Fault detection method and device for dust removal system

Abstract

The invention relates to the technical field of fault detection, and discloses a dust removal system fault detection method and device, and the method comprises the steps: obtaining first historical data corresponding to a dust removal system, and carrying out the preprocessing of the first historical data, and obtaining a first data set and a second data set; training a preset model based on the first data set to obtain an initial model; testing the initial model based on the second data set and determining a target model according to a test result; and obtaining abnormal data points based on the operation data and the target model, and carrying out statistics on the occurrence probability of the fault condition. The first data set and the second data set are obtained through the first historical data, then model training and testing are conducted through the first data set and the second data set, the target model is obtained according to the testing result, the operation data are input into the target model to obtain the abnormal data points, and then the fault occurrence probability is determined. More efficient fault detection of the dust removal system is realized, and the labor cost is also saved.

Description

Technical Field

[0001] The present application relates to the technical field of fault detection, and in particular to a dust removal system fault detection method and device. Background Art

[0002] The dust removal system of a cigarette factory is an important part of ensuring a clean production environment, protecting the health of employees, and improving product quality. A large amount of dust is generated during the cigarette production process, including tobacco dust, flavor dust, and dust generated by other auxiliary materials. These dusts not only affect the air quality of the working environment, but may also affect the normal operation of production equipment. Therefore, in order to ensure the quality of the production environment and ensure the normal operation of the dust removal system, it is becoming increasingly important to perform fault detection on the dust removal system to avoid danger and loss. The existing detection scheme is mainly manual detection, such as checking whether there is obvious damage or leakage on the outside of the dust collector, checking whether the connection is tight, observing whether the filter bag is obviously damaged, blocked or detached, and listening to whether the sound of the fan is normal.

[0003] However, this solution relies too much on manual experience and subjective judgment, which not only increases labor costs but also has low detection efficiency. Therefore, it is of great significance for cigarette factories to adopt a low-cost and efficient fault detection solution for dust removal systems. Summary of the invention

[0004] In order to solve the above-mentioned problems of high labor cost and low detection efficiency, the embodiment of the present application provides a dust removal system fault detection method and device, and its technical solution is as follows: In a first aspect, an embodiment of the present application provides a dust removal system fault detection method, comprising: Acquire first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each historical data point, and the second data set includes at least two historical data points and a label value corresponding to each historical data point, and the number of historical data points in the second data set is less than that in the first data set; Training a preset model based on all historical data points in the first data set and a label value corresponding to each historical data point to obtain an initial model; Testing the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determining the target model according to the test result; Based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the probability of occurrence of at least two fault conditions is statistically calculated based on the abnormal data points.

[0005] In an optional solution of the first aspect, preprocessing the first historical data to obtain the first data set and the second data set includes: The first historical data is supplemented according to a preset formula to obtain an initial data set; Obtain the current timestamp, and generate a random number according to the current timestamp and a preset first algorithm; The initial data set is divided according to a preset ratio and a random number to obtain a first data set and a second data set.

[0006] In another optional solution of the first aspect, the initial model is tested based on all historical data points in the second data set and the label value corresponding to each historical data point, and the target model is determined according to the test result, including: Input the historical data points in the second data set into the initial model to obtain a test label set; Calculating a first accuracy rate based on the label values ​​corresponding to the historical data points in the second data set and the test label set; The optimal hyperparameters are determined according to the first accuracy, and the target model is obtained according to the optimal hyperparameters.

[0007] In another optional solution of the first aspect, determining an optimal hyperparameter according to the first accuracy rate, and obtaining a target model according to the optimal hyperparameter includes: When it is detected that the first accuracy rate is higher than or equal to the preset accuracy rate, the initial model is used as the target model; When it is detected that the first accuracy rate is lower than a preset accuracy rate, an optimal hyperparameter is calculated according to a preset second algorithm; The target model is obtained according to the optimal hyperparameters.

[0008] In another optional solution of the first aspect, obtaining a target model according to optimal hyperparameters includes: The initial model is trained according to the optimal hyperparameters and the first data set to obtain a candidate model; Calculate a second accuracy rate of the candidate model based on the second data set; When it is detected that the second accuracy rate is higher than or equal to the preset accuracy rate, the candidate model is used as the target model; When it is detected that the second accuracy rate is lower than the preset accuracy rate, second historical data corresponding to the dust removal system is acquired; wherein the time corresponding to the first historical data is different from the time corresponding to the second historical data.

[0009] In another optional solution of the first aspect, based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the probability of occurrence of at least two fault conditions is statistically calculated based on the abnormal data points, including: Performing structured processing on the operation data corresponding to the dust removal system to obtain at least two detection data points; Input all detection data points into the target model to obtain at least two label values ​​corresponding to the detection data points; When it is identified that the label value is within the preset value range, the detection data point corresponding to the label value is regarded as a normal data point; When it is identified that the label value is not within the preset value range, the detection data point corresponding to the label value is regarded as an abnormal data point; Compare the abnormal data point with the preset standard data point, and calculate the deviation value of at least two performance parameters; wherein the abnormal data point includes at least two performance parameters, and the preset standard data point includes at least two performance parameters; According to the deviation values ​​of at least two performance parameters, the occurrence probability of the fault situation corresponding to each performance parameter is determined.

