Early warning method for equipment life, electronic equipment and storage medium
Through data acquisition and analysis of industrial equipment, and the use of presets and neural network models to predict equipment life, the inaccuracy problem of equipment life judgment in the existing technology is solved, and the efficiency of equipment maintenance and resource utilization are improved.
Patent Information
- Application Number
- CN202411201796.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2025-01-07
AI Technical Summary
The prior art cannot accurately detect the production information and working status of industrial equipment, resulting in the inability to effectively judge the service life of the equipment, resulting in high labor costs, low detection efficiency, and the inability to reasonably arrange equipment maintenance and maintenance.
By collecting equipment data and historical maintenance data of the target equipment, multiple health status indicators are determined, clustered analysis is performed, and multiple preset prediction models and neural network models are used for prediction, combining weights to determine the device life and reduce manual intervention.
It realizes accurate prediction of equipment life, reduces labor costs, improves the reliability and resource utilization of production plans, and optimizes equipment maintenance and maintenance arrangements.
Smart Images
Figure CN119272071A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of industrial equipment monitoring, and in particular to an early warning method, electronic equipment and storage medium for equipment life. Background Art
[0002] Industrial equipment refers to various types of equipment used in industrial production, and generally has complex mechanical equipment. Once an industrial equipment fails, it will take a lot of time and cost to repair it, and even serious accidents may occur. At present, most industrial equipment relies on engineers or experts to understand and analyze the equipment's operating status and historical data, and estimate the life of the equipment by considering the equipment's use conditions, maintenance conditions, and environmental factors. It is impossible to accurately detect the production information and working status of the equipment manually, resulting in the inability to judge the service life of the equipment. When detecting the production environment, it is impossible to analyze the impact of multiple components and multiple indicators on the environment on the equipment at the same time, resulting in excessive personnel investment and low-value detection equipment. Summary of the invention The purpose of the embodiments of the present application is to provide an early warning method for device life, an electronic device, and a storage medium.
[0003] In order to achieve the above-mentioned purpose, the first aspect of the present application provides an early warning method for equipment life, the early warning method comprising: Determine multiple health status indicators of the target device based on the device data and historical maintenance data of the target device; For each health status indicator, clustering the health status indicator data of the health status indicator within a preset historical time period to obtain a preset number of clustering results corresponding to the health status indicator; Sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain a first predicted equipment life; Sending all clustering results corresponding to each health status indicator to a second target prediction model to obtain a second predicted equipment life, wherein the second target prediction model is obtained by training a neural network model through a genetic algorithm according to a second preset rule; The predicted device life of the target device is determined by combining the first predicted device life and the first weight of the first target prediction model, and the second predicted device life and the second weight of the second target prediction model.
[0004] In an embodiment of the present application, the first target prediction model includes multiple trained preset prediction models, and the early warning method also includes: before sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain the first predicted equipment life, obtaining historical data of multiple sample devices, wherein the sample device is a device of the same type as the target device, and the historical data of each sample device includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample device, determining a sample data group corresponding to the sample device according to the historical data of the sample device, wherein the sample data group includes the corresponding historical health status indicator and the historical clustering results corresponding to each historical health status indicator; inputting all sample data groups into the preset prediction model to be trained to obtain the first historical predicted equipment life corresponding to each sample data group; determining the error between the first historical predicted equipment life corresponding to each sample data group and the corresponding historical actual life; adjusting the parameters of the preset prediction model until the proportion of sample data groups with errors less than a preset threshold in all sample data groups reaches a preset proportion, determining that the training of the preset prediction model to be trained is completed.
[0005] In an embodiment of the present application, the first target prediction model includes multiple trained preset prediction models, and all clustering results corresponding to each health status indicator are sent to the first target prediction model to obtain the first predicted equipment life, including: inputting all clustering results corresponding to each health status indicator into multiple trained preset prediction models respectively to obtain the intermediate predicted equipment life output by each trained preset prediction model; and performing weighted averaging on each intermediate predicted equipment life to determine the first predicted equipment life.
[0006] In the embodiment of the present application, the objective function of the preset prediction model is shown by expression (1): , expression (1); in, represents the loss function of the preset prediction model, represents the historical actual life span corresponding to the i-th sample data group, represents the first historical predicted equipment life corresponding to the i-th sample data group, where in represents the i-th sample data group in the sample data group set, Indicates the corresponding The weight of = ( ), The complexity function of the prediction model is not preset, where , k represents the number of executions, T represents the number of data splits, and is a constant coefficient, Represents weight.
[0007] In an embodiment of the present application, the early warning method also includes: sending all clustering results corresponding to each health status indicator to the second target prediction model to obtain the historical data of the sample equipment before obtaining the second predicted equipment life, wherein the sample equipment is the same type of equipment as the target equipment, and the historical data is the historical equipment data, historical maintenance data and historical actual life of the sample equipment; determining the sample data for training based on the historical data, wherein the sample data includes the historical health status indicators corresponding to the sample equipment and the historical clustering results corresponding to each historical health status indicator; training the neural network model through the sample data and the historical actual life to obtain a trained neural network model; tuning the parameters of the trained neural network model through a genetic algorithm based on the sample data and the historical actual life; and determining the neural network model after parameter tuning through the genetic algorithm as the second target prediction model.
