Neural network model selection decision-making method for mechanical fault diagnosis based on sampling inspection and fault loss

Abstract

The invention discloses a neural network model selection decision-making method for mechanical fault diagnosis based on sampling inspection and fault loss. The method comprises the following steps: constructing a plurality of neural network models for mechanical fault diagnosis; and fault loss caused by mechanical fault classification prediction errors of each neural network model is used as a basis for selecting the neural network models, and the neural network model with the minimum fault loss is selected from all the neural network models. According to the method, under the condition that the classification prediction accuracy of the neural network is basically the same, the actual production loss is used as the basis of neural network model selection, and the loss caused by classification prediction errors can be reduced to the maximum extent.

Description

Technical Field

[0001] The invention relates to a neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss, and belongs to the technical field of mechanical fault diagnosis. Background Art

[0002] In recent years, neural network algorithms have developed rapidly. In 1943, McCulloch and Pitts proposed the first model of neurons, the MP model. Subsequently, from the late 1950s to the early 1960s, Rosenblatt proposed a single-layer perception model, which put the concept of neural networks into practice. In 1986, Rumelhart et al. proposed the Back Propagation Network (BP network). In 2006, Hinton et al. introduced the research on deep learning, which set off a boom in artificial neural networks. At the same time, the application of neural networks in fault diagnosis is also developing. Taking bearing fault diagnosis as an example, Tang et al. optimized the support vector machine through the sparrow search algorithm and improved the bearing fault diagnosis accuracy to 99.25%. Zhang et al. proposed a fault diagnosis method based on the Gram angle field and particle swarm optimization convolutional neural network structure. Compared with other mainstream convolutional neural network structures, it can obtain better bearing fault diagnosis accuracy, and the average fault diagnosis accuracy can reach 99%. Zhang et al. proposed a bearing fault diagnosis model based on transfer learning, which has improved the fault diagnosis accuracy. These documents compare with other neural networks or optimization algorithms, and use the prediction accuracy as a basis to judge the performance of their own models. The prediction accuracy of these neural network models has reached more than 99%, and it is difficult to significantly improve the algorithm.

[0003] With the development of neural networks, many models with excellent performance have emerged, such as GA-BP, CNN, PSO-SVM, LSSVM, etc. Using these models to classify and predict mechanical faults is a hot topic of research. Mechanical fault diagnosis often uses prediction accuracy as the judgment standard in selecting neural network models, and pays more attention to the accuracy of classification prediction of neural network models, while ignoring the losses caused by the wrong types of neural network classification prediction to actual mechanical production. This method of selecting neural network models based on classification prediction accuracy in mechanical fault diagnosis cannot be well applied to actual production, and the selected neural network model is not the optimal solution. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art in using prediction accuracy as a judgment criterion for selecting neural network models, the purpose of the present invention is to propose a neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault losses. Under the condition that the neural network classification prediction accuracy is basically the same, the present invention uses actual production losses as the basis for selecting the neural network model, which can maximize the reduction of losses caused by classification prediction errors.

[0005] The technical solution of the present invention is achieved in this way:

[0006] A decision method for selecting a neural network model for mechanical fault diagnosis based on sampling inspection and failure loss is used to construct several neural network models for mechanical fault diagnosis. The failure loss caused by the error in the mechanical fault classification prediction of each neural network model is used as the basis for selecting the neural network model, and the neural network model with the smallest failure loss is selected among all the neural network models.

[0007] Furthermore, when constructing the neural network model, the data set is divided into a training set and a test set. The relationship between input and output features is constructed through the training set to obtain the neural network model. Using the sampling inspection principle, the test results of the test set are used to infer the number of classification prediction errors of the constructed neural network model for the actual monitoring object in mechanical fault diagnosis. Different classification prediction errors and numbers correspond to different losses, and the sum of the losses of all classification prediction errors is the fault loss.