[0010] In yet another optional solution of the first aspect, after determining the target model according to the test result, the method further includes: Receive real-time data sent by the digital twin system and obtain fault detection results through the target model and real-time data; The fault detection result is sent to the digital twin system, so that the digital twin system generates a visualization interface corresponding to the fault detection result.

[0011] In a second aspect, an embodiment of the present application provides a dust removal system fault detection device, comprising: A first processing module is used to obtain first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each historical data point, and the second data set includes at least two historical data points and a label value corresponding to each historical data point, and the number of historical data points in the second data set is less than that in the first data set; A second processing module is used to train a preset model based on all historical data points in the first data set and a label value corresponding to each historical data point to obtain an initial model; A third processing module is used to test the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determine the target model according to the test result; The fourth processing module is used to determine abnormal data points based on the operating data corresponding to the dust removal system and the target model, and to calculate the probability of occurrence of at least two fault conditions according to the abnormal data points.

[0012] In a third aspect, an embodiment of the present application further provides a dust removal system fault detection device, including a processor and a memory; The processor is connected to the memory; A memory for storing executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the dust removal system fault detection method provided by the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect.

[0013] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, the dust removal system fault detection method provided by the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect can be implemented.

[0014] The beneficial effects brought by the technical solutions provided by some embodiments of this specification include at least: In the dust removal system fault detection process, the first historical data of the dust removal system is preprocessed to obtain a first data set and a second data set, and then the preset model is trained by the first data set, and the trained model is tested by the second data set, so as to determine the target model according to the test results, and then the operating data of the dust removal system is input into the target model to obtain abnormal data points and determine the probability of occurrence of at least two fault conditions, which not only realizes more efficient dust removal system fault detection, but also saves the labor cost of manual detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 An overall flow chart of a dust removal system fault detection method provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a dust removal system fault detection device provided in an embodiment of the present application; Figure 3 A schematic structural diagram of another dust removal system fault detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0018] In the following introduction, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The following introduction provides multiple embodiments of the present application, and different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments recorded. Therefore, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing one or more of all other possible combinations of A, B, C, and D, although the embodiment may not be clearly recorded in the following text.

[0019] The following description provides examples and does not limit the scope, applicability or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present application. Various processes or components may be appropriately omitted, substituted or added to each example. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted or combined. In addition, features described in some examples may be combined in other examples.

[0020] See also Figure 1 , Figure 1 An overall flow chart of a dust removal system fault detection method provided in an embodiment of the present application is shown.

[0021] like Figure 1 As shown, the dust removal system fault detection method may at least include the following steps: Step 101: Acquire first historical data corresponding to a dust removal system, and preprocess the first historical data to obtain a first data set and a second data set.

[0022] In the embodiment of the present application, the dust removal system fault detection method can be but is not limited to being applied to terminal devices, and the terminal devices can be but are not limited to desktop computers, laptop computers, and tablet computers. In the dust removal system fault detection process, the first historical data of the dust removal system obtained is preprocessed to obtain a first data set and a second data set, and then the preset model is trained by the first data set, and the trained model is tested by the second data set, so as to determine the target model according to the test results, and then the operation data of the dust removal system is input into the target model to obtain abnormal data points and determine the probability of occurrence of at least two fault conditions, which not only realizes more efficient dust removal system fault detection, but also saves the labor cost of manual detection.

[0023] Specifically, during the dust removal system fault detection process, the real-time data of the dust removal system can be collected through various sensors (such as temperature sensors, pressure sensors, humidity sensors, dust concentration sensors, vibration sensors, etc.), data acquisition cards (such as USB data acquisition cards, PCIe data acquisition cards, etc.) and wireless sensor networks (such as Wi-Fi, Bluetooth, LoRa and other wireless communication technologies), so as to obtain the first historical data of the dust removal system. Among them, the data acquisition card is mainly used to connect sensors and convert analog signals into digital signals.

[0024] Furthermore, after obtaining the first historical data, necessary preprocessing may be performed on the first historical data, including data cleaning, format conversion, etc., and then the preprocessed data may be divided into a first data set and a second data set according to certain division rules, wherein the division rules may include but are not limited to proportional division, random sampling, etc.

[0025] It should be noted that the first data set (i.e., training set) includes at least two historical data points and the label value corresponding to each historical data point. It is a set of data used to train the model. The model will learn how to map input to output through these data. The second data set (i.e., test set) includes at least two historical data points and the label value corresponding to each historical data point. It is a set of data independent of the training set and used to evaluate the generalization ability of the model. Usually, the data in the training set accounts for a larger proportion, while the data in the test set accounts for a smaller proportion. Therefore, the number of historical data points in the second data set is less than that in the first data set.

[0026] It can be understood that a data point is an observation or instance containing a set of features, usually represented as a vector or row; a data point consists of multiple features, each of which can be numerical, categorical, or textual; a data point can be represented by a row vector, where each column is a feature. Historical data points can include, but are not limited to, features such as temperature, pressure, air velocity, filter bag resistance, and operating time, and the label value corresponding to each historical data point can be, but is not limited to, a text label (such as "normal", "abnormal") or a numerical label (such as "+1", "-1").

