Fuel injector fault detection method and device, computer equipment and storage medium

Abstract

The invention provides a fuel injector fault detection method and device, computer equipment and a storage medium, and the fuel injector fault detection method comprises the steps: obtaining a first sequence of a high-pressure fuel pipe of a fuel injector for any one of a plurality of fuel injectors of a fuel system; the first sequence comprises a plurality of first vibration signals arranged according to a time sequence; constructing a first matrix according to every m adjacent first vibration signals in the first sequence; the row number of the first matrix is N-m + 1, the column number of the first matrix is m, and N is the number of the plurality of first vibration signals; calculating the cosine similarity between every two adjacent rows in the first matrix to obtain a plurality of cosine values; calculating a first feature vector of the first sequence according to the plurality of cosine values; and performing fault detection on the oil injector according to the first feature vector. According to the embodiment of the invention, the fault feature information can be extracted more comprehensively and effectively, so that the fault diagnosis precision of the fuel injector is improved.

Description

Technical Field

[0001] The present application relates to the technical field of data processing, and particularly relates to a method, device, computer device and storage medium for detecting injector faults. Background Art

[0002] Diesel engines are widely used in fields such as marine power, construction machinery, and agricultural equipment due to their significant advantages such as high thermal efficiency, wide power range, good economy, and high reliability. As a core component of the fuel system, the health status of the injector directly determines the quality and pressure of fuel supply, and minor faults in the injector will greatly affect the performance of the diesel engine, such as efficiency and emissions, and even lead to serious accidents such as shutdowns. Therefore, realizing intelligent and efficient diagnosis of injector faults is of great significance for preventing catastrophic faults and improving the reliability of equipment operation.

[0003] Considering the complex mechanical structure and harsh working environment of the marine fuel system, it is undoubtedly a complex system that is multidisciplinary, nonlinear, and multivariable. Although the fuel pressure fluctuation in the common rail pipeline can directly characterize the state of the injector injection process and has less interference information compared with traditional vibration signals, it is still challenging to accurately extract the fault characteristics of the injector. The information entropy method is a powerful nonlinear signal analysis method that can quantify the characteristic information of a signal sequence from the perspectives of complexity and stability, and has also been proven to be a more effective method for extracting fault characteristic information. Due to its outstanding advantages such as not relying on prior knowledge, not requiring preprocessing techniques such as regularization and filtering, and having few preset parameters, this method has also been continuously introduced into the field of mechanical equipment state detection and fault diagnosis.

[0004] However, the existing information entropy method has problems of poor extraction and low fault diagnosis accuracy when extracting injector fault characteristics. Summary of the Invention

[0005] In view of this, the present application proposes a method, device, computer device and storage medium for detecting injector faults to solve the problems of poor extraction and low fault diagnosis accuracy when extracting injector fault characteristics in the related art.

[0006] The first aspect of the embodiments of the present application proposes a method for detecting injector faults, and the method includes:

[0007] For any one of multiple injectors in the fuel system, obtain a first sequence of the high-pressure fuel pipe of the injector; the first sequence includes multiple first vibration signals arranged in chronological order;

[0008] According to every adjacent m first vibration signals in the first sequence, construct a first matrix; the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the multiple first vibration signals;

[0009] Calculate the cosine similarity between each adjacent two rows in the first matrix to obtain a plurality of cosine values;

[0010] Calculate the first eigenvector of the first sequence according to the plurality of cosine values;

[0011] Perform fault detection on the fuel injector according to the first eigenvector.

[0012] In a specific embodiment of the present application, by acquiring the wall vibration signals of the high-pressure fuel pipe arranged in chronological order, constructing the first matrix, calculating the cosine similarity between each adjacent two rows in the first matrix, and calculating the first eigenvector of the first sequence according to the plurality of cosine values, it is possible to measure the dynamic complexity change of the time series at each scale. The decentralized cosine similarity method fully considers the similarity of the system mode distribution in terms of direction and value, and can extract fault feature information more comprehensively and effectively.

[0013] In an embodiment of the present application, the method further includes:

[0014] For any one of the plurality of scale factors, decompose the first sequence according to the scale factor to obtain a plurality of second sequences; each second sequence contains a plurality of adjacent first vibration signals corresponding to the scale factor;

[0015] Calculate the variance value of each second sequence to obtain a plurality of variance values corresponding to the plurality of second sequences one by one;

[0016] Calculate the second eigenvector corresponding to the scale factor according to the plurality of variance values;

[0017] Calculate the third eigenvector of the first sequence according to the plurality of second eigenvectors corresponding to the plurality of scale factors one by one and the number of scale factors.

