A Fault Diagnosis Method for Industrial Equipment Mechanism Model Based on Confusion Matrix

Through the method based on the confusion matrix, the mechanism model of industrial equipment is quickly corrected and adjusted, and the problems of low accuracy of fault diagnosis and long correction time in the prior art are solved, thereby achieving efficient fault diagnosis and alarm.

CN118734612BActive Publication Date: 2024-11-08广东知业科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411224395.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-11-08
Estimated Expiration
2044-09-03

AI Technical Summary

Technical Problem

In the prior art, industrial equipment fault diagnosis relies on regular inspections and manual experience, with low accuracy, high latency and professional skills. After the mechanism model is developed, it requires multiple on-machine verification, resulting in a long correction time, making it difficult to quickly apply to faults. diagnosis.

Method used

Using a confusion matrix-based method, we can adjust model parameters through data collection, preliminary mechanism model construction, confusion matrix construction and evaluation index analysis, until the preset threshold is reached, and quickly correction and accurate fault diagnosis are achieved.

Benefits of technology

It reduces the time for mechanism model correction, improves the accuracy and efficiency of fault diagnosis, and can be quickly applied to real-time monitoring and fault alarm of industrial equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118734612B_ABST
    Figure CN118734612B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for fault diagnosis of an industrial equipment mechanism model based on a confusion matrix, which obtains the power input parameters and equipment operation parameters of industrial equipment during large-scale normal operation in historical periods; uses the power input parameters as independent variables and the equipment operation parameters as dependent variables to construct a preliminary mechanism model; obtains a number of industrial equipment data test sets when a fault occurs and when a fault does not occur on a computer, inputs the power input parameters into the preliminary mechanism model, and outputs the predicted values ​​of the equipment operation parameters; constructs a confusion matrix through the actual values ​​and predicted values ​​of the equipment operation parameters; uses the evaluation index of the confusion matrix to verify the prediction performance of the mechanism model; analyzes the error tendency of the mechanism model prediction according to the size of the evaluation index, and adjusts the parameters of the preliminary mechanism model. The present invention solves the problem that after the development of the industrial equipment diagnosis mechanism model, it needs to be verified on the computer for many times to determine the accuracy of the model, resulting in a long correction time for the mechanism model.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of equipment fault diagnosis, and relates to an industrial equipment mechanism model fault diagnosis method based on a confusion matrix. Background Art

[0002] Industrial equipment is usually composed of multiple components, transmission paths and control systems, and its failure and failure may lead to production interruption, safety issues and increased equipment repair and replacement costs. Therefore, accurate and rapid diagnosis and location of equipment failures are of great significance for industrial production and operation and maintenance. Traditionally, fault diagnosis of industrial equipment mainly relies on regular inspections and manual experience. However, this method has problems such as low accuracy, high latency and dependence on professional skills. In recent years, with the rapid development of Internet of Things technology, the number of sensors and data acquisition devices deployed on industrial equipment has gradually increased, and a large amount of real-time data can be used for analysis and diagnosis. Among them, the mechanism model is based on the physical mechanism. By establishing models such as the working principle, physical process and transmission path of the equipment, the operating mechanism and fault behavior of the equipment can be more comprehensively understood. Through the simulation and analysis of the model, the behavior of the equipment in normal and faulty states can be predicted, and it can be compared and verified with the actual observation data to identify and locate the location and type of the fault.

[0003] However, the mechanism model-based method also faces some challenges. For example, after the mechanism model is developed, it needs to be verified multiple times on the computer to determine the accuracy of the model. As a result, the calibration time of the mechanism model is long and it is difficult to quickly apply it to fault diagnosis. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides an industrial equipment mechanism model fault diagnosis method based on confusion matrix.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A method for fault diagnosis of industrial equipment mechanism model based on confusion matrix, comprising the following steps:

[0007] S1. Data collection: obtaining industrial equipment data during large-scale normal operation in historical periods, wherein the industrial equipment data includes power input parameters and equipment operation parameters;

[0008] S2. Using power input parameters as independent variables and equipment operation parameters as dependent variables, a preliminary mechanism model is constructed;

[0009] S3, obtaining several test sets of industrial equipment data with and without failures by using a computer, inputting the power input parameters of the test sets into the preliminary mechanism model, and outputting the predicted values ​​of the equipment operation parameters;