[0027] As an option of the embodiment of the present application, preprocessing the first historical data to obtain the first data set and the second data set includes: The first historical data is supplemented according to a preset formula to obtain an initial data set; Obtain the current timestamp, and generate a random number according to the current timestamp and a preset first algorithm; The initial data set is divided according to a preset ratio and a random number to obtain a first data set and a second data set.

[0028] Specifically, after obtaining the first historical data, since there may be missing data points in the first historical data, and a complete data set is indispensable for data analysis and modeling, in order to estimate these missing data points to form a complete data set, the first historical data needs to be supplemented. Therefore, the first historical data can be supplemented according to a preset formula to obtain an initial data set, wherein the supplementation method may be but is not limited to cubic spline interpolation, piecewise linear interpolation, Hermite interpolation, and Lagrange interpolation. It should be noted that the preset formula may be but is not limited to a cubic polynomial. Since the cubic spline interpolation method can produce a continuous and smooth interpolation curve, so that the interpolation result is consistent with the original data points and has good smoothness, the cubic spline interpolation method is given priority when this scheme is implemented.

[0029] Next, after obtaining the initial data set, the current timestamp can be obtained, and then a random number can be generated based on the current timestamp and a preset first algorithm, where the current timestamp refers to data with a fixed format (such as seconds) representing the current time, and the preset first algorithm can be but is not limited to the linear congruential method, the Mersenne Twister algorithm, and the XORShift algorithm.

[0030] After generating the random number, a division rule can be determined according to the preset ratio and the random number, and then the initial data set can be divided according to the division rule to obtain the divided first data set and the second data set. For example, each data point in the initial data set can be sorted in chronological order, and each data point can be numbered in sequence (starting from 1, in order from small to large, taking integers for numbering), so as to obtain the number of data points, that is, the value of the last number. If the value is 100, the preset ratio is 80%, and the preset ratio represents the ratio of the data points in the first data set to all data points, then it can be calculated that the number of data points in the first data set should be 100*80%=80. Therefore, the random number can be limited to an integer between 1 and 100 (including both 1 and 100), and the number of random numbers generated can be limited to 80, so that the numbers of 80 data points can be obtained, and the data points corresponding to these numbers are collected as the first data set, and the remaining data points are collected as the second data set. It is worth noting that when the preset ratio is less than 50%, it can also represent the ratio of the data points in the second data set to all the data points, so as to satisfy the fact that the number of historical data points in the second data set is less than that in the first data set.

[0031] Step 102: Train a preset model based on all historical data points in the first data set and the label value corresponding to each historical data point to obtain an initial model.

[0032] Specifically, after obtaining the first data set, the first data set can be used as a training set to train a preset model to generate an initial model, wherein the preset model can be but is not limited to an OCSVM model (One-Class Support Vector Machin model).

[0033] It is understandable that during the model training process, the historical data points in the first data set can be used as the input data of the model, and the label value corresponding to each historical data point can be used as the output data of the model. The main purpose of model training is to determine several key parameters of the model, and then determine the mapping relationship between input and output, so that the model can distinguish between normal samples (i.e., the data points detected when the dust removal system is operating normally and the corresponding label values) and abnormal samples (i.e., the data points detected when the dust removal system fails and the corresponding label values). Among them, the key parameters include weight vectors, offsets, and support vectors, etc. These parameters are in an undetermined state before training and are converted into determined values ​​after training.

[0034] Step 103: Test the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determine the target model according to the test result.

[0035] Specifically, after obtaining the second data set, the second data set can be used as a test set to test the initial model, and then the determined values ​​of the hyperparameters can be obtained according to the test results, thereby generating a target model.

[0036] It is understandable that during the model testing process, the historical data points in the second data set can be used as the input data of the model, and the label value corresponding to each historical data point can be used as the output data of the model. The main purpose of model testing is to determine the performance of the model on unseen data to evaluate its generalization ability. The effectiveness and applicability of the model can be measured by calculating various performance indicators (such as accuracy, precision, recall, F1 score, AUC-ROC, etc.).

[0037] Furthermore, after obtaining the test results of the initial model (such as the accuracy A), it is also possible to determine whether the hyperparameters need to be adjusted based on the test results (such as A is less than 80%, it is considered that the hyperparameters need to be adjusted), where hyperparameters refer to configuration options or parameters set before model training. They are external settings that control the model training process and behavior and can affect the performance of the model. It is worth noting that the parameters obtained by model training are different from the hyperparameters here. The hyperparameters here need to be set before model training, while the parameters obtained by model training are learnable variables within the model, which are learned from the data during the model training process.

[0038] After obtaining the judgment result, when the hyperparameters need to be adjusted, the value of the adjusted hyperparameters can be determined through relevant algorithms or personal experience, and the hyperparameters can be updated according to the value to obtain the target model; when the hyperparameters do not need to be adjusted, the initial model can be used as the target model.

[0039] As another option of the embodiment of the present application, the initial model is tested based on all historical data points in the second data set and the label value corresponding to each historical data point, and the target model is determined according to the test result, including: Input the historical data points in the second data set into the initial model to obtain a test label set; Calculating a first accuracy rate based on the label values ​​corresponding to the historical data points in the second data set and the test label set; The optimal hyperparameters are determined according to the first accuracy, and the target model is obtained according to the optimal hyperparameters.