[0018] In an embodiment of the present application, performing fault detection on the fuel injector according to the first eigenvector includes:

[0019] Take the average value of the first eigenvector and the third eigenvector to obtain a target eigenvector;

[0020] Perform fault detection on the fuel injector according to the target eigenvector.

[0021] In an embodiment of the present application, performing fault detection on the fuel injector according to the target eigenvector includes:

[0022] Divide the plurality of target eigenvectors into training samples and test samples;

[0023] Input the training samples into a neural network model for model training to obtain a trained classifier; during the model training process, optimize the parameters of the neural network model through a preset search algorithm;

[0024] Perform fault detection on the fuel injector through the trained classifier.

[0025] In the embodiment of the present application, optimizing the parameters of the neural network model through a preset search algorithm includes:

[0026] Initialize the neural network model to obtain multiple groups of model parameters;

[0027] Calculate the prediction results of the neural network model corresponding to each group of model parameters, and calculate the error index between the prediction results and the training data as the fitness value;

[0028] Update the model parameters according to the fitness value; the discoverer is responsible for exploring the new solution space, and the follower is responsible for following the discoverer to search;

[0029] Repeat the above process until the optimal network parameters of the neural network model are output when the maximum number of iterations is reached.

[0030] In the embodiment of the present application, the preset search algorithm is the sparrow search algorithm; among them, the number of sparrows is 10, the maximum number of iterations is 20, the discoverer ratio is 0.5, and the follower ratio is 0.3.

[0031] In the embodiment of the present application, calculating the first feature vector of the first sequence according to the multiple cosine values includes:

[0032] Divide the value range of the multiple cosine values into multiple intervals;

[0033] For any one of the multiple intervals, count the number of cosine values within the interval;

[0034] Take the ratio of the number to the total number of the multiple cosine values as the state probability of the interval;

[0035] Calculate the first feature vector of the first sequence according to the multiple state probabilities corresponding to the multiple intervals one by one.

[0036] An embodiment of the second aspect of the present application provides a fuel injector fault detection device, including:

[0037] A sequence acquisition module, configured to acquire a first sequence of the high-pressure fuel pipe of any one of the multiple fuel injectors in the fuel system; the first sequence includes multiple first vibration signals arranged in chronological order;

[0038] A matrix construction module, configured to construct a first matrix according to every adjacent m first vibration signals in the first sequence; the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the multiple first vibration signals;

[0039] A cosine value calculation module, configured to calculate the cosine similarity between every adjacent two rows in the first matrix to obtain a plurality of cosine values;

[0040] A feature vector calculation module, configured to calculate a first feature vector of the first sequence according to the plurality of cosine values;

[0041] A fault detection module, configured to perform fault detection on the fuel injector according to the first feature vector.

[0042] An embodiment of the third aspect of the present application provides a computer device, which includes a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the fuel injector fault detection method described in the first aspect above.

[0043] An embodiment of the fourth aspect of the present application provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to cause a computer to execute the fuel injector fault detection method described in the first aspect above.

[0044] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present application. Description of the Drawings

[0045] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present application. Moreover, throughout the drawings, the same reference numerals are used to represent the same components.

[0046] In the drawings:

[0047] Figure 1 A flowchart showing a fuel injector fault detection method provided by an embodiment of the present application;

[0048] Figure 2 A flowchart showing a fuel injector fault diagnosis method based on GCMIDE and SSA - BP provided by an embodiment of the present application;

[0049] Figure 3 A flowchart comparison diagram showing the GCMIDE algorithm and the MIDE algorithm provided by an embodiment of the present application;

[0050] Figure 4 Shows the flowchart of the SSA - BP algorithm provided by an embodiment of the present application;

[0051] Figure 5 Shows the comparison chart of the characteristics of different injector fault types extracted by the GCMIDE algorithm provided by an embodiment of the present application;

[0052] Figure 6 Shows the comparison chart of the characteristics of different injector fault degrees extracted by the GCMIDE algorithm provided by an embodiment of the present application;

[0053] Figure 7 Shows the comparison chart of the diagnostic results of the injector fault types by the GCMIDE algorithm provided by an embodiment of the present application under different feature selection quantities;

[0054] Figure 8 Shows the comparison chart of the diagnostic results of the injector fault degrees by the GCMIDE algorithm provided by an embodiment of the present application under different feature selection quantities;

[0055] Figure 9 Shows the comparison chart of the diagnostic results of the injector fault degrees by the GCMIDE algorithm provided by an embodiment of the present application under different training set ratios;

[0056] Figure 10 Shows the structural schematic diagram of an injector fault detection device provided by an embodiment of the present application;

[0057] Figure 11 Shows the structural schematic diagram of a computer device provided by an embodiment of the present application;

[0058] Figure 12 Shows the schematic diagram of a storage medium provided by an embodiment of the present application. Detailed implementation manners

[0059] Hereinafter, the exemplary embodiments of the present application will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present application can be more thoroughly understood and the scope of the present application can be completely conveyed to those skilled in the art.

