Electrical equipment fault prediction method and system

Abstract

The invention relates to an electrical equipment fault prediction method and system, and the method improves the fault prediction precision and timeliness through a dynamic optimization model, and comprises the steps: firstly obtaining an equipment identifier, real-time data information and a historical prediction error of target electrical equipment, determining a target equipment type according to the equipment identifier, and carrying out the fault prediction of the target electrical equipment; and calling a corresponding target optimization module to dynamically optimize a preset initial fusion model, and outputting a target prediction fusion model. The method for determining the initial fusion model comprises the steps of classifying electrical equipment, collecting historical data and performing feature extraction, selecting a matched basic prediction model for training and calculating an error index, and finally selecting an optimal combination mode, is applied to actual fault prediction, and can provide powerful support for maintenance and management of the equipment. The invention further provides an electrical equipment fault prediction system, and a comprehensive, accurate and real-time solution is provided for fault prediction of the electrical equipment through efficient data management and model optimization strategies.

Description

Technical Field

[0001] The present application relates to the technical field of fault prediction, and in particular to a method and system for predicting faults of electrical equipment. Background Art

[0002] There are many kinds of electrical equipment in robot manipulators. Different types of electrical equipment, such as sensors, motors, and power systems, have significantly different fault characteristics and prediction requirements due to differences in their working principles, operating environments, and usage scenarios. Even for the same type of electrical equipment, such as temperature sensors and position sensors, there are significant differences in their failure modes, data characteristics, and requirements for prediction accuracy. This diversity and complexity makes fault prediction of electrical equipment a challenging technical problem.

[0003] Existing technologies usually use a separate monitoring method to predict faults of various types of electrical equipment, that is, to build a prediction model for each device separately. Although this method can make customized predictions based on the fault characteristics of specific equipment, it also leads to a sharp increase in the number of models, which not only increases the complexity and maintenance costs of the system, but also makes it difficult to achieve coordinated optimization between devices and comprehensive evaluation of the overall health status. Summary of the invention

[0004] The inventors of the present application have found that most of the existing fault prediction models rely on historical data for training and prediction, and the data source is usually relatively single, resulting in poor timeliness and accuracy of the model. Real-time data can reflect changes in the current operating state of the equipment, which is of great significance for capturing sudden failures and dynamically adjusting the prediction model. However, in the process of model updating and optimization, the existing methods usually adopt a fixed model structure or a static weight distribution strategy, lacking the ability to dynamically respond to real-time data, thereby limiting the prediction accuracy and adaptability of the model. The fixed model design is difficult to adapt to the dynamic changes in the operating state of the robot manipulator, especially in the high-load, high-precision operation scenarios of the manipulator. The operating conditions and workload of the equipment may change at any time, further reducing the reliability and accuracy of the prediction model, and because the association between the data is ignored or the association between the data is unclear, the application of the model is limited in terms of universality.

[0005] According to a first aspect of the present invention, there is provided a method for predicting faults of electrical equipment for a robot manipulator, comprising the following steps: Obtaining the equipment identification, real-time data information X and historical prediction error H(X) of the target electrical equipment; Determine the target device category according to the device identifier, call the target optimization module corresponding to the target device category, and make the target optimization module perform the preset initial fusion model F according to the following formula based on the real-time data information X and the historical prediction error H(X): base (X) Perform dynamic optimization to output the target prediction fusion model F opt (X), F opt (X)=F base (X)+(γ×e -β|H(X)| +λ s )×W s ×X; Among them, γ and β are coefficients, λ s represents the adaptive correction term, W s Represents the weight parameter of the target electrical equipment. One target optimization module corresponds to one equipment category. Based on the device identification and the real-time data information, the target prediction fusion model outputs a prediction result of the target electrical device.