[0008] In the embodiment of the present application, the neural network model is trained by using sample data and historical actual life to obtain a trained neural network model, including: inputting sample data into the neural network model to be trained so that the neural network model to be trained outputs a second historical predicted device life, and training the neural network model to be trained by using the second historical predicted device life and the historical actual life; in the process of training the neural network model, the model parameters of the neural network model are dynamically adjusted based on formula (2): , formula (2); in, The function represents the learning rate, where s represents the number of iterations. It represents the sum of square errors between the last prediction result and the current prediction result. The convergence function of the neural network model is shown in formula (3): , formula (3); Among them, x represents the preset bias parameter in the neural network model, and y represents the output value.
[0009] In an embodiment of the present application, parameter tuning of a trained neural network model through a genetic algorithm according to sample data and historical actual life includes: generating data to be trained through a genetic algorithm according to the sample data, so as to iteratively train the trained neural network model through the data to be trained; in the process of iteratively training the trained neural network model according to the data to be trained, when the error between the third historical predicted equipment life output by the trained neural network model in this iteration and the historical actual life and the error between the third historical predicted equipment life output by the trained neural network model in the previous iteration and the historical actual life are less than a preset value, the eliminated sample data are reversed and input into the genetic algorithm as new sample data to generate new data to be trained; the trained neural network model is trained with the new data to be trained until the number of iterations of the trained neural network model reaches a preset number.
[0010] In an embodiment of the present application, the early warning method also includes: before determining the predicted equipment life of the target equipment by combining the first predicted equipment life and the first weight of the first target prediction model, and the second predicted equipment life and the second weight of the second target prediction model, obtaining historical data of multiple sample equipment, wherein the sample equipment is equipment of the same type as the target equipment, and the historical data of each sample equipment includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample equipment, determining a test data group corresponding to the sample equipment according to the historical data of the sample equipment, wherein the test data group includes corresponding historical health status indicators and historical clustering results corresponding to each historical health status indicator; inputting all test data groups into the first target prediction model and the second target prediction model respectively to output the first test predicted life and the second test predicted life corresponding to each test data group; determining the first cumulative error sum between the first test predicted life and the historical actual life corresponding to all test data groups, and the second cumulative error sum between the second test predicted life and the historical actual life; determining the first weight of the first target prediction model and the second weight of the second target prediction model according to the first cumulative error sum and the second cumulative error sum.
[0011] A second aspect of the present application provides an electronic device comprising: a processor, and a memory communicatively connected to the processor; Memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the early warning method for equipment life as described in any one of the above embodiments.
[0012] A third aspect of the present application provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the early warning method for equipment life according to any one of the above-mentioned embodiments.
[0013] Through the above technical solution, by collecting historical data related to the life of the target equipment, determining the health status indicators of the target equipment and the clustering results of each health status indicator through the historical data, and obtaining the predicted equipment life corresponding to the target equipment through multiple target prediction models, and through equipment life warning, it is possible to reduce labor costs, better plan production plans, reasonably arrange equipment maintenance and care, optimize production processes, and improve production efficiency and resource utilization.
[0014] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings: Figure 1 A schematic diagram of a process flow of a warning method for equipment life according to an embodiment of the present application is schematically shown; Figure 2 A structural block diagram of an electronic device according to an embodiment of the present application is schematically shown; Figure 3 The internal structure diagram of a computer device according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION
[0016] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme 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. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0017] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), such directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.
[0018] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.
[0019] Figure 1 The following schematically shows a flow chart of a method for early warning of equipment life according to an embodiment of the present application. Figure 1 As shown, an embodiment of the present application provides an early warning method for equipment life, which may include the following steps.
[0020] Step 101, determining multiple health status indicators of the target device according to the device data and historical maintenance data of the target device; Step 102, for each health status indicator, clustering the health status indicator data of the health status indicator within a preset historical time period to obtain a preset number of clustering results corresponding to the health status indicator; Step 103, sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain a first predicted equipment life; Step 104, sending all clustering results corresponding to each health status indicator to a second target prediction model to obtain a second predicted equipment life, wherein the second target prediction model is obtained by training a neural network model through a genetic algorithm according to a second preset rule; Step 105 , determining the predicted device life of the target device by combining the first predicted device life and the first weight of the first target prediction model, and the second predicted device life and the second weight of the second target prediction model.