[0040] Specifically, in the process of testing the initial model with the second data set and determining the target model according to the test results, the historical data points in the second data set can be first input into the initial model as input data, and then the model outputs the prediction results (i.e., test labels) corresponding to each historical data point, thereby obtaining a test label set. It is understandable that the test label set includes multiple test labels, each of which can be, but is not limited to, a text label (such as "normal", "abnormal") or a numerical label (such as "+1", "-1").

[0041] Next, after obtaining the test label set, the test label corresponding to each data point in the test label set can be compared with the corresponding label value in the second data set. When the two are the same, the prediction result of the data point is considered to be correct, otherwise it is considered to be incorrect. After comparing all historical data points, the first accuracy rate can be calculated based on the number of historical data points with correct prediction results and the number of all historical data points, that is, the ratio of the former to the latter.

[0042] After calculating the first accuracy, if the first accuracy meets the requirements (such as reaching a known accuracy or above), the current hyperparameters are regarded as optimal hyperparameters; if the first accuracy does not meet the requirements (such as failing to reach a known accuracy), the hyperparameters may be adjusted by, but are not limited to, manual adjustment, grid search, random search, Bayesian optimization, genetic algorithm, etc., and the hyperparameters corresponding to the time when the accuracy of the model meets the requirements are regarded as optimal hyperparameters.

[0043] Thereafter, the values ​​of the hyperparameters of the initial model can be modified to the values ​​of the optimal hyperparameters, thereby obtaining a target model with the optimal hyperparameters.

[0044] As another option of the embodiment of the present application, determining the optimal hyperparameter according to the first accuracy rate, and obtaining the target model according to the optimal hyperparameter includes: When it is detected that the first accuracy rate is higher than or equal to the preset accuracy rate, the initial model is used as the target model; When it is detected that the first accuracy rate is lower than a preset accuracy rate, an optimal hyperparameter is calculated according to a preset second algorithm; The target model is obtained according to the optimal hyperparameters.

[0045] Specifically, in the process of determining the optimal hyperparameters according to the first accuracy rate and obtaining the target model according to the optimal hyperparameters, the numerical relationship between the first accuracy rate and the preset accuracy rate can be judged first. When the judgment result is that the first accuracy rate is higher than or equal to the preset accuracy rate, it means that the accuracy rate has met the requirements. The hyperparameters at this time can be regarded as the optimal hyperparameters, so the initial model at this time can be used as the target model; when the judgment result is that the first accuracy rate is lower than the preset accuracy rate, the optimal hyperparameters can be calculated by a preset second algorithm, wherein the preset second algorithm can be but is not limited to a clone selection algorithm.

[0046] It should be noted that the clone selection algorithm is a heuristic optimization algorithm based on the immune system. Its core idea is borrowed from immunology, that is, how the immune system recognizes and fights foreign antigens. During the clone selection process, when the immune system encounters an antigen, it produces a large number of antibody clones, which then undergo a process of mutation and selection, and eventually produce the most effective antibodies to fight the antigen. The working principle of the clone selection algorithm in determining the optimal hyperparameter is to find the optimal or approximately optimal hyperparameter combination by simulating the clone selection process in the biological immune system. It is worth noting that the optimal hyperparameter can be a single hyperparameter or a combination of two or more hyperparameters.

[0047] The application of the clone selection algorithm in this scheme mainly includes the following steps: 1. Initialize the population: Create a set of random hyperparameter combinations as "antibodies". These hyperparameter combinations constitute the initial population of the algorithm; 2. Evaluate fitness: Use some evaluation index (such as cross-validation score) to evaluate the performance of each set of hyperparameters on the training data. Hyperparameter combinations with high fitness are considered to be better "antibodies"; 3. Cloning process: Select a hyperparameter combination with higher fitness for cloning. The number of clones can vary depending on the fitness. Generally, the higher the fitness, the more clones there are. 4. Mutation process: Mutate the cloned hyperparameter combinations to explore new hyperparameter combinations. Mutation can be performed by random perturbation, adding or subtracting values, etc. 5. Update the population: Update the population according to the new fitness value of the hyperparameter combination after the mutation, retain the hyperparameter combination with higher fitness, and eliminate the hyperparameter combination with lower fitness; 6. Termination condition: Repeat the above steps until the preset number of iterations is reached or the fitness no longer improves significantly (such as the degree of improvement is within the preset range).

[0048] It can be understood that fitness is a quantitative indicator used to measure the effectiveness and quality of an individual in solving a problem. The fitness function defines how to calculate the fitness value of an individual, and individuals with high fitness are individuals that perform better in a given problem. In addition, cross-validation is a statistical method for evaluating the performance of machine learning models to estimate the generalization ability of the model on unseen data. It does this by dividing the dataset into several mutually exclusive subsets and training and testing the model on different subsets. The main purpose of cross-validation is to reduce the risk of overfitting and obtain a more stable estimate of model performance.

[0049] Suppose we are using the clonal selection algorithm to optimize two key hyperparameters of a support vector machine (SVM) model: the penalty coefficient C and the Gaussian kernel width γ of the kernel function. In order to find the optimal (C, γ) combination, we can define the fitness function as: Fitness ((C, γ)) = −CrossValidationScore (C, γ). Among them, CrossValidationScore is the average performance indicator (such as accuracy) obtained using cross-validation.

