A method for constructing a hybrid model for internal combustion engine fault diagnosis with few samples

Abstract

The application aims to provide a construction method of an internal combustion engine fault diagnosis hybrid model for few samples, belonging to the field of fault diagnosis, comprising a model discriminator, a sample discriminator and a parameter optimizer: the model discriminator constructs an internal combustion engine fault diagnosis model, and analyzes the accuracy of the internal combustion engine fault diagnosis model; the sample discriminator analyzes the classification result of the internal combustion engine fault diagnosis model; the parameter optimizer uses an optimization algorithm to find the extreme value of the hyperparameters and sample weight values of the internal combustion engine fault diagnosis model, and performs twice parameter optimization in the model construction process, parameter optimization for the fault diagnosis model, and weight optimization for the misclassified samples. The application can improve the processing capacity of the model for fault samples under the condition of few samples, and effectively improves the classification accuracy. The model construction framework of the application involves the application of various algorithms, can be adjusted according to the actual situation, and has high universality.

Description

Technical Field

[0001] The present invention relates to a fault diagnosis method, specifically a fault diagnosis method for internal combustion engines. Background Technology

[0002] Internal combustion engines operate for extended periods in harsh environments with corrosive substances and strong impacts. This makes them highly susceptible to failure due to factors such as degradation, aging, and wear, potentially leading to major accidents and seriously threatening people's lives and property. Fault diagnosis plays a crucial role in the prediction and health management of internal combustion engines. Fault diagnosis is the process of understanding and monitoring the machine's operating status, determining its overall or partial normal and abnormal conditions. It is a technology capable of detecting faults and their causes at an early stage and predicting the development trend of faults.

[0003] As mechanical diagnostic and prediction technologies incorporating next-generation artificial intelligence mature, they become capable of constructing more complex and sophisticated mechanical damage models, thereby improving fault diagnosis and prediction performance. Fault diagnosis models built using machine learning methods require a large amount of data samples. However, the inability of internal combustion engines to sustain faults for extended periods limits the collection of fault data. This scarcity of data hinders the development of accurate fault diagnosis models, particularly in the context of small-sample, highly imbalanced fault modeling, posing a significant and challenging problem.

[0004] Chinese invention patent CN116933656A discloses a deep learning-based method for fault diagnosis of marine propulsion systems. This invention divides the propulsion system into power system, power system modules, power system equipment, and power system parameters based on the type of the main engine. For fault diagnosis of low / zero carbon marine propulsion systems, it proposes a fault diagnosis algorithm based on data mining and deep learning applicable to multiple propulsion systems. However, this invention has limitations in its application to fault diagnosis of marine propulsion systems. Furthermore, while deep learning has advantages in data processing, it cannot solve the problem of low accuracy and low diagnostic precision in fault diagnosis models with few samples. Summary of the Invention

[0005] The purpose of this invention is to provide a method for constructing a hybrid model for internal combustion engine fault diagnosis with a small number of samples, which can improve the ability to handle fault samples in the case of small samples.

[0006] The objective of this invention is achieved as follows:

[0007] This invention discloses a method for constructing a hybrid model for internal combustion engine fault diagnosis with a small sample size. The method comprises a model discriminator, a sample discriminator, and a parameter optimizer. The model discriminator constructs an internal combustion engine fault diagnosis model, analyzes its accuracy, and selects high-accuracy fault diagnosis sub-models as the basis for constructing the hybrid model. The sample discriminator analyzes the classification results of the internal combustion engine fault diagnosis model, weights the classified samples based on the accuracy of the fault samples, marks incorrect samples, and updates the weights of incorrect samples. The parameter optimizer uses an optimization algorithm to find the optimal values ​​of the hyperparameters and sample weights of the internal combustion engine fault diagnosis model, performing two parameter optimizations during the model construction process: parameter optimization for the fault diagnosis model and weight optimization for incorrectly classified samples.

[0008] Step 1: Process the raw data acquired by the internal combustion engine sensors to extract the fault sample feature parameters F, or simulate the internal combustion engine fault state through a simulation model to obtain the thermodynamic parameters of different internal combustion engine fault types C as the fault sample feature parameters F; divide the internal combustion engine fault samples into a training set Trr and a test set Te.

[0009] Step 2: Import the training set divided in Step 1 into the machine learning algorithm to construct the internal combustion engine fault diagnosis model M, where the input variables are the extracted feature parameters and the outputs are the different fault types of the internal combustion engine. The test set is then imported into the internal combustion engine fault diagnosis model to calculate the accuracy of the classification results.

