilize fault file features training sorting algorithm at last;
(5) the DSP module is taken out signal to be checked from the IDE module, by pretreatment module signal to be checked is carried out pre-service earlier; Then by signal characteristic extraction module to be checked, the inner product of selected basis function of calculation procedure (4) and signal to be checked generates signal characteristic to be checked; By fault diagnosis module,, carry out fault diagnosis then with the sorting algorithm that the feature substitution step (4) of signal to be checked is trained; At last by the diagnostic result analysis module, according to the distance of signal characteristic to be checked to the lineoid of classifying, fault diagnosis degree of confidence thresholding is set, judge whether diagnostic result is credible, if credible, by fault file database update module, believable signal to be checked is added the fault file database, otherwise,, incredible signal to be checked is retained in Signals Data Base to be checked by Signals Data Base update module to be checked;
(6) the DSP module is by verifying Signals Data Base to be checked by judging module, if it be empty or in the front and back two-wheeled is diagnosed no change, then diagnostic result is gathered and is sent to the FPGA module; Otherwise the diagnosis of a new round is carried out in repeating step (4), (5);
(7) the FPGA module diagnostic result that will gather deposits the diagnostic result filing database in, and by Network Interface Module diagnostic result is sent to the fault diagnosis terminal.
4, the method for the automobile engine failure diagnosis based on rarefaction representation according to claim 3, it is characterized in that: when described fault file storehouse is 2 classification fault file storehouses, the described Sparse Decomposition module of step (4) is carried out adaptive sparse to pretreated fault file class differences signal and is decomposed, select the optimal base function, specifically may further comprise the steps:
A, class differences submodule calculate the difference D (t) of fault file storehouse 2 class signal averages, that is:
Wherein x (t) is a fault file storehouse engine sound signal, ω
1, ω
2The expression class formative, n
1, n
2Be respectively the quantity of two class signals;
Parameter b wherein
nThe normalization coefficient of expression basis function makes
And u
n, s
n, f
n, β
nThe peak value moment of representing basis function respectively, decay factor, frequency and initial phase;
C, basis function chooser module adopt coupling track algorithm and particle swarm optimization algorithm, select minority basis function rarefaction representation fault file storehouse class differences signal, that is:
Wherein
Be selected basis function, a
kBe rarefaction representation coefficient, a
kSimilarity degree between difference signal D (t) and the basis function has been described; Find the solution
a
kProcess as follows:
(a) establish residue signal r the 0th time
0(t)=and D (t), the initialization submodule is to the basis function parameters u
n, s
n, f
n, β
nEncode the initialization population;
(b) decompose submodule one by one and decompose residue signal one by one, establishing the p time residue signal is r
p(t), according to particle swarm optimization algorithm, select best particle { u
p, s
p, f
p, β
p, make basis function
With r
p(t) the most similar, then
Rarefaction representation coefficient a
pFor:
(c) decomposition is responsible for calculating the residue signal energy by the judgement submodule || r
p(t) || with || D (t) || ratio, when if this ratio is not less than default thresholding, then returns and decompose submodule one by one residue signal is proceeded to decompose one by one, if this ratio is during less than default thresholding, then stop to decompose, selected optimal base function is decomposed in output each time;
When described fault file database be 3 the classification more than the fault file database time, adopt differential method one by one, move above-mentioned method for diagnosing faults during to 2 classification fault file databases repeatedly, judge one by one whether signal to be checked belongs to each fault of fault file database.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810198342A CN100595548C (en) | 2008-09-05 | 2008-09-05 | Automotive engine fault diagnosis system and method based on sparse expression |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810198342A CN100595548C (en) | 2008-09-05 | 2008-09-05 | Automotive engine fault diagnosis system and method based on sparse expression |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101382468A CN101382468A (en) | 2009-03-11 |
CN100595548C true CN100595548C (en) | 2010-03-24 |
Family
ID=40462425