[0050] After determining the optimal hyperparameters, the target model can be obtained by changing the values ​​of the hyperparameters of the initial model to the values ​​of the optimal hyperparameters.

[0051] As another option of the embodiment of the present application, obtaining a target model according to optimal hyperparameters includes: The initial model is trained according to the optimal hyperparameters and the first data set to obtain a candidate model; Calculate a second accuracy rate of the candidate model based on the second data set; When it is detected that the second accuracy rate is higher than or equal to the preset accuracy rate, the candidate model is used as the target model; When it is detected that the second accuracy rate is lower than the preset accuracy rate, second historical data corresponding to the dust removal system is acquired; wherein the time corresponding to the first historical data is different from the time corresponding to the second historical data.

[0052] Specifically, in the process of obtaining the target model according to the optimal hyperparameters, the values ​​of the hyperparameters of the initial model can be first changed to the values ​​of the optimal hyperparameters, and the initial model can be trained with the first data set to obtain a candidate model. The candidate model can then be tested according to the second data set to calculate the accuracy of the candidate model according to the test results, that is, the second accuracy rate, wherein the calculation method can refer to the calculation method of the first accuracy rate mentioned above.

[0053] Then, after calculating the second accuracy rate, the numerical relationship between the second accuracy rate and the preset accuracy rate can be judged. When the judgment result is that the second accuracy rate is higher than or equal to the preset accuracy rate, it means that the accuracy rate of the candidate model obtained through training has met the requirements, and the various parameters of the model have been adjusted appropriately, so the candidate model can be used as the target model; when the judgment result is that the second accuracy rate is lower than the preset accuracy rate, it means that the accuracy rate of the candidate model obtained through training does not meet the requirements, and it may be due to the poor model performance caused by the data set used for training (such as data imbalance, too much noise data, etc.), so it is necessary to obtain a new data set for training the model. Therefore, the second historical data corresponding to the dust removal system can be obtained, and the second historical data can be preprocessed to obtain a new data set for training the model, so that the target model can be obtained through training with the new data set, wherein the time corresponding to the first historical data is different from the time corresponding to the second historical data.

[0054] Step 104: Based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the occurrence probabilities of at least two fault conditions are statistically calculated based on the abnormal data points.

[0055] Specifically, after obtaining the target model, the operating data corresponding to the dust removal system (such as the operating time of the dust removal system, the particle concentration, temperature, pressure, flow rate, etc. corresponding to the operating time) can be input into the target model. Through the calculation of the model, the prediction labels of the data points corresponding to different operating times can be obtained, and then the abnormal data points indicating dust removal system failures can be screened out according to the values ​​of all prediction labels.

[0056] Assume that the operating data is as shown in the following table, and the predicted label is "+1" which means that the data point is normal (that is, the dust removal system works normally), and the predicted label is "-1" which means that the data point is abnormal (that is, the dust removal system fails).

[0057] It can be seen from the above table that the prediction labels of the data points corresponding to the running time of 10 hours and 12 hours are both "+1", and the prediction label of the data point corresponding to the running time of 20 hours is "-1". Therefore, the data point corresponding to the running time of 20 hours is an abnormal data point.

[0058] Furthermore, after the abnormal data points are determined, the occurrence probabilities of at least two fault conditions may be statistically calculated based on the abnormal data points, wherein the statistical method may be, but is not limited to, a frequency counting method and a cross-tabulation analysis method.

[0059] 1. Frequency counting method: The probability of occurrence of each fault type is obtained by calculating the ratio of the number of occurrences of each fault type to the total number of abnormal events. For example, assuming that there are 100 abnormal data points, 30 of which correspond to the "filter bag blockage" fault, and another 20 data points correspond to the "fan failure". The remaining data points correspond to other faults or normal conditions. The corresponding rules can be set based on historical fault conditions. Then, the probability of the "filter bag blockage" fault is P (filter bag blockage) = 30 / 100 = 0.3, and the probability of the "fan failure" is P (fan failure) = 20 / 100 = 0.2.

[0060] 2. Cross-table analysis method: Cross-tables (also called contingency tables) are mainly used to study the correlation between two categorical variables. Therefore, cross-tables can be used to calculate conditional probabilities to analyze the correlation between specific abnormal data features and fault types. If you need to analyze the correlation between the two abnormal data features of "normal pressure difference" and "abnormal vibration level" and the fault type, you can build a cross-table to record the number of occurrences of "normal pressure difference" and "abnormal vibration level" in the abnormal data points and the fault types caused by them. The fault types correspond to the abnormal data points one by one, and the corresponding rules can be set based on historical fault conditions. Suppose "normal pressure difference" occurs 60 times and "abnormal vibration level" occurs 40 times. Among these 60 "normal pressure difference", 30 times led to "filter bag blockage" and 10 times led to "fan failure". Among the 40 "abnormal vibration level", 10 times led to "filter bag blockage" and 20 times led to "fan failure". Therefore, the probability of various faults corresponding to the two abnormal data features of "normal pressure difference" and "abnormal vibration level" can be calculated through this information. For example, the probability of filter bag clogging due to pressure difference is P(filter bag clogging | pressure difference normal) = 30 / 60 = 0.5; the probability of fan failure due to pressure difference is P(fan failure | pressure difference normal) = 10 / 60 = 1 / 6 ≈ 0.167; the probability of filter bag clogging due to abnormal vibration level is P(filter bag clogging | vibration level abnormal) = 10 / 40 = 0.25; the probability of fan failure due to abnormal vibration level is P(fan failure | vibration level abnormal) = 20 / 40 = 0.5.