[0021] The processor can obtain the equipment data of the target device, such as the production date, equipment name, equipment number, equipment energy consumption, power consumption, temperature value, pressure value, vibration frequency, equipment operation time, etc. of the target device. The processor can also obtain the historical maintenance data of the target device, such as the number of maintenance times and maintenance frequency of the target device. The processor can collect the relevant data of the target device through the industrial Internet platform, and after collecting the data of the target device, the collected data can be preprocessed, such as cleaning the collected data to filter invalid data. The processor can determine multiple health status indicators of the target device based on the preprocessed equipment data and historical maintenance data of the target device. For example, the processor can use the principal component (PCA) analysis algorithm to analyze the acquired data of the target device, determine the contribution rate and cumulative contribution rate of each data through the principal component (PCA) analysis algorithm, and determine the data corresponding to the contribution rate greater than the preset contribution rate as the principal component data in the target device data, so that the processor can determine the determined principal component data as the health status indicator corresponding to the target device. For example, assuming that the data used by the processor for principal component (PCA) analysis include production date, device name, device number, device energy consumption, power consumption, temperature value, pressure value, vibration frequency, device operating time and maintenance frequency, the principal component (PCA) analysis algorithm is used to determine that the temperature value, pressure value, vibration frequency, device operating time and maintenance frequency are health status indicators of the target device.
[0022] After determining the health status indicator of the target device, the processor can obtain the health status indicator data of the health status indicator within a preset historical time period for each health status indicator. For example, assuming that the historical time period set by the processor is one month, the processor can obtain the data corresponding to each health status indicator in the past month. For each health status indicator, the processor can perform cluster analysis on the health status indicator of the health status indicator within the preset historical time period to obtain a preset number of clustering results corresponding to the health status indicator. For example, for each health status indicator, assuming that the processor uses a K-means clustering model to perform cluster analysis on the health status indicator data of the health status indicator within the preset historical time period, the processor can set a preset number of cluster clusters according to the user's input data. Assuming that the preset number is 5, the processor analyzes the health status indicator data within the preset historical time period through the K-means clustering model, thereby obtaining 5 clustering result data of the health status indicator within the historical time period.
[0023] The processor may send all clustering results corresponding to each health status indicator to the first target prediction model, and output the first predicted device life of the target device according to all clustering results corresponding to each health status indicator of the target device through the first target prediction model. The processor may train the neural network model through a genetic algorithm according to the second preset rule to obtain the second target prediction model. After obtaining the second target prediction model, the processor may send all clustering results corresponding to each health status indicator to the second target prediction model, so as to output the second predicted device life of the target device according to all clustering results corresponding to each health status indicator of the target device through the second target prediction model. The processor may determine the first weight corresponding to the first target prediction model and the second weight corresponding to the second target prediction model, and determine the predicted device life of the target device according to the first predicted device life output by the first prediction target model and the first weight of the first target prediction model, and the second weight of the second predicted device life output by the second target preset model. For example, the processor may take the sum of the product of the first predicted device life and the first weight and the product of the second predicted device life and the second weight as the final predicted device life of the target device.
[0024] In one embodiment, the first target prediction model includes multiple trained preset prediction models, and the early warning method also includes: before sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain the first predicted equipment life, obtaining historical data of multiple sample devices, wherein the sample device is a device of the same type as the target device, and the historical data of each sample device includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample device, determining a sample data group corresponding to the sample device according to the historical data of the sample device, wherein the sample data group includes the corresponding historical health status indicator and the historical clustering results corresponding to each historical health status indicator; inputting all sample data groups into the preset prediction model to be trained to obtain the first historical predicted equipment life corresponding to each sample data group; determining the error between the first historical predicted equipment life corresponding to each sample data group and the corresponding historical actual life; adjusting the parameters of the preset prediction model until the proportion of sample data groups with errors less than a preset threshold in all sample data groups reaches a preset proportion, determining that the training of the preset prediction model to be trained is completed.
[0025] In one embodiment, the objective function of the preset prediction model is expressed by expression (1): , expression (1); in, represents the loss function of the preset prediction model, represents the historical actual life span corresponding to the i-th sample data group, represents the first historical predicted equipment life corresponding to the i-th sample data group, where in represents the i-th sample data group in the sample data group set, Indicates the corresponding The weight of = ( ), The complexity function of the prediction model is not preset, where , k represents the number of executions, T represents the number of data splits, and is a constant coefficient, Represents weight.
[0026] The first target prediction model includes multiple trained preset prediction models. Before the processor sends all clustering results corresponding to each health status indicator to the first target prediction model to obtain the first predicted equipment life through the first target prediction model, the processor can train the preset prediction model included in the first target prediction model. The processor can obtain historical data of multiple sample devices and train the preset prediction model through the historical data to obtain the trained preset prediction model, wherein the sample device can be a device of the same type as the target device, and the historical data of each sample device can include the historical device data, historical maintenance data and historical actual life of the sample device. For each sample device, the processor can determine the sample data group corresponding to the sample device based on the historical data of the sample device, wherein the sample data group can include the historical health status indicator corresponding to the sample data and the historical clustering results corresponding to each historical health status indicator. After obtaining the sample data group corresponding to each sample device, the processor can input all the sample data groups into the preset prediction model to be trained to obtain the first historical predicted equipment life corresponding to each sample data group. The objective function of the preset prediction model is: ,in, represents the loss function of the preset prediction model, represents the historical actual life span corresponding to the i-th sample data group, represents the first historical predicted equipment life corresponding to the i-th sample data group, where in represents the i-th sample data group in the sample data group set, Indicates the corresponding The weight of = ( ), The complexity function of the prediction model is not preset, where , k represents the number of executions, T represents the number of data splits, and is a constant coefficient, The processor can determine the error between the first historical predicted device life corresponding to each sample data group and the historical actual life of the sample device corresponding to the sample data group, and the processor can continuously adjust the parameters of the preset prediction model until the proportion of sample data groups with errors less than a preset threshold set by the processor in all sample data groups reaches a preset proportion, at which time the processor can determine that the training of the preset prediction model to be trained is completed.