[0008] Furthermore, the classification prediction error is divided into three error situations: the monitored object is in a fault state, and the classification prediction is a healthy state; the monitored object is in a healthy state, and the classification prediction is a fault state; the monitored object is in a fault state, and the classification prediction is a fault state but the fault type is judged incorrectly; the former situation belongs to a safety failure, recorded as a type I misjudgment, and the corresponding loss is called a safety failure loss; the latter two situations belong to efficiency failures, recorded as type II misjudgments, and the corresponding losses are called efficiency failure losses; using the sampling inspection principle, based on the occurrence frequency of type I misjudgments and type II misjudgments in the test set samples, the frequency of type I misjudgments and type II misjudgments of the corresponding neural network model in the actual monitoring process is inferred, and then the safety failure loss and efficiency failure loss of each neural network model are inferred, and the sum of the safety failure loss and the efficiency failure loss is the failure loss.

[0009] Compared with the prior art, the present invention has the following beneficial effects:

[0010] The present invention combines the mechanical production process to obtain a new basis for selecting a neural network model. While considering the accuracy of neural network prediction, it pays more attention to the error types of neural network model classification prediction in mechanical production and the production losses caused. The present invention solves the problem that the neural network model is limited to theory. Based on the failure losses caused in actual production, it truly applies it to mechanical production, and minimizes failure losses in the mechanical production process. The present invention compares the principles of applying neural networks to mechanical fault diagnosis with the principles of sampling inspection, and makes a detailed division of the types of neural network classification prediction errors, and proposes a fault loss function based on this. Experimental results show that the smaller the value of the fault loss function, the smaller the loss caused by the classification prediction errors of the neural network model in the actual mechanical production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 This is the principle diagram of sampling inspection.

[0012] Figure 2 This is the schematic diagram of mechanical fault diagnosis using a neural network model.

[0013] Figure 3 Schematic diagram of the neural network classification prediction error type of the present invention.

[0014] Figure 4 This is a sampling characteristic curve diagram of the present invention.

[0015] Figure 5 This is a schematic diagram of the bearing fault software operation interface I according to Example 1 of the present invention.

[0016] Figure 6 This is a schematic diagram of the bearing fault software operation interface II according to Example 1 of the present invention.

[0017] Figure 7 This is a schematic diagram of the bearing fault software operation interface III of Example 1 of the present invention.

[0018] Figure 8 This is a schematic diagram of the bearing fault software operation interface IV of Example 1 of the present invention.

[0019] Fig. 9 This is a schematic diagram of the bearing fault software operation interface V according to Example 1 of the present invention.

[0020] Fig.10 This is a schematic diagram of the equipment installation of Example 2 of the present invention.

[0021] Fig.11 This is a schematic diagram comparing the prediction accuracy of the BP neural network and the GA-BP neural network in Example 2 of the present invention.

[0022] Fig.12Schematic diagram for comparing confusion matrices of the BP neural network and the GA-BP neural network in Example 2 of the present invention.

[0023] Fig.13 Schematic diagram of the comparison of prediction accuracy of convolutional neural network and GA-BP neural network in Example 2 of the present invention.

[0024] Fig.14 Schematic diagram of comparison of confusion matrices of convolutional neural network and GA-BP neural network in Example 2 of the present invention. DETAILED DESCRIPTION

[0025] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0026] The present invention constructs a plurality of neural network models for mechanical fault diagnosis based on a decision method for selecting a neural network model for mechanical fault diagnosis based on sampling inspection and fault loss. The fault loss caused by errors in the mechanical fault classification prediction of each neural network model is used as the basis for selecting the neural network model, and the neural network model with the smallest fault loss is selected from all the neural network models.

[0027] The present invention is based on the failure losses caused in actual production and truly applies it to mechanical production, thus maximizing the reduction of failure losses in the mechanical production process. The present invention compares the principle of applying neural networks to mechanical fault diagnosis with the principle of sampling inspection, and makes a detailed classification of the types of neural network classification prediction errors, and proposes a failure loss function based on this. The details are as follows.

[0028] About the basic principles of neural network and sampling inspection. The application principle of neural network model in mechanical fault diagnosis is similar to the principle of sampling inspection. Sampling inspection is to randomly select a certain number of samples from a batch of products to be inspected, and then inspect all the samples, and then determine the quality status of the products based on the inspection results of the samples. The inspection principle is as follows: Figure 1 As shown in the figure. In terms of neural networks, a limited data set is first extracted from the monitored object and divided into a training set and a test set. The relationship between input and output features is constructed through the training set. The test set test results are used to infer the number of classification prediction errors of the neural network model for the actual monitored object in mechanical fault diagnosis. The test principle is as follows Figure 2 As shown in Figure 2. Both of them use the idea of ​​"inferring the population from the sample".