[0061] It is worth noting that the above calculated probability of occurrence of fault conditions can achieve more efficient fault detection of the dust collection system, which in turn can help the maintenance team prioritize the most likely problem areas, thereby improving maintenance efficiency.

[0062] As another option of the embodiment of the present application, based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the probability of occurrence of at least two fault conditions is statistically calculated based on the abnormal data points, including: Performing structured processing on the operation data corresponding to the dust removal system to obtain at least two detection data points; Input all detection data points into the target model to obtain at least two label values ​​corresponding to the detection data points; When it is identified that the label value is within the preset value range, the detection data point corresponding to the label value is regarded as a normal data point; When it is identified that the label value is not within the preset value range, the detection data point corresponding to the label value is regarded as an abnormal data point; Compare the abnormal data point with the preset standard data point, and calculate the deviation value of at least two performance parameters; wherein the abnormal data point includes at least two performance parameters, and the preset standard data point includes at least two performance parameters; According to the deviation values ​​of at least two performance parameters, the occurrence probability of the fault situation corresponding to each performance parameter is determined.

[0063] Specifically, in the process of inputting the operation data corresponding to the dust removal system into the target model to obtain abnormal data points, since the operation data exists in the form of original log files, it is impossible to analyze it directly. Therefore, it is necessary to structure it first for further analysis. The structured processing method can include but is not limited to the following 5 steps: 1. Data import: reading data into a computer program, usually done using a data processing library (such as Pandas) in a programming language (such as Python); 2. Data cleaning: Check whether there are missing values ​​or outliers in the data, and perform corresponding processing, such as deleting or filling missing values, and correcting outliers; 3. Data conversion: Convert data into a suitable format to facilitate subsequent analysis. For example, convert a string-formatted timestamp into a date-time object to facilitate operations such as sorting by date and time window analysis. 4. Feature engineering: Extract useful features as needed, for example, calculate dust removal efficiency, identify abnormal pressure loss, etc. 5. Data storage: Save the processed data into a suitable data structure or database for subsequent use.

[0064] After the operation data is structured, at least two detection data points can be obtained, each of which includes information such as temperature, humidity, pressure, and operation time. After obtaining the detection data points, all detection data points can be input into the target model, and the model will output at least two label values ​​corresponding to the detection data points after calculation.

[0065] Next, after obtaining the label value corresponding to each detection data point, an identification operation can be performed on the label value. When it is identified that the label value is in a preset numerical range (such as the label value is in a non-negative range), it means that the dust removal system is not faulty, and the detection data point corresponding to the label value can be used as a normal data point; when it is identified that the label value is not in the preset numerical range (such as the label value is in a negative range), it means that the dust removal system is faulty, and the detection data point corresponding to the label value can be used as an abnormal data point.

[0066] After obtaining the abnormal data point, the abnormal data point can also be compared with the preset standard data point, and the deviation value of at least two performance parameters can be calculated, wherein the abnormal data point and the preset standard data point both include at least two performance parameters, and the performance parameters can be but are not limited to temperature, humidity, pressure value, air flow velocity, filter bag resistance, etc. For example, if the temperature value in the abnormal data point is 40°C, and the temperature value at the preset standard data point is 35°C, then the deviation value is 40-35=5.

[0067] After calculating the deviation values ​​of at least two performance parameters, the failure probabilities corresponding to different value ranges can be estimated based on the range of the deviation values ​​and historical failure conditions, and stored in a database, which includes at least two failure probabilities corresponding to the deviation values ​​of at least two performance parameters, and then the probability of occurrence of the failure condition corresponding to each performance parameter can be determined through the database. For example, when the deviation value of humidity is less than 1, the failure probability is 5%; when the deviation value of humidity is greater than or equal to 1 and less than 2, the failure probability is 15%; when the deviation value of humidity is greater than or equal to 2 and less than 3, the failure probability is 30%; when the deviation value of humidity is greater than or equal to 3 and less than 4, the failure probability is 50%; when the deviation value of humidity is greater than or equal to 4, the failure probability is 75%.

[0068] As another option of the embodiment of the present application, after determining the target model according to the test results, the method further includes: Receive real-time data sent by the digital twin system and obtain fault detection results through the target model and real-time data; The fault detection result is sent to the digital twin system, so that the digital twin system generates a visualization interface corresponding to the fault detection result.

[0069] Specifically, after obtaining the target model, the target model can also be combined with the digital twin system to detect the faults of the dust removal system. In this process, the digital twin system will perform real-time detection on the dust removal system and send the real-time data generated by the real-time detection to the terminal device where the target model is located, so as to calculate all the label values ​​corresponding to the real-time data through the target model, and obtain the fault detection results according to all the label values ​​(such as the label value is -1 when the running time is X, and the temperature of the corresponding part A of the dust removal system exceeds the normal threshold), and then the fault detection results can be sent to the digital twin system, so that the digital twin system generates a visual interface corresponding to the fault detection results (such as generating a temperature change curve diagram of part A of the dust removal system or a temperature distribution diagram of part A), so that the administrator of the dust removal system can understand the fault situation of the dust removal system more intuitively, clearly and quickly, so as to timely adjust the dust removal system and make targeted maintenance decisions for the dust removal system in real time.