[0010] S4. Construct a confusion matrix using the actual values ​​and predicted values ​​of the equipment operating parameters of the test set;

[0011] S5. Using the evaluation indicators of the confusion matrix to verify the prediction performance of the mechanism model, the evaluation indicators include accuracy, recall, precision and F1 value;

[0012] S6. Analyze the error tendency of the mechanism model prediction based on the size of the evaluation index;

[0013] S7. Adjust the parameters of the preliminary mechanism model according to the error tendency;

[0014] S8, repeating steps S3-S7 until the evaluation index reaches a preset threshold, and determining the final mechanism model;

[0015] S9. Apply the final mechanism model to the real-time monitoring of industrial equipment. When the relative error between the predicted value and the actual value of the equipment operating parameter exceeds the preset threshold, an equipment failure alarm is issued.

[0016] Specifically, in step S4, the confusion matrix is ​​constructed, and the specific contents are as follows:

[0017] Determine true positives: When the relative error between the predicted value of the equipment operating parameter at the time of the fault by the preliminary mechanism model and the actual value exceeds the fault error threshold, it is marked as a true positive;

[0018] Determine true negative examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when the fault occurs by the preliminary mechanism model does not exceed the fault error threshold, it is marked as a true negative example;

[0019] Determine false positive examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when no fault occurs by the preliminary mechanism model does not exceed the normal error threshold, it is marked as a false positive example;

[0020] Determine false negative examples: When the relative error between the predicted value of the equipment operating parameter when no fault occurs and the actual value of the preliminary mechanism model exceeds the normal error threshold, it is marked as a false negative example.

[0021] Furthermore, the fault error threshold indicates the minimum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment fails; the normal error threshold indicates the maximum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment does not fail.

[0022] Furthermore, in step S5, the accuracy rate is calculated as follows:

[0023] ,

[0024] In the formula, ACC represents accuracy; TP represents true positive examples; TN represents true negative examples; FP represents false positive examples; and FN represents false negative examples.

[0025] Furthermore, in step S5, the accuracy is calculated as follows:

[0026] ,

[0027] Where PPV represents precision, TP represents true positives, and FP represents false positives.

[0028] Furthermore, in step S5, the recall rate is calculated as follows:

[0029] ,

[0030] Where TRP represents the recall rate; TP represents the true positive examples; and FN represents the false negative examples.

[0031] Furthermore, in step S5, the F1 value is calculated by the formula:

[0032] ,

[0033] Where, F1 represents the F1 value; PPV represents the precision; TRP represents the recall rate.

[0034] Specifically, in step S6, the error tendency predicted by the mechanism model is analyzed according to the size of the evaluation index, specifically by inputting the evaluation index into a preset clustering unit and outputting the corresponding error tendency level. The preset clustering unit includes the following construction steps:

[0035] S61. Collect evaluation indicator data samples of preliminary mechanism models for large-scale historical periods;

[0036] S62, normalization processing: normalizing each evaluation index;

[0037] S63, initializing cluster centers: randomly selecting several data objects in the data space as initial cluster centers;

[0038] S64, initializing clusters: calculating the Euclidean distances between all data objects and the initial cluster centers, and dividing each data object into the initial cluster centers with the smallest Euclidean distances to form initial clusters;

[0039] S65, updating the cluster center: calculating the average value of the data objects of each initial cluster, and taking the average value as the new cluster center;

[0040] S66. Determine the final cluster: repeatedly calculate the Euclidean distance between all data objects and the new cluster center, divide each data object into the new cluster center with the smallest Euclidean distance to form a new cluster; calculate the average value of each new cluster as the new cluster center again until the new cluster center no longer changes, determine the final cluster, and determine the error tendency level corresponding to each final cluster according to the numerical range of the evaluation index of the final cluster.

[0041] Furthermore, in step S62, the normalization process is calculated using the following formula:

[0042] ,

[0043] In the formula, Y i After normalization, i Evaluation indicator data samples; K i For the i Evaluation indicator data samples; K max Represents the maximum value of all evaluation indicator data samples; K min Represents the minimum value of all evaluation indicator data samples.