[0029] Neural network model classification prediction error types

[0030] In the actual mechanical production process, there are three kinds of errors in the neural network classification prediction: the monitored object is in a fault state, but the classification prediction is a healthy state; the monitored object is in a healthy state, but the classification prediction is a fault state; the monitored object is in a fault state, but the classification prediction is a fault state but the fault type is misjudged. The former situation is a safety failure, recorded as a type I misjudgment, and the latter two situations are efficiency failures, recorded as type II misjudgments. The neural network classification prediction error types of the present invention are as follows: Figure 3 shown.

[0031] The confusion matrix is ​​used as a visualization tool in the field of machine learning. It displays the true value of the sample and the predicted value of the model in a confusion matrix, which can intuitively determine the fault type and conduct diagnostic analysis. Take the following confusion matrix as an example:

[0032]

[0033] Each column of the confusion matrix represents a predicted category, and the total number of each column represents the number of data predicted to be of that category; each row represents the true category of the data, and the total number of data in each row represents the number of data instances of that category; the value in each column represents the number of true data predicted to be of that category.

[0034] When the true state of the sample is fault type 3, 5 samples are predicted by the model as healthy state, which belongs to Class I misjudgment; when the true state of the sample is healthy state, 3 samples are predicted by the model as fault type 1, which belongs to Class II misjudgment; when the true state of the sample is fault type 1, 2 samples are predicted by the model as fault type 2, which belongs to Class II misjudgment.

[0035] The sampling characteristic curve uses the percentage of defective products p as the horizontal axis. The key is to find out the number of defective products in the batch of products, obtain the acceptance probability L(p) as the vertical axis through probability distribution, and use it to judge the quality status of the entire batch of products. Figure 4 This is a sampling characteristic curve diagram of the present invention.

[0036] p=D / N×100% (1)

[0037]

[0038] In formula (1), p is the percentage of defective products; D is the number of defective products in a batch of products; N is the total number of products in a batch. Formula (2) is a Poisson distribution, and the sample size n and the acceptance number (i.e., the judgment standard) Ac are the two most basic parameters in the sampling plan.

[0039] For example, using the sampling plan (30, 3), when L(p) = L(5%) = 0.943, it means that out of every 100 batches of products with this quality, approximately 94 batches will be judged as qualified products.

[0040] However, there are many aspects to judge whether a product is unqualified, such as size, appearance, performance, etc. If the basis for each product's unqualified is used as the horizontal axis, the acceptance probability L(p) is obtained through probability distribution, then 1-L(p) is the probability of defects such as size, appearance, and performance appearing in the entire batch of products. Similarly, in the neural network model, the percentage of type I misjudgment and the percentage of type II misjudgment can be obtained, and the calculation formula is:

[0041] P1=D1 / N×100% (3)

[0042] P2=D2 / N×100% (4)

[0043] P1 is the percentage of type I misjudgment; P2 is the percentage of type II misjudgment; D1 is the number of type I misjudgments in the validation set; D2 is the number of type II misjudgments in the validation set; N is the total number of validation sets.

[0044] By selecting a fixed sampling scheme (n, AC), the acceptance probabilities L(P1) and L(P2) can be obtained. At this time, the acceptance probability L(P1) indicates the probability that no type I misjudgment will occur in the overall operation of the monitored object, and the acceptance probability L(P2) indicates the probability that no type II misjudgment will occur in the overall operation of the monitored object. Therefore, α1 and α2 are respectively the safety risk coefficient and the efficiency risk coefficient, and their respective calculation formulas are:

[0045] α1=1-L(P1) (5)

[0046] α2=1-L(P2) (6)

[0047] In the formula, the safety risk coefficient α1 represents the probability of type I misjudgment in the overall operation of the monitored object, and α2 represents the probability of type II misjudgment in the overall operation of the monitored object.