[0060] It should be noted that unless otherwise specified, the technical terms or scientific terms used in the present application should have the ordinary meanings understood by those skilled in the art to which the present application belongs.

[0061] The following describes the technical scenarios involved in the embodiments of the present application.

[0062] Considering the complex mechanical structure and harsh working environment of the marine fuel system, it is undoubtedly a complex system with multiple disciplines, nonlinearity, and multiple variables. Although the fuel pressure fluctuation in the common rail pipeline can directly characterize the state of the fuel injection process of the injector and has less interference information compared with traditional vibration signals, it is still challenging to accurately extract the fault characteristics of the injector. The information entropy method is a powerful nonlinear signal analysis method that can quantify the characteristic information of a signal sequence from the perspectives of complexity and stability, and has also been proven to be a more effective method for extracting fault characteristic information. Due to its prominent advantages such as not relying on prior knowledge, not requiring preprocessing techniques such as regularization and filtering, and having few preset parameters, this method has also been continuously introduced into the field of mechanical equipment state detection and fault diagnosis. In recent years, many new entropy methods have been proposed and introduced into mechanical fault diagnosis, such as sample entropy (SE), fuzzy entropy (FE), permutation entropy (PE), dispersion entropy (DispE), etc. Unfortunately, these existing entropy methods all have some common disadvantages, namely low computational efficiency and undefined values in some specific systems or due to short-time series analysis. In addition, the entropy values of some specific systems do not match their dynamic complexity, which is inconsistent with the basic meaning of entropy. Wang et al. discussed this in detail and proposed the diversity entropy (DE) method, which is different from other existing information entropy methods and has three advantages: high consistency, high computational efficiency, and high robustness. However, DE only considers the angular relationship when calculating the cosine similarity between adjacent orbits and does not consider the magnitudes and influences of each dimension. Considering that the higher the orbit dimension, the more different orbit states with the same included angle, DE cannot well distinguish multiple different state distributions with the same included angle, so it will greatly affect the sensitivity and accuracy of DE in analyzing the change of signal complexity.

[0063] To solve the above related technical problems, the present application provides an injector fault diagnosis method based on GCMIDE and SSA - BP, that is, the above injector fault detection method, which is beneficial to solving the problems of poor extraction of injector fault characteristics and low fault diagnosis accuracy in the diesel engine fuel system under the background of a complex and harsh working environment.

[0064] According to an embodiment of the present application, an embodiment of an injector fault detection method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0065] Embodiment 1:

[0066] In this embodiment, an injector fault detection method is provided. Figure 1is a flowchart of an injector fault detection method according to an embodiment of the present application. As Figure 1 shown, the process includes the following steps:

[0067] Step S101, for any one of a plurality of injectors in a fuel system, obtain a first sequence of the high-pressure fuel pipe of the injector.

[0068] In an embodiment of the present application, the fuel system includes a plurality of injectors, and each injector nozzle is connected to a corresponding high-pressure fuel pipe. In the present application, a clamping pressure sensor with high precision and strong anti-interference ability is fixed at each high-pressure fuel pipe to obtain the wall vibration signal of the high-pressure fuel pipe (including the wall vibration signals of the high-pressure fuel pipe in the normal state and different fault states of the injector).

[0069] In an embodiment of the present application, the first sequence includes a plurality of first vibration signals arranged in chronological order. For example: X = {x1, x2,..., x N}, where x N represents the wall vibration signal at the Nth moment. The meaning of the moment is not specifically limited and can be understood as seconds or milliseconds.

[0070] Step S102, construct a first matrix according to every adjacent m first vibration signals in the first sequence.

[0071] In an embodiment of the present application, the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the plurality of first vibration signals. Among them, m can be artificially set according to the actual situation.