[0044] Furthermore, in step S64, the Euclidean distance is calculated by the formula:

[0045] ,

[0046] In the formula, d ( x, C i ) represents the Euclidean distance function; x is the normalized data object; C i Indicates i Cluster centers; n is the number of samples in the data set; x j Indicates the data set j A data object with a variable; C ij Indicates j The variable i Cluster centers.

[0047] Furthermore, in step S7, the parameters of the preliminary mechanism model are adjusted, specifically by traversing the influence of all model parameters of the preliminary mechanism model on the evaluation index, and selecting model parameters with significant influence for adjustment.

[0048] Beneficial effects of the present invention:

[0049] (1) The confusion matrix is ​​used to evaluate the accuracy of the mechanism model of industrial equipment fault diagnosis. Only a small number of data samples are used in the process. This solves the problem that after the mechanism model is developed, it needs to be verified on the computer multiple times to determine the accuracy of the model, which leads to a long calibration time for the mechanism model.

[0050] (2) Using historical data, a clustering method is used to classify the evaluation indicators of the confusion matrix to obtain the error tendency of the mechanism model prediction. The error tendency is used to adjust the mechanism model parameters, thereby improving the efficiency and accuracy of the mechanism model correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0052] Figure 1 The present invention is a flowchart of a method for fault diagnosis of an industrial equipment mechanism model based on a confusion matrix. DETAILED DESCRIPTION

[0053] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0054] See also Figure 1 , the present application provides an industrial equipment mechanism model fault diagnosis method based on confusion matrix, comprising the following steps:

[0055] S1. Data collection: obtaining industrial equipment data during large-scale normal operation in historical periods, wherein the industrial equipment data includes power input parameters and equipment operation parameters;

[0056] In this embodiment, the power input parameters refer to the key parameters used to describe the energy supply required by the device. They are usually used to indicate the power or energy consumed by the device. Common power input parameters include: input voltage, input current, input power, input frequency, etc. These power input parameters are crucial to the normal operation and performance of the device. The operating parameters of the device refer to the parameters used to describe the operating characteristics and performance of the device when the device is in normal working condition. These parameters can cover speed, pressure, temperature, flow rate and other aspects. It can be seen that there is a close connection between the power input parameters and the device operating parameters. By monitoring the device operating parameters under specific power input parameters, it can be used to determine whether the device has a fault.

[0057] S2. Using power input parameters as independent variables and equipment operation parameters as dependent variables, a preliminary mechanism model is constructed;

[0058] In this embodiment, the mechanism model based on the power input parameter as the independent variable and the device operating parameter as the dependent variable is intended to study and explain the behavior and performance of the device. By analyzing the relationship between the power input and the device operating parameters, the model can predict the operating status of the device under given power input conditions. The mechanism model is a mathematical model based on known physical, chemical or other scientific principles and mechanisms to describe and explain the behavior of the system. It is not just a statistical model that describes the observed phenomenon, but through theoretical and empirical research, it explains the cause and effect relationship and interaction within the system from the bottom. The mechanism model can predict and explain the behavior of the system even under unobserved conditions. The mechanism model usually consists of one or more equations, each of which describes the relationship between a specific process or variables in the system. These equations can be linear or nonlinear and depend on system characteristics, parameters and initial conditions. The mechanism model can involve continuous time or discrete time, and can be deterministic or probabilistic, depending on the nature and requirements of the system under study. The establishment of the mechanism model requires steps such as careful selection and verification of model parameters, experimental verification and model fitting to ensure that the model is consistent with the actual situation and has predictive performance.

[0059] The establishment of the model includes the following steps: First, the collected equipment operation data under a series of different power input conditions are preprocessed and sorted. Then, the mathematical relationship between the independent variable (power input parameter) and the dependent variable (equipment operation parameter) is established by statistical analysis, correlation analysis and model fitting of the data. The mechanism model can be in the form of a function, such as a linear function, a nonlinear function or a polynomial function, and the specific form depends on the specific circumstances of the data and the problem. When establishing the model, the variables need to be transformed, normalized or smoothed to improve the model's fitting ability and prediction accuracy. After the mechanism model is established, the model can be used to predict the equipment operation parameters under unknown power input conditions. By inputting a specific power input value, the model can output the corresponding equipment operation parameter estimate, which helps us understand the working principle and performance characteristics of the equipment.