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN200810198342A Expired - Fee Related CN100595548C (en) | 2008-09-05 | 2008-09-05 | Automotive engine fault diagnosis system and method based on sparse expression |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100595548C (en) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI426242B (en) * | 2010-11-03 | 2014-02-11 | Ind Tech Res Inst | Diagnosing device and an associated method for a motor device |
CN102494894A (en) * | 2011-11-17 | 2012-06-13 | 高丙团 | Audio monitoring and fault diagnosis system for wind generating set and audio monitoring and fault diagnosis method for same |
CN102521613B (en) * | 2011-12-17 | 2014-01-08 | 山东省科学院自动化研究所 | Method for fault diagnosis of automobile electronic system |
CN102661866A (en) * | 2012-05-11 | 2012-09-12 | 天津工业大学 | Engine fault identification method based on time-domain energy and support vector machine |
JP5948418B2 (en) * | 2012-07-25 | 2016-07-06 | 株式会社日立製作所 | Abnormal sound detection system |
CN104182642B (en) * | 2014-08-28 | 2017-06-09 | 清华大学 | A kind of fault detection method based on rarefaction representation |
CN104598654A (en) * | 2014-10-07 | 2015-05-06 | 芜湖扬宇机电技术开发有限公司 | Ignition advance angle prediction system and method thereof |
CN105242110B (en) * | 2015-09-08 | 2018-04-10 | 广东电网有限责任公司东莞供电局 | A kind of power distribution network multiple spot harmonic pollution source tracing method |
CN106527403B (en) * | 2016-12-13 | 2019-01-08 | 象翌微链科技发展有限公司 | Vehicular intelligent diagnostic method and device |
CN110360024A (en) * | 2019-07-29 | 2019-10-22 | 西北工业大学 | A kind of airborne trouble-shooter of rocket engine based on FPGA+DSP |
CN110763487A (en) * | 2019-11-08 | 2020-02-07 | 江苏食品药品职业技术学院 | Remote intelligent diagnosis system for automobile detection line |
CN111999064A (en) * | 2020-06-30 | 2020-11-27 | 中国航发南方工业有限公司 | Engine health monitoring unit |
-
2008
- 2008-09-05 CN CN200810198342A patent/CN100595548C/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN101382468A (en) | 2009-03-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100595548C (en) | Automotive engine fault diagnosis system and method based on sparse expression | |
CN112161784B (en) | Mechanical fault diagnosis method based on multi-sensor information fusion migration network | |
CN101770219B (en) | Knowledge acquisition method of fault diagnosis knowledge library of turn-milling combined machine tool | |
CN100485342C (en) | Integrated supporting vector machine mixed intelligent diagnosing method for mechanical fault | |
CN102507230B (en) | Method for diagnosing fault of automatic ammunition supply and transportation device | |
CN102819239A (en) | Intelligent fault diagnosis method of numerical control machine tool | |
CN104929864A (en) | Field programmable gate array (FPGA)-based embedded type operating state monitoring and fault diagnosis system for wind generating set | |
CN201666248U (en) | Natural gas compressor state detection and failure predication system | |
CN102103036A (en) | Automobile engine fault diagnosis system based on vibration and audio frequency information | |
CN111504385A (en) | Multi-parameter monitoring device and method suitable for abnormal state of mechanical equipment | |
CN111810395B (en) | Reciprocating compressor energy efficiency analysis system based on P-V graph test | |
Zhang et al. | A novel denoising algorithm based on TVF-EMD and its application in fault classification of rotating machinery | |
CN111878322B (en) | Wind power generator device | |
CN114893360B (en) | Method and system for identifying abnormal vibration of tower of wind turbine generator and monitoring running state | |
CN115979661A (en) | Marine diesel engine fault diagnosis method based on deep learning and improved evidence fusion theory | |
CN116956215A (en) | Fault diagnosis method and system for transmission system | |
CN116625683A (en) | Wind turbine generator system bearing fault identification method, system and device and electronic equipment | |
CN114611940A (en) | Decision-making method and system for multi-source coal-based solid waste resource utilization direction | |
CN117392824A (en) | Steel structure monitoring and early warning platform of offshore wind power booster station | |
CN110377009B (en) | Method and system for automobile fault diagnosis | |
CN112270209A (en) | Method for extracting vibration signal characteristics of threshing cylinder of combine harvester | |
Ren et al. | Research on fault feature extraction of hydropower units based on adaptive stochastic resonance and fourier decomposition method | |
CN101071159A (en) | Electric automobile energy collecting system and method | |
Joshuva et al. | Logistic model tree classifier for condition monitoring of wind turbine blades | |
CN112799378A (en) | Diagnosis and simulation method for hardware signal of wind generating set |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100324 Termination date: 20150905 |
|
EXPY | Termination of patent right or utility model |