[0010] Step 3: Import the internal combustion engine fault diagnosis model M constructed in Step 2 into the model discriminator MD, use the classification accuracy as the screening criterion, set a threshold, and retain internal combustion engine fault diagnosis models that exceed the threshold as sub-models of the internal combustion engine fault diagnosis hybrid model. The number of sub-models is greater than or equal to 2.

[0011] Step 4: Use the optimization algorithm in the parameter optimizer OP to optimize the hyperparameters of the sub-models selected in Step 3, and use the accuracy of the classification results as the optimization index to find the hyperparameters of the internal combustion engine fault diagnosis model with the highest classification accuracy.

[0012] Step 5: Analyze the classification results of the optimized sub-model in Step 4, import the classification results into the sample discriminator SD, perform weighted combination of all fault samples in the classification results, initialize the weight values, and mark the fault samples that are misclassified.

[0013] Step 6: Import the erroneous samples marked in Step 5 into the parameter optimizer OP, and use the optimization algorithm to find the best solution for the erroneous samples;

[0014] Step 7: Combine the weights from Step 6 with the optimized sub-model to establish a hybrid model for internal combustion engine fault diagnosis.

[0015] The present invention may also include:

[0016] 1. The processing of the raw data acquired by the internal combustion engine sensor in step 1 includes denoising the data using principal component analysis and mode decomposition methods, extracting the fault sample feature parameters F, including converting the time domain signal into the frequency domain signal using Fourier transform and wavelet Fourier transform, extracting time domain and frequency domain feature parameters, and extracting fault sample feature parameters using convolutional neural networks.

[0017] 2. In step 1, the internal combustion engine fault samples are divided into training and test sets. An equal number of samples for each fault type are randomly selected and combined to construct the training set Tr. The remaining samples are used as the test set Te. The training set should be a set {Tr1, Tr2, Tr3, ..., Tr...} containing different fault samples. i}, Test set {Te1,Te2,Te3,…,Te i For example, if an internal combustion engine fault type C contains 5 fault types C = {c1, c2, c3, c4, c5}, and each fault type contains 5 feature parameters F = {f1, f2, f3, f4, f5}, and 100 samples are collected for each fault type, for a total of 500 fault samples, then the training set Tr1 is a fault sample randomly selected from the fault samples of the 5 fault types C, containing 5 fault samples. The remaining 490 fault samples are used as the test set Te1, and so on.

[0018] 3. The model discriminator utilizes machine learning algorithms to construct various internal combustion engine fault diagnosis models, including neural networks, support vector machines, decision trees, and random forests.

[0019] 4. In step 3, the internal combustion engine fault diagnosis model M is imported into the model discriminator MD, and sub-models are selected based on the accuracy of the results, as shown below:

[0020]

[0021] In the formula, M is the internal combustion engine fault diagnosis model, R is the model classification accuracy, and value is the set accuracy threshold.

[0022] 5. In step 5, the classification results are imported into the sample discriminator SD. Faulty samples are weighted and combined, and the sample weights are initialized. The weights of correctly classified samples are retained, while the weights of incorrectly classified samples are updated by the parameter optimizer.

[0023]

[0024] Where w is the sample weight, i is the sample, N is the number of sub-models selected in step 4, and w' is the weight after the optimization algorithm is updated.

[0025] Error samples are marked, and their weights are imported into the parameter optimizer for weight updates. The weights w of the correct samples in each sub-model are then recombined with the updated error sample weights w', so that the sum of the weights w of the correct samples in each sub-model is 1, and the sum of the updated error sample weights w' is also 1.

[0026]

[0027] In the formula, Sample is the weighted combination of samples, R is the number of correctly classified samples, E is the number of misclassified samples, Sr is the number of correctly classified samples, Se is the number of misclassified samples, and Ms is the high-accuracy sub-model selected by the model discriminator.

[0028] 6. The sample discriminator process in step 6 is as follows:

[0029] (1) The samples of the N high-accuracy sub-models after being screened by the model discriminator are weighted with a weight of w;

[0030] (2) Initialize the weights w to 1 / N, the total weights to 1, and mark the erroneous samples;

[0031] (3) Use the parameter optimizer to update the weight values ​​of the erroneous samples and find the optimal weight combination w'(N);

[0032] (4) Based on the optimal weight w'(N), perform weighted combination of different sub-models;

[0033] (5) Obtain a high-accuracy hybrid model for internal combustion engine fault diagnosis.

[0034] 7. The parameter optimizer uses optimization algorithms to find the optimal values ​​of hyperparameters and sample weights of the internal combustion engine fault diagnosis model. The optimization algorithms include particle swarm optimization algorithm, genetic algorithm, gray wolf optimization algorithm, and snake optimization algorithm. The parameter optimizer performs two optimizations and adopts different optimization algorithms according to different parameters.