[0060] It should be noted that the initial mechanism model may have certain errors and uncertainties when directly applied to the computer. Therefore, it is important to verify and optimize the model. However, the existing technology often requires a large amount of computer data to verify the model. This verification method is slow and it is difficult to adjust the appropriate model parameters for application in predicting faults.

[0061] S3, obtaining several test sets of industrial equipment data with and without failures by using a computer, inputting the power input parameters of the test sets into the preliminary mechanism model, and outputting the predicted values ​​of the equipment operation parameters;

[0062] S4. Construct a confusion matrix using the actual values ​​and predicted values ​​of the equipment operating parameters of the test set;

[0063] In this embodiment, the confusion matrix, also known as the error matrix, is a tabular form used to evaluate the performance of the classification model. It visualizes and counts the prediction results of the classification model based on the actual category and the predicted category. The verification of the prediction results of the preliminary mechanism model by the confusion matrix is ​​divided into four categories, namely, true positive examples TP, true negative examples TN, false positive examples FP and false negative examples FN. Positive examples and negative examples depend on the specific problems and the goals of concern. The two are contradictory results. For example, in this embodiment, the failure of the test set can be regarded as a positive example, and the failure can be regarded as a negative example. Among them, TP (True Positive) indicates the number of positive examples correctly predicted by the model as positive examples; FN (False Negative) indicates the number of positive examples incorrectly predicted by the model as negative examples; FP (False Positive) indicates the number of negative examples incorrectly predicted by the model as positive examples; TN (True Negative) indicates the number of negative examples correctly predicted by the model as negative examples.

[0064] Specifically, in step S4, the confusion matrix is ​​constructed, and the specific contents are as follows:

[0065] Determine true positives: When the relative error between the predicted value of the equipment operating parameter at the time of the fault by the preliminary mechanism model and the actual value exceeds the fault error threshold, it is marked as a true positive;

[0066] Determine true negative examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when the fault occurs by the preliminary mechanism model does not exceed the fault error threshold, it is marked as a true negative example;

[0067] In this embodiment, the preliminary mechanism model is constructed based on the industrial equipment data during normal operation. When the power input parameters at the time of the fault are input into the preliminary mechanism model, the output equipment operation parameter prediction value should have a large deviation from the actual value. Therefore, when the relative error between the predicted value of the equipment operation parameter at the time of the fault by the preliminary mechanism model and the actual value exceeds the fault error threshold (a large error indicates a fault), it means that the model correctly predicts the positive example of the fault as a positive example. Similarly, when the predicted value and the actual value are the same or similar, it means that the preliminary mechanism model mistakenly believes that the operation data at the time of the equipment fault is normal, that is, a true negative example.

[0068] Determine false positive examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when no fault occurs by the preliminary mechanism model does not exceed the normal error threshold, it is marked as a false positive example;

[0069] Determine false negative examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when no fault occurs by the preliminary mechanism model exceeds the normal error threshold, it is marked as a false negative example.

[0070] It should be noted that, in this embodiment, the fault error threshold indicates the minimum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment fails; the normal error threshold indicates the maximum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment does not fail.

[0071] S5. Using the evaluation indicators of the confusion matrix to verify the prediction performance of the mechanism model, the evaluation indicators include accuracy, recall, precision and F1 value;

[0072] In this embodiment, accuracy is the ratio of the number of samples correctly predicted by the model to the total number of samples, and is one of the most commonly used evaluation indicators; precision measures how many of the positive examples predicted by the model are truly positive examples, and is an important indicator for evaluating the accuracy of model predictions; recall measures the ratio of the number of positive examples that the model can correctly predict to the actual number of positive examples, also known as recall; F1 value is the harmonic mean of precision and recall, which comprehensively considers the accuracy and recall of the model. The higher the F1 value, the better the balance between precision and recall.

[0073] Furthermore, in step S5, the accuracy rate is calculated as follows:

[0074] ,

[0075] In the formula, ACC represents accuracy; TP represents true positive examples; TN represents true negative examples; FP represents false positive examples; and FN represents false negative examples.

[0076] Furthermore, in step S5, the accuracy is calculated as follows:

[0077] ,

[0078] Where PPV represents precision, TP represents true positives, and FP represents false positives.

[0079] Furthermore, in step S5, the recall rate is calculated as follows:

[0080] ,

[0081] Where TRP represents the recall rate; TP represents the true positive examples; and FN represents the false negative examples.