[0006] Optionally, the adaptive correction term λ s The method for obtaining includes the following steps: Obtain the historical prediction errors and preset basic correction items λ of all electrical equipment in the target equipment category s (0) ; Calculate the average error H of all electrical devices in the target device category according to the following formula: s , ; Where N is the number of electrical devices in the target device category, and H i represents the historical prediction error of the i-th electrical equipment in the target equipment category; Define the impact factor α of the target device category s and the influencing factors of the target electrical equipment , α s =e -k|Hs| , , k and p both represent adjustment coefficients; According to the formula Calculate the adaptive correction term λ s .

[0007] Optionally, the adjustment coefficient k is obtained according to the following formula: ; ; The adjustment coefficient p is obtained according to the following formula: ; ; Among them, k0 and p0 are the initial adjustment coefficients, t k and t p These are adjustment parameters.

[0008] Optionally, the method for obtaining the historical prediction error H(X) of the target electrical equipment comprises the following steps: Obtain the prediction error H of the target electrical equipment at time t-1 t-1 (X), prediction error H at time t-2 t-2 (X) and the mean prediction error H of the M moments before time t-1 m (X); Based on the weighted moving average method, the first historical forecast error H is calculated according to the following formula: EWMA,t-1 (X), ; Wherein, q represents the smoothing factor, and 0<q<1; The second historical prediction error T is calculated according to the following formula t (X), T t-1 (X)=r×(H t-1 (X)-H t-2 (X)); Where r represents the trend adjustment coefficient, and 0.5<r<1.5; According to the formula H(X)=θ1×H EWMA,t-1 (X)+θ2×T t-1 (X) Calculate the historical prediction error H(X), where θ1 and θ2 are both constants, and their sum is equal to 1.

[0009] Optionally, the initial fusion model F base The method for determining (X) comprises the following steps: Classify the electrical equipment of the robot manipulator into categories, thereby obtaining multiple equipment categories; For each device category, historical data is collected and feature extraction is performed; Selecting at least two basic prediction models matching the feature type from a model database according to the extracted feature type; The selected basic prediction model is trained using the historical data of the equipment category, and the error index is calculated to select the optimal combination mode of the equipment category as the initial fusion model F of the equipment category. base (X).

[0010] Optionally, the selected basic prediction model is trained using the historical data of the device category, and the error index is calculated to select the optimal combination mode of the device category as the initial fusion model F of the device category. base (X), comprising the following steps: The historical data of each device category is segmented according to time periods, and the selected basic model is trained using the historical data of each time period; The error of each basic model in each time period is calculated according to the following formula: ; in, represents the i-th basic model f i In time period T i The error within Indicates time period T i The number of samples within Indicates the actual value, represents the predicted value; For each time period, the basic model with the smallest error is selected as the optimal model in that time period; The frequency of each basic model being selected as the optimal model in all time periods is counted according to the following formula , ; in, represents the jth basic model, For time period T i The optimal model within represents the indicator function; Select at least two basic models with the highest frequency to be fused as the target basic model. If the frequencies of multiple basic models are the same, select at least two basic models with the smallest error as the target basic model. Perform weighted fusion on the selected target basic models to obtain the initial fusion model F base (X).

[0011] Optionally, the selected target basic models are weightedly fused to obtain an initial fusion model F base (X), comprising the following steps: For each target base model, calculate its average error over all time periods; According to the target basic model and the error, the i-th basic model is calculated according to the following formula: Weight , ; Among them, S represents the set of target basic models, ε represents the adjustment coefficient, Represents the i-th basic model The average error of Get the initial fusion model F base The formula for (X) is ; in, , k represents the number of selected target base models.

[0012] Optionally, the initial fusion model F base After the formula of (X) is determined, the following steps are also included: At every preset time interval, calculate the average error of each basic model on the latest historical data; After the average error of the latest historical data exceeds the threshold error, the model update is triggered to reselect the optimal basic model or adjust the weight of the target basic model to obtain an updated initial fusion model.