[0035] The advantages of this invention are: it can improve the model's ability to process fault samples in the case of small sample sizes, and its classification accuracy is effectively improved; the model construction framework of this invention involves the application of multiple algorithms and can be adjusted according to the actual situation, and its versatility is high. Attached Figure Description

[0036] Figure 1 This is a flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the sample discriminator. Detailed Implementation

[0038] The invention will now be described in more detail with reference to the accompanying drawings:

[0039] Combination Figure 1-2 The present invention includes a model discriminator, a sample discriminator, and a parameter optimizer:

[0040] The model discriminator is used to construct internal combustion engine fault diagnosis models using various machine learning algorithms, analyze the accuracy of various internal combustion engine fault diagnosis models, and select high-accuracy fault diagnosis sub-models as the basis for building hybrid models.

[0041] The sample discriminator is used to analyze the classification results of the internal combustion engine fault diagnosis model. Based on the accuracy of the fault samples of the internal combustion engine fault diagnosis model, the samples of the classification results are weighted and combined, the erroneous samples are marked, and the weights of the erroneous samples are updated to improve the internal combustion engine fault diagnosis model's ability to process samples.

[0042] The parameter optimizer is used to find the optimal values ​​of hyperparameters and sample weights of the internal combustion engine fault diagnosis model using optimization algorithms. It performs two parameter optimizations during the model building process: parameter optimization for the fault diagnosis model and weight optimization for misclassified samples.

[0043] This invention includes the following steps:

[0044] Step 1: Process the raw data acquired by the internal combustion engine sensors to extract the fault sample feature parameters F, or simulate the internal combustion engine fault state through a simulation model to obtain the thermodynamic parameters of different internal combustion engine fault types C as the fault sample feature parameters F; divide the internal combustion engine fault samples into training set Tr and test set Te.

[0045] Step 2: Import the training set divided in Step 1 into various machine learning algorithms to construct various internal combustion engine fault diagnosis models M, where the input variables are the extracted feature parameters and the outputs are different fault types of internal combustion engines. The test set is then imported into the various internal combustion engine fault diagnosis models to calculate the accuracy of the classification results.

[0046] Step 3: Import the various internal combustion engine fault diagnosis models M constructed in Step 2 into the model discriminator MD, use the classification accuracy as the screening criterion, set a threshold, and retain the internal combustion engine fault diagnosis models that exceed the threshold as sub-models of the internal combustion engine fault diagnosis hybrid model. The number of sub-models should be greater than or equal to 2.

[0047] Step 4: Optimize the hyperparameters of the sub-models selected in Step 3 using the optimization algorithm in the parameter optimizer OP. Use the accuracy of the classification results as the optimization index to find the hyperparameters of the internal combustion engine fault diagnosis model with the highest classification accuracy, and further improve the accuracy of the internal combustion engine fault diagnosis model.

[0048] Step 5: Analyze the classification results of the optimized sub-model in Step 4, import the classification results into the sample discriminator SD, perform weighted combination of all fault samples in the classification results, initialize the weight values, and mark the fault samples that are misclassified.

[0049] Step 6: Import the erroneous samples marked in Step 5 into the parameter optimizer OP, and use the optimization algorithm to optimize the erroneous samples to further improve the accuracy of the optimized sub-model.

[0050] Step 7: Combine the weights from Step 6 with the optimized sub-model to establish a hybrid model for internal combustion engine fault diagnosis. Export the model and save it.

[0051] Step 1 involves processing the raw data acquired by the internal combustion engine sensors, including denoising the data using principal component analysis and mode decomposition methods, extracting the fault sample feature parameters F, including converting the time domain signal into the frequency domain signal using Fourier transform and wavelet Fourier transform, extracting the time domain and frequency domain feature parameters using formulas, and extracting the fault sample feature parameters using convolutional neural networks.

[0052] In step 1, the internal combustion engine fault samples are divided into training and test sets. To ensure an even distribution between the training and test sets, equal numbers of samples for each fault type of the internal combustion engine are randomly selected and combined to construct the training set Tr. The remaining samples are used as the test set Te. To study the impact of different sample sizes on the internal combustion engine fault diagnosis model, the training set should be designed as a set containing different fault samples {Tr1, Tr2, Tr3, ..., Tr...}. i}, test set {Te1,Te2,Te3,…,Te i For example, if an internal combustion engine fault type C contains 5 fault types C = {c1, c2, c3, c4, c5}, and each fault type contains 5 feature parameters F = {f1, f2, f3, f4, f5}, and 100 samples are collected for each fault type, for a total of 500 fault samples, then the training set Tr1 is a fault sample randomly selected from the fault samples of the 5 fault types C, containing 5 fault samples. The remaining 490 fault samples are used as the test set Te1, and so on.