[0072] In an embodiment of the present application, for the first sequence X = {x1, x2,..., x N}, where the number of first vibration signals included in the first sequence is N. When the number of rows of the first matrix is set to m, the first sequence is reconstructed to obtain N - m + 1 vectors {y i (m)} as follows:

[0073] yi(m) = {xi, xi +τ ,..., xi + (m - 1)τ}, 1 ≤ i ≤ N - m + 1 (1)

[0074] {y i (m)} constitutes a sequence Y(m) = {y1(m), y2(m),... y N-m+1 (m)}.

[0075] Among them, τ can be understood as the delay time. For example, x1 is the pipe wall vibration signal collected at the first moment, x2 is the pipe wall vibration signal collected at the second moment, and the time difference between the first moment and the second moment is the above-mentioned delay time. The delay time is not specifically limited and can be in seconds or milliseconds.

[0076] Each {y i (m)} can be understood as the first vibration signal included in the corresponding row of the first matrix. An example is given to illustrate this:

[0077] When N = 10 and m = 3, when the number of rows of the first matrix is N - m + 1 = 8, the first matrix is:

[0078]

[0079] When i = 1, {y1(m)} = {x1 x2 x3}

[0080] When i = 2, {y2(m)} = {x2 x3 x4} ......

[0082] Step S103, calculate the cosine similarity between each adjacent two rows in the first matrix to obtain a plurality of cosine values.

[0083] Among them, the cosine similarity refers to the de-centered cosine similarity.

[0084] In the embodiments of the present application, the cosine similarity between each adjacent two rows in the first matrix can be calculated through the following calculation formula to obtain a plurality of cosine values:

[0085]

[0086]

[0087] Step S104, calculate the first eigenvector of the first sequence according to the plurality of cosine values;

[0088] In some specific embodiments, the above step S104 includes steps S1041 - S1044:

[0089] Step S1041, divide the value range of the plurality of cosine values into a plurality of intervals.

[0090] Specifically, divide the value range interval [-1, 1] of D(m) into e equal sub-intervals.

[0091] Step S1042, for any one of the plurality of intervals, count the number of cosine values within the interval.

[0092] Specifically, for example: D(m) = {d1, d2, …, d N-m}, if the cosine values within the first equal sub - interval include d1, d3, d5, then the number of cosine values contained in the first equal sub - interval is 3.

[0093] Step S1043: Take the ratio of the number to the total number of the multiple cosine values as the state probability of the interval.

[0094] Specifically, for example: When N = 10 and m = 3, the total number of cosine values is 10 - 3 = 7, then the state probability P1 of the first equal sub - interval is 3 / 7.

[0095] Step S1044: Calculate the first eigenvector of the first sequence according to the multiple state probabilities corresponding to the multiple intervals one by one.

[0096] Specifically, when obtaining the multiple state probabilities {P1, P2, …, P e} corresponding to the multiple intervals through the above - mentioned steps S1042 - step S1043, the first eigenvector IDE(X, m, τ, e) of the first sequence can be calculated by the following formula:

[0097]

[0098] In some specific embodiments, the method further includes steps S201 - step S204:

[0099] Step S201: For any one of the multiple scale factors, decompose the first sequence according to the scale factor to obtain multiple second sequences.

[0100] In the embodiments of the present application, the scale factor can be artificially set according to the actual situation. For example, the scale factor can be 5.

[0101] In some specific embodiments, after setting the scale factor, it is necessary to calculate the number of the decomposed multiple second sequences according to the scale factor and the first sequence X = {x1, x2, …, x N} through the following formula:

[0102]

[0103] Where j represents the number of the decomposed second sequences, N represents the number of the first vibration signals included in the first sequence, and s represents the scale factor.

[0104] When the scale factor s is 5 and the number N of the first vibration signals included in the first sequence is 10, the number j of the decomposed second sequences is 10 / 5 = 2. For example, the first sequence X = {x1, x2, …, x 10} can be decomposed into:

[0105]

[0106] Among them, represents the first second sequence obtained by decomposition when the scale factor s is 5.

[0107]

[0108] Among them, represents the second second sequence obtained by decomposition when the scale factor s is 5.

[0109] In some specific embodiments, after setting the scale factor and decomposing the first sequence X = {x1, x2,..., x N} into multiple second sequences by using the set scale factor, it is also necessary to consider the second sequences obtained by decomposing with other scale factors smaller than the set scale factor (when the set scale factor is 5, the other scale factors are 1 - 5) Among them, s represents the set scale factor, k represents other scale factors smaller than the set scale factor s, and j represents the number of second sequences obtained by decomposing the first sequence with other scale factors k. For example:

[0110] When the set scale factor s is 5, the other scale factor k is 2, and the number N of the first vibration signals included in the first sequence is 10, the number j of the second sequences obtained by decomposing the first sequence with other scale factor k is 10 / 2 = 5. The first sequence X = {x1, x2,..., x 10} can be decomposed into:

[0111]

[0112]

[0113] Among them, represents the first second sequence obtained by decomposing the first sequence with other scale factor k = 2, represents the second second sequence obtained by decomposing the first sequence with other scale factor k = 2, and so on for others.