[0070] See also Figure 2 , Figure 2 A schematic structural diagram of a dust removal system fault detection device provided in an embodiment of the present application is shown.

[0071] like Figure 2 As shown, the dust removal system fault detection device may include at least a first processing module 201, a second processing module 202, a third processing module 203 and a fourth processing module 204, wherein: The first processing module 201 is used to obtain first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each historical data point, and the second data set includes at least two historical data points and a label value corresponding to each historical data point, and the number of historical data points in the second data set is less than that in the first data set; The second processing module 202 is used to train a preset model based on all historical data points in the first data set and the label value corresponding to each historical data point to obtain an initial model; The third processing module 203 is used to test the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determine the target model according to the test result; The fourth processing module 204 is used to determine abnormal data points based on the operation data corresponding to the dust removal system and the target model, and to calculate the probability of occurrence of at least two fault conditions according to the abnormal data points.

[0072] In some possible embodiments, preprocessing the first historical data to obtain the first data set and the second data set includes: The first processing module 201 is specifically used for: The first historical data is supplemented according to a preset formula to obtain an initial data set; Obtain the current timestamp, and generate a random number according to the current timestamp and a preset first algorithm; The initial data set is divided according to a preset ratio and a random number to obtain a first data set and a second data set.

[0073] In some possible embodiments, testing the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determining the target model according to the test result includes: The third processing module 203 is specifically used for: Input the historical data points in the second data set into the initial model to obtain a test label set; Calculating a first accuracy rate based on the label values ​​corresponding to the historical data points in the second data set and the test label set; The optimal hyperparameters are determined according to the first accuracy, and the target model is obtained according to the optimal hyperparameters.

[0074] In some possible embodiments, determining an optimal hyperparameter according to the first accuracy rate, and obtaining a target model according to the optimal hyperparameter includes: The third processing module 203 is specifically used for: When it is detected that the first accuracy rate is higher than or equal to the preset accuracy rate, the initial model is used as the target model; When it is detected that the first accuracy rate is lower than a preset accuracy rate, an optimal hyperparameter is calculated according to a preset second algorithm; The target model is obtained according to the optimal hyperparameters.

[0075] In some possible embodiments, obtaining a target model according to optimal hyperparameters includes: The third processing module 203 is specifically used for: The initial model is trained according to the optimal hyperparameters and the first data set to obtain a candidate model; Calculate a second accuracy rate of the candidate model based on the second data set; When it is detected that the second accuracy rate is higher than or equal to the preset accuracy rate, the candidate model is used as the target model; When it is detected that the second accuracy rate is lower than the preset accuracy rate, second historical data corresponding to the dust removal system is acquired; wherein the time corresponding to the first historical data is different from the time corresponding to the second historical data.

[0076] In some possible embodiments, based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the probability of occurrence of at least two fault conditions is statistically calculated based on the abnormal data points, including: The fourth processing module 204 is specifically used for: Performing structured processing on the operation data corresponding to the dust removal system to obtain at least two detection data points; Input all detection data points into the target model to obtain at least two label values ​​corresponding to the detection data points; When it is identified that the label value is within the preset value range, the detection data point corresponding to the label value is regarded as a normal data point; When it is identified that the label value is not within the preset value range, the detection data point corresponding to the label value is regarded as an abnormal data point; Compare the abnormal data point with the preset standard data point, and calculate the deviation value of at least two performance parameters; wherein the abnormal data point includes at least two performance parameters, and the preset standard data point includes at least two performance parameters; According to the deviation values ​​of at least two performance parameters, the occurrence probability of the fault situation corresponding to each performance parameter is determined.

[0077] In some possible embodiments, after determining the target model according to the test results, the method further includes: The fourth processing module 204 is specifically used for: Receive real-time data sent by the digital twin system and obtain fault detection results through the target model and real-time data; The fault detection result is sent to the digital twin system, so that the digital twin system generates a visualization interface corresponding to the fault detection result.

[0078] See also Figure 3 , Figure 3 A structural schematic diagram of another dust removal system fault detection device provided in an embodiment of the present application is shown.

[0079] like Figure 3 As shown, the dust removal system fault detection device 300 may include at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 and at least one communication bus 302 .

[0080] The communication bus 302 may be used to realize the connection and communication among the above-mentioned components.

[0081] The user interface 303 may include buttons, and the optional user interface may also include a standard wired interface or a wireless interface.

[0082] The network interface 304 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0083] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect the various parts of the entire dust removal system fault detection device 300, and executes various functions and processes data of the dust removal system fault detection device 300 by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of DSP, FPGA, and PLA. The processor 301 can integrate one or a combination of CPU, GPU, modem, etc. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0084] The memory 305 may include a RAM or a ROM. Optionally, the memory 305 includes a non-transitory computer-readable medium. The memory 305 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may optionally be at least one storage device located away from the aforementioned processor 301. As Figure 3 As shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a dust removal system fault detection application.