[0048] As an emerging discipline, mechanical fault diagnosis technology plays an important role in monitoring, diagnosing and predicting the operating status of mechanical equipment. At the same time, understanding the manifestation of faults is the basis of mechanical fault diagnosis. Faults often appear in two forms in mechanical production: sudden failures caused by failure of mechanical parts and short-term failures caused by repairing or maintaining mechanical equipment. The former type of fault focuses on analyzing the losses caused to the mechanical equipment itself and the life safety of operators, which is called safety failure loss; the latter type of fault focuses on analyzing the production window period loss caused by repairing or maintaining mechanical equipment, which is called efficiency failure loss.

[0049] According to the ratio of safety failure loss to efficiency failure loss, the failure loss magnitude t can be obtained. Failure loss magnitude t = 1000, that is, a safety failure causes 1000 unit losses, and an efficiency failure causes 1 unit loss.

[0050] When different levels of attention are paid to safety failures and efficiency failures, the magnitude of failure loss t will change. In actual production processes, it can be divided into three situations:

[0051] ① When the loss caused by safety failure is obviously greater than the loss caused by efficiency failure or when it is hoped to avoid casualties caused by safety accidents, more attention should be paid to the loss caused by safety failure. Therefore, the value of t should be as large as possible (t tends to infinity) to increase the proportion of safety failure loss in failure loss.

[0052] ② When the safety failure loss is significantly smaller than the efficiency failure loss or the safety accidents that may occur in the production process cause less harm to personnel, more attention should be paid to the failure losses caused by efficiency problems. Therefore, the t value should be as small as possible (t tends to 0) to increase the proportion of efficiency loss in failure losses.

[0053] ③ When the failure loss caused by safety issues is approximately the same as the failure loss caused by efficiency issues, the two are given equal attention, the t value is 1, and the proportion of the two losses in the failure loss is approximately the same. The following table is a direct comparison of the three situations.

[0054] The scale between safety and efficiency Failure loss ratio Focus on safety t tends to positive infinity Safety failure losses account for a large proportion Focus on efficiency t tends to 0 Efficiency failure losses account for a large proportion Both are equally important t＝1 The loss ratio is approximately the same

[0055] In the process of mechanical production, once mechanical equipment fails, it often causes huge economic losses and casualties. At this time, the loss due to safety failure is obviously greater than the loss due to efficiency failure, and the magnitude of the failure loss t should be as large as possible.

[0056] When the monitoring object runs m times, mα1 and mα2 represent the number of type I and type II misjudgments of the monitoring object, respectively. At this time, the safety failure loss and efficiency failure loss are expressed as tmα1 and mα2, respectively. Let η1 and η2 be the safety failure loss function and efficiency failure loss function, respectively, η1 = safety failure loss / batch number of products; η2 = efficiency failure loss / batch number of products, so the calculation formula is as follows:

[0057] η1=t*a1 (7)

[0058] η2=a2 (8)

[0059] The safety failure loss function and the efficiency failure loss function represent the loss caused by each sample. t is the magnitude of safety and efficiency (i.e., the magnitude of failure loss); α1 is the safety risk coefficient; α2 is the efficiency risk coefficient.

[0060] Combined with the above analysis, the fault loss function η is proposed, which is the sum of the safety fault loss function and the efficiency fault loss function. Its calculation formula is as follows:

[0061] η=η1+η2(9)

[0062] Therefore, according to formulas (2)(3)(4)(5)(6)(7)(8)(9), from the perspective of safety and efficiency, according to the degree of failure loss of safety and efficiency, we can flexibly select the magnitude between safety and efficiency, and obtain the failure loss function η as the criterion for selecting the neural network model. Generally speaking, the smaller the failure loss function η, the smaller the failure loss caused. However, in the actual mechanical production process, in order to avoid major safety accidents, safety is often ensured by sacrificing efficiency. The calculation principle can be expressed as follows: artificially increase the failure loss magnitude t so that the efficiency loss function η2 can be ignored. At this time, the failure loss function is approximately equal to the safety loss function, that is, η=η1=t*α1. Therefore, the selection of the neural network is determined by the safety risk coefficient α1.

[0063] The present invention is described in detail below through two embodiments.

[0064] Embodiment 1:

[0065] A software for bearing failure was designed using Matlab R2024a and applied to a machine tool production line of a certain enterprise. According to the investigation, the safety accident loss caused by bearing failure in this production line is equivalent to RMB 2 million, and the production window loss caused by regular replacement of faulty bearings is equivalent to RMB 10,000. The failure loss magnitude is t=200.