[0114] Step S202, calculate the variance value of each second sequence to obtain multiple variance values corresponding to the multiple second sequences one by one.

[0115] Specifically, the variance value of each second sequence can be calculated by the following formula:

[0116]

[0117] Based on the following formula, the variance value can be obtained by first calculating the average value, then calculating the square of the difference between each first vibration signal and the average value, and finally calculating the average value of multiple mean values.

[0118] Step S203: Calculate the second eigenvector corresponding to the scale factor according to the multiple variance values.

[0119] Step S204: Calculate the third eigenvector of the first sequence according to the multiple second eigenvectors corresponding one by one to the multiple scale factors and the number of scale factors.

[0120] In the above steps S203 - S204, the third eigenvector of the first sequence can be calculated by the following formula:

[0121]

[0122] where GCMIDE(X, m, τ, e, s) represents the third eigenvector of the first sequence. can represent the variance value of the j - th second sequence obtained by decomposing the first sequence by other scale factor k when the scale factor is s.

[0123] Step S105: Perform fault detection on the fuel injector according to the first eigenvector.

[0124] In the application embodiment, the average value of the first eigenvector and the third eigenvector is taken to obtain the target eigenvector; and fault detection is performed on the fuel injector according to the target eigenvector.

[0125] In some specific embodiments, performing fault detection on the fuel injector according to the target eigenvector includes steps S301 - S303:

[0126] Step S301: Divide multiple target eigenvectors into training samples and test samples.

[0127] Before step S301, the Relieff algorithm can be used to perform feature selection on the target eigenvector to improve the quality of the target eigenvector and the influence of the number of selected features on the diagnostic accuracy.

[0128] Step S302: Input the training samples into a neural network model for model training to obtain a trained classifier.

[0129] In the embodiments of the present disclosure, all training samples can be input into the BP network model for training, and the SSA is used to optimize the parameters of the BP neural network; the parameters of the sparrow search algorithm are as follows: the number of sparrows is 10, the maximum number of iterations is 20, the proportion of discoverers is 0.5, the proportion of followers is 0.3, and the root mean square error between the prediction result and the training data is used as the fitness value.

[0130] In the embodiments of the present disclosure, during the model training process, the parameters of the neural network model are optimized through a preset search algorithm (such as the sparrow search algorithm), including:

[0131] Step S3021, initialize the neural network model to obtain multiple groups of model parameters;

[0132] Step S3022, calculate the prediction result of the neural network model corresponding to each group of model parameters, and calculate the error index between the prediction result and the training data as the fitness value;

[0133] Step S3023, update the model parameters according to the fitness value; the discoverers are responsible for exploring new solution spaces, and the followers are responsible for following the discoverers to search;

[0134] Step S3024, repeat the above process until the optimal network parameters of the neural network model are output when the maximum number of iterations is reached.

[0135] Step S303, perform fault detection on the fuel injector through the trained classifier.

[0136] Specifically, the trained SSA-BP classifier is used to diagnose the faults of the fuel injector components in the diesel fuel system for the test samples.

[0137] Embodiment 2:

[0138] The embodiments of the present application provide a fuel injector fault diagnosis method and device based on GCMIDE and SSA-BP. The overall technical framework is as Figure 2 shown, including the following steps:

[0139] S1. A high-precision and strong anti-interference clamping pressure sensor is fixed at the high-pressure fuel pipe of the fuel system to obtain the vibration signal of the high-pressure fuel pipe wall. The vibration signals in the normal state and different fault states of the fuel injector are obtained and converted into electrical signals. Taking the high-pressure common rail fuel injector component of the diesel fuel system as an example, the fuel pressure fluctuation signals of the high-pressure fuel pipe in the normal state and different fuel injector fault states are collected.

[0140] S2. The electrical signals converted by the vibration sensor in step 1 are digitally processed by a high-performance FPGA and DSP integrated with an adaptive filter and transmitted to the upper computer.