[0027] In one embodiment, the first target prediction model includes multiple trained preset prediction models, and all clustering results corresponding to each health status indicator are sent to the first target prediction model to obtain the first predicted equipment life, including: inputting all clustering results corresponding to each health status indicator into multiple trained preset prediction models respectively to obtain the intermediate predicted equipment life output by each trained preset prediction model; and taking a weighted average of each intermediate predicted equipment life to determine the first predicted equipment life.
[0028] The first target prediction model includes multiple trained preset prediction models. After determining all clustering results corresponding to each health status indicator of the target device, the processor can input all clustering results corresponding to each health status indicator of the target device into multiple trained preset prediction models respectively, so as to output the intermediate predicted device life through the trained preset prediction model. After obtaining the intermediate predicted device life output by each trained preset prediction model, the processor can perform weighted average on each intermediate predicted device life to determine the first predicted device life. For example, the processor can determine the weight of each trained preset prediction model according to the prediction accuracy of each trained preset prediction model, or the processor can determine the weight of each trained preset prediction model according to the data input by the user. After each trained preset prediction model outputs the intermediate predicted device life according to all clustering results corresponding to each health status indicator of the target device, the processor can perform weighted average in combination with the weights, thereby finally determining the first predicted life of the first target prediction model.
[0029] In one embodiment, all clustering results corresponding to each health status indicator are sent to the second target prediction model to obtain the historical data of the sample equipment before obtaining the second predicted equipment life, wherein the sample equipment is the same type of equipment as the target equipment, and the historical data is the historical equipment data, historical maintenance data and historical actual life of the sample equipment; the sample data for training is determined based on the historical data, wherein the sample data includes the historical health status indicators corresponding to the sample equipment and the historical clustering results corresponding to each historical health status indicator; the neural network model is trained through the sample data and the historical actual life to obtain a trained neural network model; the parameters of the trained neural network model are tuned through a genetic algorithm based on the sample data and the historical actual life; and the neural network model after parameter tuning through the genetic algorithm is determined as the second target prediction model.
[0030] Before sending all clustering results corresponding to each health status indicator to the second target prediction model to obtain the second predicted equipment life, the processor will obtain the historical data of the sample equipment, and train the neural network model through the genetic algorithm according to the second preset rule through the historical data to obtain the second target prediction model. Among them, the sample equipment can be a device of the same type as the target equipment, and the historical data of the sample equipment can be the historical equipment data, historical maintenance data and historical actual life of the sample equipment. The processor can determine the sample data for training based on the historical data, wherein the sample data for training can be the historical health status indicators of the sample equipment determined by the processor based on the historical data and the historical clustering results corresponding to each historical health status indicator. The processor can train the neural network model through the sample data and the historical actual life of the sample equipment to obtain a trained neural network model. For example, assuming that the neural network model used for training is a BP neural network model, the processor can input the sample data into the BP neural network to output the predicted life through the BP neural network, and adjust the parameters of the BP neural network model according to the predicted life output by the BP neural network model and the historical actual life until the accuracy of the predicted life output by the BP neural network model meets the preset requirements, and then determine that the BP neural network model training is completed. After the processor has determined the trained neural network model, it can use the genetic algorithm to optimize the parameters of the trained neural network model again according to the sample data and the historical actual life span, so as to obtain a neural network model with more accurate prediction accuracy. The processor can determine the neural network model after the parameter tuning through the genetic algorithm as the second target prediction model.
[0031] In one embodiment, training a neural network model by using sample data and historical actual life to obtain a trained neural network model includes: inputting sample data into the neural network model to be trained so that the neural network model to be trained outputs a second historical predicted device life, and training the neural network model to be trained by using the second historical predicted device life and the historical actual life; in the process of training the neural network model, the model parameters of the neural network model are dynamically adjusted based on formula (2): , formula (2); in, The function represents the learning rate, where s represents the number of iterations. It represents the sum of square errors between the last prediction result and the current prediction result. The convergence function of the neural network model is shown in formula (3): , formula (3); Among them, x represents the preset bias parameter in the neural network model, and y represents the output value.