[0082] Furthermore, in step S5, the F1 value is calculated by the formula:

[0083] ,

[0084] Where F1 represents the F1 value; PPV represents the precision; TRP represents the recall rate.

[0085] S6. Analyze the error tendency of the mechanism model prediction based on the size of the evaluation index;

[0086] In this embodiment, the error tendency of the mechanism model prediction is analyzed according to the size of the evaluation index: (1) If the accuracy of the mechanism model is very high, it means that the model performs well in classification as a whole, that is, the number of correctly predicted samples accounts for a high proportion of the total number of samples. If the accuracy is low, there may be a large number of misclassifications, that is, the model misclassifies positive examples as negative examples or misclassifies negative examples as positive examples, resulting in a decrease in overall performance; (2) A high recall rate means that the model can effectively identify true positive examples, that is, among all true positive examples, the model successfully finds a high proportion. If the recall rate is low, there may be omissions, that is, the model fails to correctly identify all true positive examples, resulting in some positive examples being misclassified as negative examples. (3) A high precision rate means that when the model classifies a sample as a positive example, it is more likely that it is a positive example, that is, the model has a high precision rate. If the precision rate is low, there may be false positives, that is, the model misclassifies some negative examples as positive examples, resulting in a high false positive rate. (4) The F1 score takes into account both precision and recall and is the harmonic mean of the two. Therefore, if the F1 score of the model is high, it means that the model has good performance in both precision and recall, that is, the overall performance of the model is good. If the F1 score is low, there may be a balance problem between precision and recall, that is, the model has not achieved a good balance between precision and recall. Among them, precision and recall are more suitable for analyzing the error tendency of the mechanism model, that is, the tendency to predict positive examples as negative examples or negative examples as positive examples.

[0087] Specifically, in step S6, the error tendency predicted by the mechanism model is analyzed according to the size of the evaluation index, specifically by inputting the evaluation index into a preset clustering unit and outputting the corresponding error tendency level. The preset clustering unit includes the following construction steps:

[0088] In this embodiment, by collecting large-scale historical period confusion matrix evaluation index data samples, a reference can be provided for the error tendency of the current mechanism model, and a clustering method can be used to classify data samples with similar evaluation index characteristics into one category, that is, such data samples have similar error tendencies. Among them, K-means clustering is a common unsupervised learning algorithm for dividing data points into different clusters. By calculating the similarity between data points, the K-means algorithm can classify data points into the clusters with the closest distance. This method helps to discover hidden patterns and structures in the data, thereby realizing cluster analysis and classification of data. The K-means algorithm is used to divide the evaluation index data samples into different clusters, each cluster represents a range of evaluation indicators with similar characteristics, and these characteristics can reflect the accuracy of the mechanism model. By analyzing the range of evaluation indicators in different clusters, the error tendency can be classified, thereby assisting management decision-making and problem handling.

[0089] S61. Collect evaluation indicator data samples of preliminary mechanism models for large-scale historical periods;

[0090] S62, normalization processing: normalizing each evaluation index;

[0091] In this embodiment, since there are multiple types of evaluation indicators (accuracy, recall rate, precision and F1 value), that is, multiple variables, the purpose of normalizing the data is to unify the data ranges between different features so that they have the same scale. This can avoid the problem of weight imbalance between different features due to different data ranges, and ensure that each feature can be properly reflected in the clustering process. Through normalization, the dimensional influence between data can be eliminated, making the data easier to compare and analyze, thereby improving the accuracy and efficiency of clustering. In addition, normalization can also make the clustering algorithm more stable and converge faster, which is conducive to obtaining more accurate and reliable clustering results. Therefore, before performing cluster analysis on the evaluation indicators, the data is usually normalized to achieve better clustering effects.

[0092] S63, initializing cluster centers: randomly selecting several data objects in the data space as initial cluster centers;

[0093] In this embodiment, when initializing the cluster centers, a common practice is to randomly select several data objects in the data space as the initial cluster centers. The reason for this is that by randomly selecting the initial cluster centers, it can be ensured that the algorithm will not be affected by the position of a specific initial cluster center, thereby avoiding falling into a local optimal solution. By randomly selecting data objects in the data space as the initial cluster centers, the algorithm can have a certain degree of randomness, which helps to explore different aspects of the data, and can increase the robustness and reliability of the algorithm, and improve the quality of the final clustering results. Randomly selecting data objects as initial cluster centers is a common and effective initialization method.