[0141] S3. The intelligent fault diagnosis algorithm of the host computer system completes fault diagnosis. Specifically, it includes the following processes: calculating the generalized composite multi-scale improved diversity entropy of the original signal, and using GCMIDE as the feature vector. The calculation steps of the generalized composite multi-scale improved diversity entropy are as follows, and its algorithm flow chart is as Figure 3 shown:

[0142] The first step: For a given time series X = {x1, x2,..., x N} with a length of N, when the embedding dimension is m, the original time series is reconstructed to obtain N - m + 1 vectors {y i (m)} as follows:

[0143] yi(m) = {xi, xi +τ , …, xi + (m - 1)τ}), 1 ≤ i ≤ N - m + 1.

[0144] {y i (m)} constitutes the sequence Y(m) = {y1(m), y2(m), … y N-m+1 (m)}.

[0145] The second step: By calculating the decentralized cosine similarity between adjacent trajectories, a series of decentralized cosine values D(m) are obtained, and its calculation formula is:

[0146] D(m) = {d1, d2, …, d N-m}

[0147] = {d(y1(m), y2(m)),

[0148] d(y2(m), y3(m)), …, d(y N-m (m), y N-m+1 (m))}.

[0149]

[0150] The third step: Divide the value range interval [-1, 1] of D(m) into e equal sub-intervals, and calculate the ratio of the number of decentralized cosine similarity values in different intervals to the total number of decentralized cosine similarity values as the state probability {P1, P2, … P e} to construct the improved diversity entropy algorithm:

[0151]

[0152] The fourth step: Combine the generalized composite multi-scale decomposition method with the improved diversity entropy, and use the variance composite coarse-graining method to obtain all decomposed subsequences at each scale factor s Its calculation formula is:

[0153]

[0154] Step 5: Under the scale factor s, calculate the average improved diversity entropy value of the generalized composite coarse-grained sequence to obtain the generalized composite multi-scale improved diversity entropy value at scale s:

[0155]

[0156] S4. The algorithm flowchart of SSA-optimized BP is as Figure 4 shown. The comparison diagram of GCMIDE eigenvalues of the high-pressure fuel pipe pressure fluctuation signal under different injector states is as Figure 5 、 Figure 6 shown. Use the Relieff algorithm to perform necessary feature selection to obtain the final feature vector, and divide the calculated GCMIDE feature vector into a training set and a test set. The GCMIDE of all training samples is input into the BP network model for training, and use SSA to optimize the parameters of the BP neural network; Sparrow Search Algorithm parameters: the number of sparrows is 10, the maximum number of iterations is 20, the proportion of discoverers is 0.5, the proportion of followers is 0.3, and the root mean square error between the prediction result and the training data is used as the fitness value.

[0157] S5. Use the trained SSA-BP classifier to perform fault diagnosis on the test samples of the diesel engine fuel system injector components to determine whether an injector fault occurs, as well as the location and degree of the fault, so as to output the injector fault diagnosis result. The comparison diagrams of the fault diagnosis results output under different feature selection numbers and different training ratios are as Figure 7 、 Figure 8 and Figure 9 shown.

[0158] The technical solution of the present invention has the following advantages: A method and device for injector fault diagnosis based on GCMIDE and SSA-BP are proposed, which is used to measure the dynamic complexity changes of time series at each scale. The decentralized cosine similarity method fully considers the similarity of the system mode distribution in terms of direction and value, and can extract fault feature information more comprehensively and effectively. Use the Relieff algorithm to perform feature selection on the feature vector calculated by GCMIDE, and the quality of the features extracted by GCMIDE and the influence of the number of selected features on the diagnostic accuracy can be studied. Use BP optimized by the SSA algorithm for pattern recognition, which can avoid falling into local optima and improve the fault diagnosis accuracy.

[0159] Example 3:

[0160] Corresponding to the implementation manner of the above fuel injector fault detection method, an embodiment of the present application further provides a fuel injector fault detection device for performing the fuel injector fault detection method described in the above embodiment. As Figure 10 shown, the fuel injector fault detection device includes:

[0161] A sequence acquisition module, configured to acquire a first sequence of a high-pressure fuel pipe of any one of a plurality of fuel injectors in a fuel system; the first sequence includes a plurality of first vibration signals arranged in chronological order;

[0162] A matrix construction module, configured to construct a first matrix according to every adjacent m first vibration signals in the first sequence; the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the plurality of first vibration signals;

[0163] A cosine value calculation module, configured to calculate the cosine similarity between every adjacent two rows in the first matrix to obtain a plurality of cosine values;

[0164] A feature vector calculation module, configured to calculate a first feature vector of the first sequence according to the plurality of cosine values;

[0165] A fault detection module, configured to perform fault detection on the fuel injector according to the first feature vector.