[0066] (1) Input safety failure loss and efficiency failure loss, such as Figure 5 shown.

[0067] (2) Click 'Click' to jump to the interface for selecting the neural network model, such as Figure 6 The neural network models selected in this embodiment include BP neural network, GA-BP neural network and CNN neural network.

[0068] (3) Import the bearing data set, such as Figure 7 shown.

[0069] (4) Select a neural network model, such as Figure 8 shown.

[0070] (5) Calculation results, such as Fig. 9 shown.

[0071] The analysis found that the failure loss value of the BP neural network model was 44,930 yuan, which was significantly lower than the failure loss value of 32,390 yuan of the GA-BP neural network and the failure loss value of 79,880 yuan of the CNN neural network. Therefore, the BP neural network model is more suitable as the neural network model for fault diagnosis of the machine tool production line of this enterprise.

[0072] Embodiment 2:

[0073] Example 2 uses the Canada-Ottawa University bearing dataset for model training and testing to verify the effectiveness of the method and ideas. The dataset contains vibration signals collected from bearings in five health conditions under time-varying speeds. The five health conditions are healthy (H), inner ring defect fault (IF), outer ring defect fault (OF), ball defect fault (BF), and inner ring, outer ring and ball combined defect fault (CF). Each dataset contains two channels: "Channel_1" is the vibration data measured by the accelerometer, and "Channel_2" is the speed data measured by the encoder. All of these data are sampled at a frequency of 200,000 Hz, and the sampling time lasts for 10s. The CPR (cycles per revolution) of the encoder is 1024. Example 2 The equipment is installed as follows: Fig.10 shown.

[0074] This experiment uses vibration data and speed data as input features, and health status as output features. In order to analyze the health status of the bearing at the start-up stage, the signal values ​​of the vibration data of 5 health statuses under the condition of increasing speed are selected. In the initial stage of the test, the speed is small and the vibration difference is not obvious, so the vibration data when the speed increases to the maximum value and gradually stabilizes is selected. It is also found through observation of the data that the outer ring defect fault (OF) and the ball defect fault (BF) data are similar and not suitable for analysis together, so the following four categories are mainly studied: health (H), inner ring defect fault (IF), outer ring defect fault (OF) and inner ring, outer ring and ball combined defect fault (CF). Each health status is in units of 10 cycles, and a total of 284 groups of data (including vibration data and speed data) are selected, and each health status is numbered 1, 2, 3, and 4 in sequence. 199 groups are selected as training sets to construct a GA-BP neural network, and the remaining 85 groups are used as test sets for performance verification, and the fault classification test graph is output. Then, the confusion matrix is ​​used to display the statistical information of the classification results of the model for the fault category samples.

[0075] In MATLAB, the collected data is first divided into training set and test set, and the mapminmax function is used to normalize the data and map the data to the [0,1] interval. The newff function is used to establish a BP neural network, and the number of hidden layers is set to 10, forming a 2-10-1 network structure. The training parameters are then determined: the maximum number of iterations is 1000, the target training error is 1×10^(-6), and the learning rate is 0.01. Finally, simulation tests and data processing are performed to obtain the test set prediction results. After optimizing the weights and thresholds of the BP neural network through a genetic algorithm, the above steps are repeated to obtain a comparison of the test set prediction results, as shown in the figure below. Fig.11 and Fig.12 shown.

[0076] By comparing the BP neural network and the GA-BP neural network, it is found that the prediction accuracy of the two is the same, both 92.9412%, and they have strong similarities in terms of confusion matrix. The only difference is the number of error types they make. In the case of 85 total samples, the BP neural network makes 1 type I misjudgment and 5 type II misjudgments, while the GA-BP neural network makes 0 type I misjudgments and 6 type II misjudgments. The type I misjudgment percentage P1 and type II misjudgment percentage P2 of each of the two can be obtained by formula (3) and (4), and then the acceptance probability L(p) can be obtained by formula (2). The fault loss function η is obtained by formulas (5), (6), (7), (8), and (9) to judge the performance of the model.