[0013] Optionally, the initial fusion model F corresponding to different device categories base The weight parameter W in (X) s Satisfies the following formula: ; Among them, W s (0) Indicates the initial weight parameter of the corresponding device category.

[0014] According to a second aspect of the present invention, there is provided an electrical equipment fault prediction system, comprising at least one processor, wherein the at least one processor is used to execute computer programs or instructions to perform operations corresponding to the aforementioned electrical equipment fault prediction method.

[0015] According to the solution of the present application, by obtaining the device identification, real-time data information X and historical prediction error H(X) of the target electrical equipment, and calling the target optimization module corresponding to the target equipment category based on the device identification, the initial fusion model F is optimized by using the real-time data information X and the historical prediction error H(X). base (X) Perform dynamic optimization and output the target prediction fusion model F opt (X), thereby solving the problems in the prior art that the fault prediction model is overly dependent on historical data, underutilized real-time data, and fixed models lead to poor prediction accuracy and timeliness. In addition, for different equipment categories (such as sensors, motors, power supply systems, etc.), the embodiment of the present invention extracts a prediction fusion model F that can be applied to key electrical equipment of the robot manipulator opt The general formula of (X) is obtained by dynamically adjusting the adaptive correction term λ s and weight parameter W s, which can adapt to the fault characteristics and prediction requirements of different equipment categories, thereby significantly improving the prediction accuracy and reliability while maintaining model consistency. In addition, the unified prediction fusion model F opt (X) It not only avoids building and maintaining multiple models for each device category, thus improving the versatility and scalability of the model, but also enables continuous adjustment of the model and the correction of the model with the assistance of real-time data, thereby continuously improving its accuracy, while also reducing the consumption of computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic flow chart of a method for predicting electrical equipment faults according to an embodiment of the present invention; Figure 2 is a schematic flow chart of a method for determining an initial fusion model in a fault prediction method of the present invention; Figure 3 It is a schematic flow chart of a specific method for obtaining the initial fusion model in the fault prediction method of the present invention. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some structures related to the present application are shown in the accompanying drawings, rather than all structures. 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.

[0018] The terms "including" and "having" and any variations thereof in this application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these methods, products or devices.

[0019] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] An embodiment of the present invention provides an electrical equipment fault prediction method for predicting faults of electrical equipment of a robot manipulator. The electrical equipment is a key electrical equipment of the robot manipulator, including multiple categories, such as sensor category, motor category, controller category, power system category, etc., each category contains multiple electrical equipment. Sensor category may include temperature sensor, position sensor, force sensor, vision sensor, etc. Motor category may include servo motor, stepper motor, DC motor, etc. Controller category may include logic controller, microprogram controller, etc. Power system category may include power module, battery pack, regulator, etc.

[0021] Prior to this application, the prior art generally believed that due to the differences in the categories of electrical equipment of the robot manipulator and the complexity of data, it is impossible to design a universal prediction model to meet the needs of all equipment categories or all electrical equipment. This technical bias leads to the fact that the existing methods usually adopt a separate monitoring method, that is, to build a fault prediction model for each device separately. Based on such a technical bias, the inventors of this application have tried many ways to solve the problem of fault prediction of electrical equipment of the robot manipulator, but all failed to achieve the desired effect. The following are the main directions and problems that the inventors have tried under this technical bias. For example, on the basis of the existing model, the feature dimension and real-time data are added to improve the real-time and accuracy of the prediction results. Although this method improves the prediction accuracy to a certain extent, it cannot fundamentally solve the problems of model fixation, model complexity and management difficulties. For another example, the inventors tried to classify the same type of electrical equipment into the same equipment category (for example, temperature sensors, position sensors, etc. are divided into sensor categories, and servo motors, stepper motors, etc. are divided into motor categories), thereby dividing the electrical equipment of the robot manipulator into multiple equipment categories. For each equipment category, a separate fused prediction model is set, and these prediction models are independent of each other, without correlation and universality. This approach reduces the computational cost to a certain extent and improves the pertinence of the prediction model. However, this division still fails to fundamentally solve the scalability problems of a large number of models, high system complexity, and other electrical categories.