[0166] Optionally, the device further includes:

[0167] A sequence decomposition module, configured to decompose the first sequence according to any one of a plurality of scale factors to obtain a plurality of second sequences; each second sequence includes a plurality of adjacent first vibration signals corresponding to the scale factor;

[0168] A variance calculation module, configured to calculate the variance value of each second sequence to obtain a plurality of variance values corresponding to the plurality of second sequences one by one;

[0169] A second feature vector calculation module, configured to calculate a second feature vector corresponding to the scale factor according to the plurality of variance values;

[0170] A third feature vector calculation module, configured to calculate a third feature vector of the first sequence according to the plurality of second feature vectors corresponding to the plurality of scale factors one by one and the number of scale factors.

[0171] Optionally, the fault detection module is further configured to take an average of the first feature vector and the third feature vector to obtain a target feature vector; and perform fault detection on the fuel injector according to the target feature vector.

[0172] Optionally, the fault detection module is further configured to divide the multiple target feature vectors into training samples and test samples; input the training samples into a neural network model for model training to obtain a trained classifier; during the model training process, optimize the parameters of the neural network model through a preset search algorithm; and perform fault detection on the fuel injector through the trained classifier.

[0173] Optionally, the fault detection module is further configured to initialize the neural network model to obtain multiple sets of model parameters; calculate the prediction results of the neural network models corresponding to each set of model parameters, and calculate the error index between the prediction results and the training data as the fitness value; update the model parameters according to the fitness value; the discoverer is responsible for exploring a new solution space, and the follower is responsible for following the discoverer to perform a search; repeat the above process until the optimal network parameters of the neural network model are output when the maximum number of iterations is reached.

[0174] Optionally, the feature vector calculation module is further configured to divide the value range of the multiple cosine values into multiple intervals; for any one of the multiple intervals, count the number of cosine values within the interval; use the ratio of the number to the total number of the multiple cosine values as the state probability of the interval; and calculate the first feature vector of the first sequence according to the multiple state probabilities corresponding to the multiple intervals one by one.

[0175] The fuel injector fault detection device provided in the above embodiments of the present application and the fuel injector fault detection method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run, or implemented by the application programs stored therein.

[0176] The embodiments of the present application also provide a computer device for executing the above fuel injector fault detection method. Please refer to Figure 11 which shows a schematic diagram of a computer device provided in some embodiments of the present application. As Figure 11 shown, the computer device 11 includes: a processor 1100, a memory 1101, a bus 1102, and a communication interface 1103. The processor 1100, the communication interface 1103, and the memory 1101 are connected through the bus 1102; a computer program that can run on the processor 1100 is stored in the memory 1101, and when the processor 1100 runs the computer program, it executes the fuel injector fault detection method provided in the foregoing embodiments of the present application.

[0177] Among them, the memory 1101 may include high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk memory. The communication connection between this system network element and at least one other network element is realized through at least one communication interface 1103 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used.

[0178] The bus 1102 can be an ISA bus, a PCI bus, an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. Among them, the memory 1101 is used to store a program. After receiving an execution instruction, the processor 1100 executes the program. The fuel injector fault detection method disclosed in the foregoing embodiments can be applied to the processor 1100 or implemented by the processor 1100.

[0179] The processor 1100 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit in hardware or instructions in software form in the processor 1100. The above-mentioned processor 1100 can be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers, etc. This storage medium is located in the memory 1101, and the processor 1100 reads the information in the memory 1101 and combines its hardware to complete the steps of the above method.

[0180] The computer device provided by the embodiments of the present application and the fuel injector fault detection method provided by the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run, or implemented by them.

[0181] An embodiment of the present application also provides a computer-readable storage medium corresponding to the fuel injector fault detection method provided in the foregoing embodiment. Please refer to Figure 12 , which shows that the computer-readable storage medium is an optical disc 30, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it will execute the fuel injector fault detection method provided in any of the foregoing embodiments.

[0182] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other optical and magnetic storage media, which will not be elaborated here one by one.

[0183] The computer-readable storage medium provided in the above embodiments of the present application and the fuel injector fault detection method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run, or implemented by the application programs stored therein.

[0184] It should be noted that:

[0185] In the specification provided here, a large number of specific details are described. However, it can be understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this specification.