[0032] When the processor trains the neural network model through the sample data and the historical actual life of the sample device to obtain the trained neural network model, the processor can output the sample data to the neural network model to be trained. During the training process of the neural network model, the model parameters in the neural network model are dynamically adjusted based on formula (2). Formula (2) is as follows: , formula (2); in, The function represents the learning rate, where s represents the number of iterations, = It represents the sum of square errors between the last prediction result and the current prediction result. The convergence formula of the neural network model is , where x represents the preset bias parameter in the neural network model, and y represents the output value. The convergence function in the neural network is a periodic function, and as x increases, the amplitude decays exponentially. The neural network model in the prior art is prone to fall into local extreme points and converges slowly. The convergence function of the neural network model in this embodiment is as follows: As shown, The exponential decay characteristic of helps to reduce the risk of falling into local extreme values, thereby improving the training effect of the neural network.
[0033] In one embodiment, parameter tuning of a trained neural network model using a genetic algorithm according to sample data and historical actual lifespan includes: generating data to be trained using a genetic algorithm according to sample data, so as to iteratively train the trained neural network model using the data to be trained; in the process of iteratively training the trained neural network model according to the data to be trained, when the difference between the error between the third historical predicted equipment lifespan output by the trained neural network model in this iteration and the historical actual lifespan and the error between the third historical predicted equipment lifespan output by the trained neural network model in the previous iteration and the historical actual lifespan is less than a preset value, the eliminated sample data is reversed and input into the genetic algorithm as new sample data to generate new data to be trained; the trained neural network model is trained using the new data to be trained until the number of iterations of the trained neural network model reaches a preset number.
[0034] After the processor completes the training of the neural network model through sample data and historical actual life, it can generate data to be trained through a genetic algorithm based on the sample data, and iteratively train the trained neural network model again through the data to be trained. In the process of iteratively training the trained neural network model based on the data to be trained, when the difference between the error between the third historical predicted equipment life output by the trained neural network model in this iteration and the historical actual life and the error between the third historical predicted equipment life output by the trained neural network model in the previous iteration and the historical actual life is less than a preset value, the eliminated sample data is reversed and input into the genetic algorithm as new sample data to generate new data to be trained, and the trained neural network model is trained with the new data to be trained until the number of iterations of the trained neural network model reaches a preset number. For example, assuming that the neural network model is a BP neural network model, the processor performs parameter tuning on the trained neural network model through a genetic algorithm based on sample data and historical actual life, such as Figure 2As shown, step 1, according to each historical health status indicator in the sample data, determine the data interval corresponding to each historical health status indicator, convert this interval data into the corresponding 01 code, so as to generate the population corresponding to the sample data; step 2, take the inverse of the sum of squared errors as the fitness function, if the error is smaller, the fitness is greater, and the error is larger, the fitness is smaller, so as to screen and pass on the individuals with large fitness to the next generation; step 3 uses the crossover, mutation and genetic operations of the genetic algorithm to evolve the current population and generate the next generation population; step 4, repeat steps 2 and 3 until the specified accuracy or evolutionary algebra is reached; step 5 input the optimized data as the training data of the trained BP neural network into the trained BP neural network, obtain the prediction result (i.e., the third historical predicted equipment life of the sample equipment) through the trained BP neural network, judge whether the error between the prediction result and the historical actual life meets the preset requirements, if the error does not meet the requirements, it is fed back to the trained BP neural network, and the parameters in the trained BP neural network are tuned, and the error is fed back to the genetic algorithm, and the fitness function of the genetic algorithm iteratively selects the population according to the error size. During the iteration process of the genetic algorithm, when the difference between the error between the third historical predicted equipment life output by the BP neural network model trained in this iteration based on the training data provided by the genetic algorithm and the historical actual life, and the error between the third historical predicted equipment life output by the BP neural network model trained in the previous iteration based on the training data provided by the genetic algorithm and the historical actual life is less than a preset value, that is, the difference between the error of this iteration and the error of the previous iteration is less than the preset value, the genetic algorithm can re-extract the eliminated sample data, reverse the eliminated sample data as new sample data, and generate new data to be trained through the genetic algorithm until the number of iterations of the trained neural network model reaches the preset number.
[0035] In one embodiment, the early warning method also includes: before determining the predicted equipment life of the target equipment by combining the first predicted equipment life and the first weight of the first target prediction model, and the second predicted equipment life and the second weight of the second target prediction model, obtaining historical data of multiple sample equipment, wherein the sample equipment is the same type of equipment as the target equipment, and the historical data of each sample equipment includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample equipment, determining a test data group corresponding to the sample equipment according to the historical data of the sample equipment, wherein the test data group includes corresponding historical health status indicators and historical clustering results corresponding to each historical health status indicator; inputting all test data groups into the first target prediction model and the second target prediction model respectively to output the first test predicted life and the second test predicted life corresponding to each test data group; determining the first cumulative error sum between the first test predicted life and the historical actual life corresponding to all test data groups, and the second cumulative error sum between the second test predicted life and the historical actual life; determining the first weight of the first target prediction model and the second weight of the second target prediction model according to the first cumulative error sum and the second cumulative error sum.