[0094] S64, initializing clusters: calculating the Euclidean distances between all data objects and the initial cluster centers, and dividing each data object into the initial cluster centers with the smallest Euclidean distances to form initial clusters;

[0095] S65, updating the cluster center: calculating the average value of the data objects of each initial cluster, and taking the average value as the new cluster center;

[0096] S66. Determine the final cluster: repeatedly calculate the Euclidean distance between all data objects and the new cluster center, divide each data object into the new cluster center with the smallest Euclidean distance to form a new cluster; calculate the average value of each new cluster as the new cluster center again until the new cluster center no longer changes, determine the final cluster, and determine the error tendency level corresponding to each final cluster according to the numerical range of the evaluation index of the final cluster.

[0097] Furthermore, in step S62, the normalization process is calculated using the following formula:

[0098] ,

[0099] In the formula, Y i After normalization, i Evaluation indicator data samples; K i For the i Evaluation indicator data samples; K max Represents the maximum value of all evaluation indicator data samples; K min Represents the minimum value of all evaluation indicator data samples.

[0100] Furthermore, in step S64, the Euclidean distance is calculated by the formula:

[0101] ,

[0102] In the formula,d ( x, C i ) represents the Euclidean distance function; x is the normalized data object; C i Indicates i Cluster centers; n is the number of samples in the data set; x j Indicates the data set j A data object with a variable; C ij Indicates j The variable i Cluster centers.

[0103] In this embodiment, since the mechanism model is a model with certain parameters, after determining the error tendency of the mechanism model prediction, the prediction errors of the positive and negative examples can be corrected by adjusting the parameters of the model to improve the accuracy of the model.

[0104] S7. Adjust the parameters of the preliminary mechanism model according to the error tendency;

[0105] S8, repeating steps S3-S7 until the evaluation index reaches a preset threshold, and determining the final mechanism model;

[0106] S9. Apply the final mechanism model to the real-time monitoring of industrial equipment. When the relative error between the predicted value and the actual value of the equipment operating parameter exceeds the preset threshold, an equipment failure alarm is issued.

[0107] Furthermore, in step S7, the parameters of the preliminary mechanism model are adjusted, specifically by traversing the influence of all model parameters of the preliminary mechanism model on the evaluation index, and selecting model parameters with significant influence for adjustment.

[0108] In this embodiment, the verification and correction of the mechanism model is an iterative process, that is, the error tendency of the prediction can be obtained by verifying the mechanism model, and the model is corrected by the tendency, and the corrected model gradually tends to be accurate, until the prediction accuracy of the mechanism model reaches the requirements of the confusion matrix evaluation index.

[0109] Beneficial effects of the present invention:

[0110] (1) The confusion matrix is ​​used to evaluate the accuracy of the mechanism model of industrial equipment fault diagnosis. Only a small number of data samples are used in the process. This solves the problem that after the mechanism model is developed, it needs to be verified on the computer multiple times to determine the accuracy of the model, which leads to a long calibration time for the mechanism model.

[0111] (2) Using historical data, a clustering method is used to classify the evaluation indicators of the confusion matrix to obtain the error tendency of the mechanism model prediction. The error tendency is used to adjust the mechanism model parameters, thereby improving the efficiency and accuracy of the mechanism model correction.