[0053] The model discriminator utilizes various machine learning algorithms to construct multiple internal combustion engine fault diagnosis models. These machine learning algorithms include the application of various classification algorithms, such as neural networks, support vector machines, decision trees, and random forests.

[0054] In step 3, multiple internal combustion engine fault diagnosis models M are imported into the model discriminator MD, and sub-models are selected based on the accuracy of the results, as shown below:

[0055]

[0056] In the formula, M is a fault diagnosis model constructed based on different machine learning methods, R is the model classification accuracy, and value is a set accuracy threshold, which is set to 95%.

[0057] In step 5, the classification results are imported into the sample discriminator (SD), and the faulty samples are weighted and combined to initialize the sample weights. The weights of correctly classified samples are retained, while the weights of incorrectly classified samples are updated by the parameter optimizer.

[0058]

[0059] Where w is the sample weight, i is the sample, N is the number of sub-models selected in step 4, and w' is the weight after the optimization algorithm is updated.

[0060] Error samples are marked, and their weights are imported into the parameter optimizer for weight updates. The weights w of the correct samples in each sub-model are then recombined with the updated error sample weights w', so that the sum of the weights w of the correct samples in each sub-model is 1, and the sum of the updated error sample weights w' is also 1.

[0061]

[0062] In the formula, Sample is the weighted combination of samples, R is the number of correctly classified samples, E is the number of misclassified samples, Sr is the number of correctly classified samples, Se is the number of misclassified samples, and Ms is the high-accuracy sub-model selected by the model discriminator.

[0063] The sample discriminator process in step 6 is as follows:

[0064] Step 1: Weight the samples of the N high-accuracy sub-models selected by the model discriminator, with a weight value of w;

[0065] Step 2: Initialize the weights w to 1 / N, the sum of the weights to 1, and mark the erroneous samples;

[0066] Step 3: Use the parameter optimizer to update the weight values ​​of the erroneous samples, find the optimal weight combination w'(N), and improve the accuracy of the erroneous samples;

[0067] Step 4: Based on the optimal weights w'(N), perform a weighted combination of different sub-models;

[0068] Step 5: Obtain a high-accuracy hybrid model for internal combustion engine fault diagnosis to further improve the model's accuracy.

[0069] The parameter optimizer uses optimization algorithms to find the optimal values ​​of hyperparameters and sample weights of the internal combustion engine fault diagnosis model. There are many types of optimization algorithms, including traditional ones such as particle swarm optimization and genetic algorithms, and newer ones such as gray wolf optimization and snake optimization. The parameter optimizer performs two optimizations. Since the effects of optimization algorithms differ, different optimization algorithms need to be adopted according to different parameters.

Claims

1. A method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples, characterized by: Including models Discriminator, sample discriminator, and parameter optimizer: The model discriminator constructs an internal combustion engine fault diagnosis model, analyzes the accuracy of the internal combustion engine fault diagnosis model, and selects high-accuracy fault diagnosis sub-models as the construction basis for the hybrid model; the sample discriminator analyzes the classification results of the internal combustion engine fault diagnosis model, performs weighted combination of the classification results samples based on the accuracy of the fault samples of the internal combustion engine fault diagnosis model, marks the incorrect samples, and updates the weights of the incorrect samples; The parameter optimizer uses optimization algorithms to find the optimal values ​​of hyperparameters and sample weights of the internal combustion engine fault diagnosis model. It performs two parameter optimizations during the model construction process: parameter optimization for the fault diagnosis model and weight optimization for misclassified samples. Step 1: Process the raw data acquired by the internal combustion engine sensors to extract the fault sample feature parameter F, or simulate the internal combustion engine fault state through a simulation model to obtain the thermodynamic parameters of different internal combustion engine fault types C as the fault sample feature parameter F; The internal combustion engine fault samples are divided into a training set Trr and a test set Te; Step 2: Import the training set divided in Step 1 into the machine learning algorithm to construct the internal combustion engine fault diagnosis model M, where the input variables are the extracted feature parameters and the outputs are the different fault types of the internal combustion engine. The test set is then imported into the internal combustion engine fault diagnosis model to calculate the accuracy of the classification results. Step 3: Import the internal combustion engine fault diagnosis model M constructed in Step 2 into the model discriminator MD, use the classification accuracy as the screening criterion, set a threshold, and retain internal combustion engine fault diagnosis models that exceed the threshold as sub-models of the internal combustion engine fault diagnosis hybrid model. The number of sub-models is greater than or equal to 2. Step 4: Use the optimization algorithm in the parameter optimizer OP to optimize the hyperparameters of the sub-models selected in Step 3, and use the accuracy of the classification results as the optimization index to find the hyperparameters of the internal combustion engine fault diagnosis model with the highest classification accuracy. Step 5: Analyze the classification results of the optimized sub-model in Step 4, import the classification results into the sample discriminator SD, perform weighted combination of all fault samples in the classification results, initialize the weight values, and mark the fault samples that are misclassified. Step 6: Import the erroneous samples marked in Step 5 into the parameter optimizer OP, and use the optimization algorithm to find the best solution for the erroneous samples; Step 7: Combine the weights from Step 6 with the optimized sub-model to establish a hybrid model for internal combustion engine fault diagnosis.

2. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples as described in claim 1, characterized in that: Step 1 involves processing the raw data acquired by the internal combustion engine sensors, including denoising the data using principal component analysis and mode decomposition methods, extracting the fault sample feature parameters F, including converting the time domain signal into the frequency domain signal using Fourier transform and wavelet Fourier transform, extracting time domain and frequency domain feature parameters, and extracting fault sample feature parameters using convolutional neural networks.

3. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples as described in claim 1, characterized in that: In step 1, the internal combustion engine fault samples are divided into training and test sets. An equal number of samples for each fault type are randomly selected and combined to construct the training set Tr. The remaining samples are used as the test set Te. The training set should be a set {Tr1, Tr2, Tr3, ..., Tr...} containing different fault samples. i }, Test set {Te1,Te2,Te3,…,Te i For example, if an internal combustion engine fault type C contains 5 fault types C = {c1, c2, c3, c4, c5}, and each fault type contains 5 feature parameters F = {f1, f2, f3, f4, f5}, and 100 samples are collected for each fault type, for a total of 500 fault samples, then the training set Tr1 is a fault sample randomly selected from the fault samples of the 5 fault types C, containing 5 fault samples. The remaining 490 fault samples are used as the test set Te1, and so on.

4. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples as described in claim 1, characterized in that: The model discriminator utilizes machine learning algorithms to construct various internal combustion engine fault diagnosis models, including neural networks, support vector machines, decision trees, and random forests.

5. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples according to claim 1, characterized in that: In step 3, the internal combustion engine fault diagnosis model M is imported into the model discriminator MD, and sub-models are selected based on the accuracy of the results, as shown below: In the formula, M is the internal combustion engine fault diagnosis model, R is the model classification accuracy, and value is the set accuracy threshold.

6. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples according to claim 1, characterized in that: In step 5, the classification results are imported into the sample discriminator SD, the faulty samples are weighted and combined, the sample weights are initialized, the weights of correctly classified samples are retained, and the weights of misclassified samples are updated by the parameter optimizer: Where w is the sample weight, i is the sample, and N is the number of sub-models selected in step 4. ' To optimize the weights after the algorithm update; Error samples are marked, and their weights are imported into the parameter optimizer for weight updates. The weights w of the correct samples in each sub-model are then compared with the updated weights w of the error samples. ' The model is reorganized so that the sum of the weights w of the correct samples in each sub-model is 1, and the updated weights w of the incorrect samples are... ' The sum of is 1: In the formula, Sample is the weighted combination of samples, R is the number of correctly classified samples, E is the number of misclassified samples, Sr is the number of correctly classified samples, Se is the number of misclassified samples, and Ms is the high-accuracy sub-model selected by the model discriminator.

7. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples according to claim 1, characterized in that: The sample discriminator process in step 6 is as follows: (1) The samples of the N high-accuracy sub-models after being screened by the model discriminator are weighted with a weight of w; (2) Initialize the weights w to 1 / N, the total weights to 1, and mark the erroneous samples; (3) Use the parameter optimizer to update the weight values ​​of the erroneous samples and find the optimal weight combination w. ' (N); (4) Based on the optimal weight w ' (N), weighted combination of different sub-models; (5) Obtain a high-accuracy hybrid model for internal combustion engine fault diagnosis.

8. The method for constructing a hybrid model for fault diagnosis of internal combustion engines with few samples according to claim 1, characterized in that: The parameter optimizer uses optimization algorithms to find the optimal values ​​of hyperparameters and sample weights of the internal combustion engine fault diagnosis model. The optimization algorithms include particle swarm optimization, genetic algorithm, gray wolf optimization, and snake optimization. The parameter optimizer performs two optimizations, using different optimization algorithms for different parameters.

Images