[0036] Before determining the predicted device life of the target device according to the first predicted life and the first weight of the first target prediction model, the second predicted device life and the second weight of the second target prediction model, the processor may confirm the first weight and the second weight of the first target prediction model and the second target prediction model respectively, and the processor may obtain the historical data of multiple sample devices, wherein the sample device is a device of the same type as the target device, and the historical data of each sample device may include the historical device data, historical maintenance data and historical actual life of the sample device. For each sample device, the processor may determine the test data group corresponding to each sample device according to the historical data of each sample device, wherein the test data group includes the historical health status indicator corresponding to each sample device and the historical clustering result corresponding to each historical health status indicator. The processor may input all the test data groups into the first target prediction model and the second target prediction model respectively, and output the first test predicted life and the second test predicted life corresponding to each test data group through the first target prediction model and the second target prediction model. The processor may determine the first error between the first test predicted life corresponding to each test data group and the historical actual life, and the second error between the second test predicted life corresponding to each test data group and the historical actual life. After determining the first error and the second error corresponding to each test data group, for all test data groups, the processor can determine the first error cumulative sum of the first error corresponding to each test data group, and the second error cumulative sum of the second error corresponding to each test data group. The processor can determine the weights of the first target prediction model and the second target prediction model based on the first error cumulative sum corresponding to the first target prediction model and the second error cumulative sum corresponding to the second target prediction model. For example, the processor can determine the sum of the first error cumulative sum and the second error cumulative sum, determine the ratio of the first error cumulative sum of the first target prediction model to the sum value as the second weight of the second target prediction model, and determine the ratio of the second error cumulative sum of the second target prediction model to the sum value as the first weight of the first target prediction model. After the processor sends all clustering results corresponding to each health status indicator of the target device to the first target prediction model and the second target prediction model, and obtains the first predicted device life and the second predicted device life, the processor can determine the predicted device life of the target device in combination with the first weight of the first target prediction model and the second weight of the second target prediction model.
[0037] In one embodiment, Figure 2 As shown, an electronic device 200 is provided, comprising: a processor 201, and a memory 202 communicatively connected to the processor 201; the memory 202 stores computer-executable instructions; the processor 201 executes the computer-executable instructions stored in the memory 202 to implement a method for identifying a mine slope deformation area as described above in any one of the above.
[0038] Through the above scheme, by collecting historical data related to the life of the target equipment, determining the health status indicators of the target equipment and the clustering results of each health status indicator through the historical data, and obtaining the predicted equipment life corresponding to the target equipment through multiple target prediction models, and through equipment life warning, it is possible to reduce labor costs, better plan production plans, reasonably arrange equipment maintenance and care, optimize production processes, and improve production efficiency and resource utilization.
[0039] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown in the figure) and a database (not shown in the figure) connected through a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02 and a database (not shown in the figure). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A04. The database of the computer device is used to store relevant data obtained in the process of judging the deformation area. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program B02 is executed by the processor A01, a method for identifying the deformation area of a mine slope is implemented.
[0040] Figure 1 FIG. 1 is a flow chart of an early warning method for equipment life in one embodiment. It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0041] An embodiment of the present application provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: determining multiple health status indicators of the target device based on the device data and historical maintenance data of the target device; for each health status indicator, clustering the health status indicator data of the health status indicator within a preset historical time period to obtain a preset number of clustering results corresponding to the health status indicator; sending all clustering results corresponding to each health status indicator to a first target prediction model to obtain a first predicted device life; sending all clustering results corresponding to each health status indicator to a second target prediction model to obtain a second predicted device life, wherein the second target prediction model is a second target prediction model obtained by training a neural network model through a genetic algorithm according to a second preset rule; and determining the predicted device life of the target device by combining the first predicted device life and the first weight of the first target prediction model, as well as the second predicted device life and the second weight of the second target prediction model.
[0042] In one embodiment, the first target prediction model includes multiple trained preset prediction models, and the early warning method also includes: before sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain the first predicted equipment life, obtaining historical data of multiple sample devices, wherein the sample device is a device of the same type as the target device, and the historical data of each sample device includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample device, determining a sample data group corresponding to the sample device according to the historical data of the sample device, wherein the sample data group includes the corresponding historical health status indicator and the historical clustering results corresponding to each historical health status indicator; inputting all sample data groups into the preset prediction model to be trained to obtain the first historical predicted equipment life corresponding to each sample data group; determining the error between the first historical predicted equipment life corresponding to each sample data group and the corresponding historical actual life; adjusting the parameters of the preset prediction model until the proportion of sample data groups with errors less than a preset threshold in all sample data groups reaches a preset proportion, determining that the training of the preset prediction model to be trained is completed.
[0043] In one embodiment, the first target prediction model includes multiple trained preset prediction models, and all clustering results corresponding to each health status indicator are sent to the first target prediction model to obtain the first predicted equipment life, including: inputting all clustering results corresponding to each health status indicator into multiple trained preset prediction models respectively to obtain the intermediate predicted equipment life output by each trained preset prediction model; and taking a weighted average of each intermediate predicted equipment life to determine the first predicted equipment life.