[0077] When the sampling scheme is (85, 30, 3) and t=100, the data obtained by the two models are shown in the following table:

[0078]

[0079] When the sampling scheme is (85, 30, 3) and t=1000, the data obtained by the two models are shown in the following table:

[0080]

[0081] From the table analysis, when the fault loss magnitude t = 100, that is, the safety fault loss causes 100 unit losses and the efficiency fault loss causes 1 unit loss, the BP neural network is selected because the fault loss function η = 0.153 of the BP neural network is less than the fault loss function η = 0.1648 of the GA-BP neural network. Similarly, when the fault loss magnitude t = 1000, that is, the safety fault loss causes 1000 unit losses and the efficiency fault loss causes 1 unit loss, the GA-BP neural network is selected because the fault loss function η = 0.603 of the BP neural network is greater than the fault loss function η = 0.1648 of the GA-BP neural network. It can be seen that as the fault loss magnitude t increases, the selection of neural networks will change, and at the same time, the advantages of neural networks with a small percentage of Class I misjudgment will become more and more obvious.

[0082] According to the above experimental method, the experimental results of CNN neural network are obtained. By comparing CNN neural network and GA-BP neural network, it is found that Fig.13 and Fig.14As shown in the figure, the prediction accuracy of the CNN neural network is significantly higher than that of the GA-BP neural network. However, through the confusion matrix, it can be intuitively found that when the total sample size is 85, the CNN neural network makes 2 type I errors and 3 type II errors, while the GA-BP neural network makes 0 type I errors and 6 type II errors. The type I misjudgment percentage P1 and type II misjudgment percentage P2 of the two can be obtained by formula (3) and (4), and the acceptance probability L(p) can be obtained by formula (2). The fault loss function η is obtained by formulas (5), (6), (7), (8), and (9) to judge the model performance.

[0083] When the sampling scheme is (85, 30, 3) and t=100, the data obtained by the two models are shown in the following table:

[0084]

[0085] When the sampling scheme is (85, 30, 3) and t=1000, the data obtained by the two models are shown in the following table:

[0086]

[0087] From the table, no matter the fault loss magnitude t = 100 or 1000, the fault loss function η of the GA-BP neural network is always smaller than the fault loss function η of the CNN, so the GA-BP neural network is selected. This also directly shows that even if the prediction accuracy of the CNN is higher than that of the GA-BP, the selection of the neural network model will change after introducing the possible fault losses in actual production.

[0088] From the above experiments, it can be seen that although the prediction accuracy of the CNN neural network is higher than that of the BP neural network and the GA-BP neural network, by analyzing the prediction error types and introducing the concept of fault loss, the selection of the neural network will change according to the change of the fault loss magnitude t and will be determined by the fault loss function.

[0089] The present invention starts from the perspective of safety and efficiency, and obtains the fault loss magnitude t according to the losses caused by safety faults and efficiency faults in the mechanical production process; divides the error prediction results of the neural network model into safety faults and efficiency faults through the confusion matrix, which are recorded as Class I misjudgments and Class II misjudgments respectively; uses the sampling inspection principle, based on the occurrence frequency of Class I misjudgments and Class II misjudgments in the test set samples, infers the frequency of Class I misjudgments and Class II misjudgments of each neural network model in the actual monitoring object process, and multiplies it with the fault loss magnitude t to obtain the fault loss function η; through the fault loss function, judge the best neural network suitable for fault diagnosis. In general, the selection of neural network models in fault diagnosis will change according to the change of the fault loss magnitude t.

[0090] Under the condition that the classification prediction accuracy of the neural network model reaches more than 90%, the present invention distinguishes the classification prediction error types. Taking into account the need to avoid losses caused by safety accidents in the mechanical production process as much as possible, the basis for selecting the neural network model is no longer based on the classification prediction accuracy, but the losses caused by failures in the mechanical production process are used as the basis for model selection. This is more in line with actual selection considerations and can also reduce losses caused by failures.