[0186] Similarly, it should be understood that, in order to streamline the present application and help understand one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, the various features of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be construed as reflecting the following schematic diagram: that is, the claimed present application requires more features than those expressly recited in each claim. Rather, as reflected in the following claims, the inventive aspects lie in less than all the features of the single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into the detailed description, where each claim itself serves as a separate embodiment of the present application.

[0187] In addition, those skilled in the art can understand that although some of the embodiments described herein include certain features included in other embodiments rather than other features, the combination of features of different embodiments means that it is within the scope of the present application and forms different embodiments. For example, in the following claims, any one of the claimed embodiments can be used in any combination.

[0188] As described above, the above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for detecting injector faults, characterized in that, The method includes: For any one of a plurality of fuel injectors in a fuel system, obtaining a first sequence of the high-pressure fuel pipe of the fuel injector; the first sequence includes a plurality of first vibration signals arranged in chronological order; According to every adjacent m first vibration signals in the first sequence, constructing a first matrix; the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the plurality of first vibration signals; Calculating the cosine similarity between every adjacent two rows in the first matrix to obtain a plurality of cosine values; Calculating a first eigenvector of the first sequence according to the plurality of cosine values; Performing fault detection on the fuel injector according to the first eigenvector.

2. The method according to claim 1, wherein The method further includes: For any one of a plurality of scale factors, decomposing the first sequence according to the scale factor to obtain a plurality of second sequences; each second sequence contains a plurality of adjacent first vibration signals corresponding to the scale factor; Calculating the variance value of each second sequence to obtain a plurality of variance values corresponding one by one to the plurality of second sequences; Calculating a second eigenvector corresponding to the scale factor according to the plurality of variance values; Calculating a third eigenvector of the first sequence according to the plurality of second eigenvectors corresponding one by one to the plurality of scale factors and the number of scale factors.

3. The method according to claim 2, wherein Performing fault detection on the fuel injector according to the first eigenvector, including: Taking the average of the first eigenvector and the third eigenvector to obtain a target eigenvector; Performing fault detection on the fuel injector according to the target eigenvector.

4. The method according to claim 3, wherein Performing fault detection on the fuel injector according to the target eigenvector, including: Dividing a plurality of target eigenvectors into training samples and test samples; Inputting the training samples into a neural network model for model training to obtain a trained classifier; during the model training process, optimizing the parameters of the neural network model through a preset search algorithm; Performing fault detection on the fuel injector through the trained classifier.

5. The method according to claim 4, wherein Optimizing the parameters of the neural network model through a preset search algorithm, including: Initializing the neural network model to obtain multiple groups of model parameters; Calculating the prediction results of the neural network model corresponding to each group of model parameters, and calculating the error index between the prediction results and the training data as the fitness value; Updating the model parameters according to the fitness value; the discoverer is responsible for exploring a new solution space, and the follower is responsible for following the discoverer to search; Repeating the above process until the optimal network parameters of the neural network model are output when the maximum number of iterations is reached.

6. The method according to claim 4 or 5, characterized in that, The preset search algorithm is a sparrow search algorithm; wherein, the number of sparrows is 10, the maximum number of iterations is 20, the discoverer ratio is 0.5, and the follower ratio is 0.

3.

7. The method according to claim 1, characterized in that Calculating a first eigenvector of the first sequence according to the plurality of cosine values, including: Dividing the value range of the plurality of cosine values into a plurality of intervals; For any one of the plurality of intervals, counting the number of cosine values within the interval; Taking the ratio of the number to the total number of the plurality of cosine values as the state probability of the interval; Calculate the first eigenvector of the first sequence according to the multiple state probabilities corresponding to the multiple intervals one by one.

8. An injector fault detection device, characterized in that, The device includes: A sequence acquisition module, configured to acquire a first sequence of the high-pressure fuel pipe of any one of a plurality of fuel injectors in a fuel system; the first sequence includes a plurality of first vibration signals arranged in chronological order; A matrix construction module, configured to construct a first matrix according to every adjacent m first vibration signals in the first sequence; the number of rows of the first matrix is N - m + 1, the number of columns of the first matrix is m, and N is the number of the plurality of first vibration signals; A cosine value calculation module, configured to calculate the cosine similarity between every adjacent two rows in the first matrix to obtain a plurality of cosine values; An eigenvector calculation module, configured to calculate the first eigenvector of the first sequence according to the plurality of cosine values; A fault detection module, configured to perform fault detection on the fuel injector according to the first eigenvector.

9. A computer device, characterized in that, including: A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the fuel injector fault detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to cause a computer to execute the fuel injector fault detection method according to any one of claims 1 to 7.

Images