[0044] In one embodiment, the objective function of the preset prediction model is expressed by expression (1): , expression (1); in, represents the loss function of the preset prediction model, represents the historical actual life span corresponding to the i-th sample data group, represents the first historical predicted equipment life corresponding to the i-th sample data group, where in represents the i-th sample data group in the sample data group set, Indicates the corresponding The weight of = ( ), The complexity function of the prediction model is not preset, where , k represents the number of executions, T represents the number of data splits, and is a constant coefficient, Represents weight.
[0045] In one embodiment, the early warning method also includes: sending all clustering results corresponding to each health status indicator to the second target prediction model to obtain the historical data of the sample equipment before obtaining the second predicted equipment life, wherein the sample equipment is the same type of equipment as the target equipment, and the historical data is the historical equipment data, historical maintenance data and historical actual life of the sample equipment; determining the sample data for training based on the historical data, wherein the sample data includes the historical health status indicators corresponding to the sample equipment and the historical clustering results corresponding to each historical health status indicator; training the neural network model through the sample data and the historical actual life to obtain a trained neural network model; tuning the parameters of the trained neural network model through a genetic algorithm based on the sample data and the historical actual life; and determining the neural network model after parameter tuning through the genetic algorithm as the second target prediction model.
[0046] In one embodiment, training a neural network model by using sample data and historical actual life to obtain a trained neural network model includes: inputting sample data into the neural network model to be trained so that the neural network model to be trained outputs a second historical predicted device life, and training the neural network model to be trained by using the second historical predicted device life and the historical actual life; in the process of training the neural network model, the model parameters of the neural network model are dynamically adjusted based on formula (2): , formula (2); in, The function represents the learning rate, where s represents the number of iterations. It represents the sum of square errors between the last prediction result and the current prediction result. The convergence function of the neural network model is shown in formula (3): , formula (3); Among them, x represents the preset bias parameter in the neural network model, and y represents the output value.
[0047] In one embodiment, parameter tuning of a trained neural network model using a genetic algorithm according to sample data and historical actual lifespan includes: generating data to be trained using a genetic algorithm according to sample data, so as to iteratively train the trained neural network model using the data to be trained; in the process of iteratively training the trained neural network model according to the data to be trained, when the difference between the error between the third historical predicted equipment lifespan output by the trained neural network model in this iteration and the historical actual lifespan and the error between the third historical predicted equipment lifespan output by the trained neural network model in the previous iteration and the historical actual lifespan is less than a preset value, the eliminated sample data is reversed and input into the genetic algorithm as new sample data to generate new data to be trained; the trained neural network model is trained using the new data to be trained until the number of iterations of the trained neural network model reaches a preset number.
[0048] In one embodiment, the early warning method also includes: before determining the predicted equipment life of the target equipment by combining the first predicted equipment life and the first weight of the first target prediction model, and the second predicted equipment life and the second weight of the second target prediction model, obtaining historical data of multiple sample equipment, wherein the sample equipment is the same type of equipment as the target equipment, and the historical data of each sample equipment includes corresponding historical equipment data, historical maintenance data and historical actual life; for each sample equipment, determining a test data group corresponding to the sample equipment according to the historical data of the sample equipment, wherein the test data group includes corresponding historical health status indicators and historical clustering results corresponding to each historical health status indicator; inputting all test data groups into the first target prediction model and the second target prediction model respectively to output the first test predicted life and the second test predicted life corresponding to each test data group; determining the first cumulative error sum between the first test predicted life and the historical actual life corresponding to all test data groups, and the second cumulative error sum between the second test predicted life and the historical actual life; determining the first weight of the first target prediction model and the second weight of the second target prediction model according to the first cumulative error sum and the second cumulative error sum.
[0049] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0050] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0051] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0052] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0053] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0054] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0055] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0056] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0057] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A warning method for equipment life, characterized in that: The early warning method comprises: Determine multiple health status indicators of the target device according to the device data and historical maintenance data of the target device; For each health status indicator, clustering the health status indicator data of the health status indicator within a preset historical time period to obtain a preset number of clustering results corresponding to the health status indicator; Sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain a first predicted equipment life; Sending all clustering results corresponding to each health status indicator to a second target prediction model to obtain a second predicted equipment life, wherein the second target prediction model is obtained by training a neural network model through a genetic algorithm according to a second preset rule; The predicted device lifetime of the target device is determined by combining the first predicted device lifetime and the first weight of the first target prediction model, and the second predicted device lifetime and the second weight of the second target prediction model.
2. The early warning method for equipment life according to claim 1, characterized in that: The first target prediction model includes a plurality of preset prediction models that have been trained, and the early warning method further includes: Before sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain the first predicted equipment life, historical data of multiple sample devices are obtained, wherein the sample devices are devices of the same type as the target device, and the historical data of each sample device includes corresponding historical device data, historical maintenance data, and historical actual life; For each sample device, determine a sample data group corresponding to the sample device according to the historical data of the sample device, wherein the sample data group includes a corresponding historical health status indicator and a historical clustering result corresponding to each historical health status indicator; Inputting all sample data groups into a preset prediction model to be trained to obtain a first historical prediction equipment life corresponding to each sample data group; Determine the error between a first historical predicted equipment life corresponding to each sample data group and a corresponding historical actual life; The parameters of the preset prediction model are adjusted until the proportion of sample data groups with errors less than a preset threshold in all sample data groups reaches a preset proportion, and it is determined that the training of the preset prediction model to be trained is completed.