[0112] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for fault diagnosis of industrial equipment mechanism model based on confusion matrix, characterized by: The following steps are involved: S1. Data collection: obtaining industrial equipment data during large-scale normal operation in historical periods, wherein the industrial equipment data includes power input parameters and equipment operation parameters; S2. Using power input parameters as independent variables and equipment operation parameters as dependent variables, a preliminary mechanism model is constructed; S3, obtaining several test sets of industrial equipment data with and without failures by using a computer, inputting the power input parameters of the test sets into the preliminary mechanism model, and outputting the predicted values ​​of the equipment operation parameters; S4. Construct a confusion matrix using the actual values ​​and predicted values ​​of the equipment operating parameters of the test set; S5. Using the evaluation indicators of the confusion matrix to verify the prediction performance of the mechanism model, the evaluation indicators include accuracy, recall, precision and F1 value; S6. Analyze the error tendency of the mechanism model prediction based on the size of the evaluation index; S7. Adjust the parameters of the preliminary mechanism model according to the error tendency; S8, repeating steps S3-S7 until the evaluation index reaches a preset threshold, and determining the final mechanism model; S9. Apply the final mechanism model to the real-time monitoring of industrial equipment. When the relative error between the predicted value and the actual value of the equipment operating parameter exceeds a preset threshold, an equipment failure alarm is issued; In step S4, the confusion matrix is ​​constructed, and the specific contents are as follows: Determine true positives: When the relative error between the predicted value of the equipment operating parameter at the time of the fault by the preliminary mechanism model and the actual value exceeds the fault error threshold, it is marked as a true positive; Determine true negative examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when the fault occurs by the preliminary mechanism model does not exceed the fault error threshold, it is marked as a true negative example; Determine false positive examples: When the relative error between the predicted value and the actual value of the equipment operating parameter when no fault occurs by the preliminary mechanism model does not exceed the normal error threshold, it is marked as a false positive example; Determine false negative examples: When the relative error between the predicted value of the equipment operating parameter when no fault occurs and the actual value of the preliminary mechanism model exceeds the normal error threshold, it is marked as a false negative example; In step S6, the error tendency predicted by the mechanism model is analyzed according to the size of the evaluation index, specifically by inputting the evaluation index into a preset clustering unit and outputting the corresponding error tendency level. The preset clustering unit includes the following construction steps: S61. Collect evaluation indicator data samples of preliminary mechanism models for large-scale historical periods; S62, normalization processing: normalizing each evaluation index; S63, initializing cluster centers: randomly selecting several data objects in the data space as initial cluster centers; S64, initializing clusters: calculating the Euclidean distances between all data objects and the initial cluster centers, and dividing each data object into the initial cluster centers with the smallest Euclidean distances to form initial clusters; S65, updating the cluster center: calculating the average value of the data objects of each initial cluster, and taking the average value as the new cluster center; S66. Determine the final cluster: repeatedly calculate the Euclidean distance between all data objects and the new cluster center, divide each data object into the new cluster center with the smallest Euclidean distance to form a new cluster; calculate the average value of each new cluster as the new cluster center again until the new cluster center no longer changes, determine the final cluster, and determine the error tendency level corresponding to each final cluster according to the numerical range of the evaluation index of the final cluster.

2. The method for industrial equipment mechanism model fault diagnosis based on confusion matrix according to claim 1, characterized in that: The fault error threshold represents the minimum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment fails; The normal error threshold represents the maximum value allowed for the relative error between the predicted value and the actual value of the monitored equipment operating parameter when the equipment is not faulty.

3. The method for industrial equipment mechanism model fault diagnosis based on confusion matrix according to claim 1 is characterized in that: In step S5, the accuracy is calculated using the following formula: , , , , In the formula, ACC represents accuracy; PPV represents precision; TRP represents recall; F1 represents F1 value; TP represents true positive examples; TN represents true negative examples; FP represents false positive examples; and FN represents false negative examples.

4. The method for industrial equipment mechanism model fault diagnosis based on confusion matrix according to claim 1 is characterized in that: In step S62, the normalization process is performed using the following calculation formula: , In the formula, Y i After normalization, i Evaluation indicator data samples; K i For the i Evaluation indicator data samples; K max Represents the maximum value of all evaluation indicator data samples; K min Represents the minimum value of all evaluation indicator data samples.

5. The method for industrial equipment mechanism model fault diagnosis based on confusion matrix according to claim 1 is characterized in that: In step S64, the Euclidean distance is calculated using the following formula: , In the formula, d ( x, C i ) represents the Euclidean distance function; x is the normalized data object; C i Indicates i Cluster centers; n is the number of samples in the data set; x j Indicates the data set j A data object with a variable; C ij Indicates j The variable i Cluster centers.

6. The method for industrial equipment mechanism model fault diagnosis based on confusion matrix according to claim 1, characterized in that: In step S7, the parameters of the preliminary mechanism model are adjusted, specifically by traversing the influence of all model parameters of the preliminary mechanism model on the evaluation index, and selecting model parameters with significant influence for adjustment.