[0085] Specifically, the processor 301 may be used to call the dust removal system fault detection application stored in the memory 305, and specifically perform the following operations: Acquire first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each historical data point, and the second data set includes at least two historical data points and a label value corresponding to each historical data point, and the number of historical data points in the second data set is less than that in the first data set; Training a preset model based on all historical data points in the first data set and a label value corresponding to each historical data point to obtain an initial model; Testing the initial model based on all historical data points in the second data set and the label value corresponding to each historical data point, and determining the target model according to the test result; Based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the probability of occurrence of at least two fault conditions is statistically calculated based on the abnormal data points.

[0086] The present application also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0087] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0088] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0089] In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0090] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0092] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk and other media that can store program codes.

[0093] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by entering a program to instruct related hardware, and the program may be stored in a computer-readable memory, which may include a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0094] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A dust removal system fault detection method, characterized in that: include: Acquire first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each of the historical data points, and the second data set includes at least two historical data points and a label value corresponding to each of the historical data points, and the number of historical data points in the second data set is less than that in the first data set; Training a preset model based on all historical data points in the first data set and a label value corresponding to each of the historical data points to obtain an initial model; Testing the initial model based on all historical data points in the second data set and the label value corresponding to each of the historical data points, and determining the target model according to the test result; Based on the operation data corresponding to the dust removal system and the target model, abnormal data points are determined, and the occurrence probabilities of at least two fault conditions are statistically calculated based on the abnormal data points.

2. The method according to claim 1, characterized in that The preprocessing of the first historical data to obtain a first data set and a second data set includes: The first historical data is supplemented according to a preset formula to obtain an initial data set; Obtaining a current timestamp, and generating a random number according to the current timestamp and a preset first algorithm; The initial data set is divided according to a preset ratio and the random number to obtain a first data set and a second data set.

3. The method according to claim 1, characterized in that The step of testing the initial model based on all historical data points in the second data set and the label value corresponding to each of the historical data points, and determining the target model according to the test result includes: Inputting the historical data points in the second data set into the initial model to obtain a test label set; Calculating a first accuracy rate according to the label value corresponding to the historical data point in the second data set and the test label set; An optimal hyperparameter is determined according to the first accuracy, and a target model is obtained according to the optimal hyperparameter.

4. The method according to claim 3, characterized in that Determining an optimal hyperparameter according to the first accuracy rate, and obtaining a target model according to the optimal hyperparameter, includes: When it is detected that the first accuracy rate is higher than or equal to a preset accuracy rate, taking the initial model as a target model; When it is detected that the first accuracy rate is lower than the preset accuracy rate, calculating the optimal hyperparameter according to a preset second algorithm; A target model is obtained according to the optimal hyperparameters.

5. The method according to claim 4, characterized in that The step of obtaining a target model according to the optimal hyperparameters includes: Training the initial model according to the optimal hyperparameters and the first data set to obtain a candidate model; Calculating a second accuracy rate of the candidate model according to the second data set; When it is detected that the second accuracy rate is higher than or equal to the preset accuracy rate, taking the candidate model as the target model; When it is detected that the second accuracy rate is lower than the preset accuracy rate, second historical data corresponding to the dust removal system is acquired; wherein the time corresponding to the first historical data is different from the time corresponding to the second historical data.

6. The method according to claim 1, characterized in that The determining of abnormal data points based on the operation data corresponding to the dust removal system and the target model, and calculating the occurrence probabilities of at least two fault conditions according to the abnormal data points, includes: Performing structured processing on the operation data corresponding to the dust removal system to obtain at least two detection data points; Inputting all the detection data points into the target model to obtain at least two label values ​​corresponding to the detection data points; When it is identified that the tag value is within a preset value range, the detection data point corresponding to the tag value is regarded as a normal data point; When it is identified that the label value is not within the preset value range, the detection data point corresponding to the label value is taken as an abnormal data point; Compare the abnormal data point with a preset standard data point, and calculate the deviation value of at least two performance parameters; wherein the abnormal data point includes at least two of the performance parameters, and the preset standard data point includes at least two of the performance parameters; According to the deviation values ​​of at least two of the performance parameters, the occurrence probability of the fault situation corresponding to each of the performance parameters is determined.

7. The method according to claim 1, characterized in that After determining the target model according to the test results, the method further includes: Receiving real-time data sent by the digital twin system, and obtaining a fault detection result through the target model and the real-time data; The fault detection result is sent to the digital twin system so that the digital twin system generates a visualization interface corresponding to the fault detection result.

8. A dust removal system fault detection device, characterized in that: include: a first processing module, configured to obtain first historical data corresponding to the dust removal system, and preprocess the first historical data to obtain a first data set and a second data set; wherein the first data set includes at least two historical data points and a label value corresponding to each of the historical data points, and the second data set includes at least two historical data points and a label value corresponding to each of the historical data points, and the number of historical data points in the second data set is less than that in the first data set; A second processing module, configured to train a preset model based on all historical data points in the first data set and a label value corresponding to each of the historical data points to obtain an initial model; A third processing module, configured to test the initial model based on all historical data points in the second data set and a label value corresponding to each of the historical data points, and determine a target model according to a result of the test; The fourth processing module is used to determine abnormal data points based on the operating data corresponding to the dust removal system and the target model, and to calculate the probability of occurrence of at least two fault conditions according to the abnormal data points.

9. A dust removal system fault detection device, characterized in that: including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a computer or a processor, the computer or the processor executes the steps of the method according to any one of claims 1 to 7.

Images