3. The early warning method for equipment life according to claim 1, characterized in that: The first target prediction model includes a plurality of preset prediction models that have been trained, and sending all clustering results corresponding to each health status indicator to the first target prediction model to obtain a first predicted equipment life includes: Inputting all clustering results corresponding to each health status indicator into a plurality of trained preset prediction models respectively, so as to obtain an intermediate predicted equipment life output by each trained preset prediction model; A weighted average is taken of each intermediate predicted equipment life to determine the first predicted equipment life.
4. The early warning method for equipment life according to claim 2 or 3, characterized in that: The objective function of the preset prediction model is shown by expression (1): , expression (1); in, represents the loss function of the preset prediction model, represents the historical actual life span corresponding to the i-th sample data group, represents the first historical predicted equipment life corresponding to the i-th sample data group, where in represents the i-th sample data group in the sample data group set, Indicates the corresponding The weight of = ( ), The complexity function of the preset prediction model is not described, wherein, , k represents the number of executions, T represents the number of data splits, and is a constant coefficient, Represents weight.
5. The early warning method for equipment life according to claim 1, characterized in that: The early warning method also includes: Before sending all clustering results corresponding to each health status indicator to the second target prediction model to obtain the second predicted equipment life, historical data of sample equipment is obtained, wherein the sample equipment is equipment of the same type as the target equipment, and the historical data is historical equipment data, historical maintenance data, and historical actual life of the sample equipment; Determining sample data for training according to the historical data, wherein the sample data includes historical health status indicators corresponding to the sample devices and historical clustering results corresponding to each historical health status indicator; Training the neural network model using the sample data and the historical actual lifespan to obtain a trained neural network model; Optimizing parameters of the trained neural network model using a genetic algorithm according to the sample data and the historical actual lifespan; The neural network model after parameter tuning through genetic algorithm is determined as the second target prediction model.
6. The early warning method for equipment life according to claim 5, characterized in that: The neural network model is trained by using the sample data and the historical actual lifespan to obtain a trained neural network model, including: Inputting the sample data into a neural network model to be trained so that the neural network model to be trained outputs a second historical predicted equipment life, and training the neural network model to be trained by the second historical predicted equipment life and the historical actual life; In the process of training the neural network model, the model parameters of the neural network model are dynamically adjusted based on formula (2): , Formula (2); in, The function represents the learning rate, where s represents the number of iterations. Represents the sum of squared errors between the last prediction result and the current prediction result; The convergence function of the neural network model is shown in formula (3): , Formula (3); Wherein, x represents the preset bias parameter in the neural network model, and y represents the output value.
7. The early warning method for equipment life according to claim 5, characterized in that: The step of optimizing the parameters of the trained neural network model by using a genetic algorithm according to the sample data and the historical actual lifespan includes: Generate training data through the genetic algorithm according to the sample data, so as to iteratively train the trained neural network model through the training data; In the process of iteratively training the trained neural network model according to the data to be trained, when the difference between the error between the third historical predicted equipment life output by the trained neural network model in this iteration and the historical actual life and the error between the third historical predicted equipment life output by the trained neural network model in the previous iteration and the historical actual life is less than a preset value, the eliminated sample data are reversed and input into the genetic algorithm as new sample data to generate new data to be trained; The trained neural network model is trained using the new data to be trained until the number of iterations of the trained neural network model reaches a preset number.
8. The early warning method for equipment life according to claim 1, characterized in that: The early warning method also includes: Before determining the predicted device life of the target device by combining the first predicted device life and the first weight of the first target prediction model, and the second predicted device life and the second weight of the second target prediction model, acquiring historical data of a plurality of sample devices, wherein the sample devices are devices of the same type as the target device, and the historical data of each sample device includes corresponding historical device data, historical maintenance data, and historical actual life; For each sample device, determine a test data group corresponding to the sample device according to the historical data of the sample device, wherein the test data group includes the corresponding historical health status indicator and the historical clustering result corresponding to each historical health status indicator; Inputting all test data groups into the first target prediction model and the second target prediction model respectively to output a first test predicted life and a second test predicted life corresponding to each test data group; Determine a first cumulative error sum between a first test predicted life and a historical actual life corresponding to all test data groups, and a second cumulative error sum between a second test predicted life and a historical actual life; The first weight of the first target prediction model and the second weight of the second target prediction model are determined according to the first cumulative error sum and the second cumulative error sum.
9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the early warning method for equipment life according to any one of claims 1 to 8.
10. A machine-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to execute the early warning method for device life according to any one of claims 1 to 8.