[0091] Finally, it should be noted that the above examples of the present invention are merely examples for illustrating the present invention, and are not intended to limit the embodiments of the present invention. Although the applicant has described the present invention in detail with reference to the preferred embodiments, for those of ordinary skill in the art, other different forms of changes and modifications can be made based on the above description. It is impossible to list all the embodiments here. Any obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A decision-making method for selecting a neural network model for mechanical fault diagnosis based on sampling inspection and fault loss, characterized in that: Construct several neural network models for mechanical fault diagnosis; use the failure loss caused by the error in the mechanical fault classification prediction of each neural network model as the basis for selecting the neural network model, and select the neural network model with the smallest failure loss among all the neural network models.

2. The neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss according to claim 1 is characterized in that: When constructing the neural network model, the data set is divided into a training set and a test set. The relationship between input and output features is constructed through the training set to obtain the neural network model. Using the sampling inspection principle, the test results of the test set are used to infer the number of classification prediction errors of the constructed neural network model for the actual monitoring object in mechanical fault diagnosis. Different classification prediction errors and numbers correspond to different losses, and the sum of the losses of all classification prediction errors is the fault loss.

3. The neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss according to claim 2 is characterized by: The classification prediction error is divided into three error situations: the monitoring object is in a fault state, and the classification prediction is a healthy state; the monitoring object is in a healthy state, and the classification prediction is a fault state; the monitoring object is in a fault state, and the classification prediction is a fault state but the fault type is misjudged; the former situation belongs to a safety fault, which is recorded as a type I misjudgment, and the corresponding loss is called a safety fault loss; the latter two situations belong to an efficiency fault, which is recorded as a type II misjudgment, and the corresponding loss is called an efficiency fault loss; Utilizing the sampling inspection principle, based on the occurrence frequency of type I and type II misjudgments in the test set samples, the frequency of type I and type II misjudgments of the corresponding neural network model in the actual monitoring process is inferred, and then the safety failure loss and efficiency failure loss of each neural network model are inferred. The sum of the safety failure loss and the efficiency failure loss is the failure loss.

4. The neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss according to claim 3 is characterized by: The percentage of type I misjudgment and the percentage of type II misjudgment are calculated as follows: P1=D1 / N×100% (3) P2=D2 / N×100% (4) Among them, P1 is the percentage of type I misjudgment; P2 is the percentage of type II misjudgment; D1 is the number of type I misjudgments in the validation set; D2 is the number of type II misjudgments in the validation set; N is the total number of validation sets; Assume that the acceptance probability L(P1) represents the probability that no type I misjudgment will occur in the overall operation of the monitored object, and the acceptance probability L(P2) represents the probability that no type II misjudgment will occur in the overall operation of the monitored object; α1 and α2 are the safety risk coefficient and efficiency risk coefficient, respectively, and the calculation formulas are: α1=1-L(P1) (5) α2=1-L(P2) (6) The safety risk coefficient α1 represents the probability of type I misjudgment in the overall operation of the monitored object, and α2 represents the probability of type II misjudgment in the overall operation of the monitored object; The fault loss magnitude t is defined as the ratio of the safety fault loss to the efficiency fault loss. Let mα1 and mα2 represent the number of type I and type II misjudgments of the monitored object when the monitored object is run m times, respectively. At this time, the safety fault loss and efficiency fault loss are expressed as tmα1 and mα2, respectively. Let η1 and η2 be the safety fault loss function and the efficiency fault loss function, respectively, representing the safety fault loss and efficiency fault loss caused by each sample. It is defined as η1 = safety fault loss / number of batch products; η2 = efficiency fault loss / number of batch products. The calculation formula is simplified as follows: η1=t*a1 (7) η2=a2 (8) Let the failure loss function η be the sum of the safety failure loss function η1 and the efficiency failure loss function η2, that is: η=η1+η2(9) Calculate the fault loss function η of all neural network models, and select the neural network model with the smallest fault loss function η among all neural network models.

5. The neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss according to claim 2 is characterized by: In the mechanical production process, the fault loss magnitude t is increased so that the efficiency fault loss function η2 is ignored, that is, η=η1=t*α1. At this time, the selection of the neural network model is determined by the safety risk coefficient α1. Among all the neural network models, the neural network model with the smallest safety risk coefficient α1 can be selected.

6. The neural network model selection decision method for mechanical fault diagnosis based on sampling inspection and fault loss according to claim 1 is characterized in that: The classification prediction accuracy of all neural network models reached over 90%.

Images