[0022] In the process of exploring solutions to the above problems, the inventors tried to use a universal model to meet the fault prediction needs of various electrical equipment. After verification, this method significantly improved the prediction performance, and the universal model can take into account the characteristics of different types of electrical equipment, improve the scalability and maintenance efficiency of the system, reduce the requirement for different types of equipment to be equipped with a single equipment model, and continuously use the data obtained during operation for model optimization, thereby improving versatility and accuracy. Therefore, the solution of the embodiment of the present invention is proposed. Figure 1FIG. 2 shows a schematic flow chart of a method for predicting electrical equipment faults according to an embodiment of the present invention. Figure 1 As shown, the electrical equipment fault prediction method includes: Step S100, obtaining the device identification, real-time data information X and historical prediction error H(X) of the target electrical device; Step S200, determine the target device category according to the device identifier, call the target optimization module corresponding to the target device category, and make the target optimization module optimize the preset initial fusion model F according to the following formula based on the real-time data information X and the historical prediction error H(X): base (X) Perform dynamic optimization to output the target prediction fusion model F opt (X), F opt (X)=F base (X)+(γ×e -β|H(X)| +λ s )×W s ×X; Among them, γ and β are coefficients, λ s represents the adaptive correction term, W s Represents the weight parameter of the target electrical equipment. One target optimization module corresponds to one equipment category. Step S300: Based on the equipment identification and real-time data information, the target prediction fusion model outputs the fault prediction result of the target electrical equipment.

[0023] According to the solution of the embodiment of the present invention, by obtaining the device identification, real-time data information X and historical prediction error H(X) of the target electrical device, and calling the target optimization module corresponding to the target device category based on the device identification, the initial fusion model F is optimized by using the real-time data information X and the historical prediction error H(X). base (X) Dynamically adjust and output the target prediction fusion model F opt (X), thereby solving the problems in the prior art of fault prediction models that are overly dependent on historical data, underutilized real-time data, and fixed models leading to poor prediction accuracy and timeliness. In addition, for different equipment categories (such as sensors, motors, controllers, power supply systems, etc.), the embodiment of the present invention extracts a prediction fusion model F that can be applied to key electrical equipment of the robot manipulator opt The general formula of (X) is obtained by dynamically adjusting the adaptive correction term λ s and weight parameter W s , which can adapt to the fault characteristics and prediction requirements of different equipment categories, thereby significantly improving the prediction accuracy and reliability while maintaining model consistency. In addition, the unified prediction fusion model F opt (X) It avoids building and maintaining multiple models for each device category.

[0024] In step S100, the device identification, real-time data information X and historical prediction error H(X) of the target electrical device are obtained from the storage query database. The device identification is a code or name used to uniquely identify the target electrical device. In the storage query database, the device identification is usually associated with basic information of the device such as model, manufacturer, installation location, etc. The device identification can be obtained by, for example, directly querying the corresponding device identification in the storage query database by inputting specific information of the device such as serial number, model code, etc., or by selecting the target device from the device list in the storage query database, and the system automatically extracts its device identification.

[0025] Real-time data information X refers to the real-time changing data such as the current working status and operating parameters of the target electrical equipment. These data are usually collected in real time through sensors, monitoring equipment, etc., or they can be data in the process of electrical equipment interacting with other electrical equipment. These are stored in the storage query database. The way to obtain real-time data information X can be the storage query database synchronization method. The storage query database system is synchronized with the monitoring system of the electrical equipment in real time to ensure that the real-time data information X in the storage query database is the latest. In addition, the API interface can be called. By calling the API interface provided by the storage query database, the model can obtain the real-time data information X of the target electrical equipment in real time. Another way is to set a scheduled task to regularly query and update the real-time data information X of the target electrical equipment from the storage query database.

[0026] The historical prediction error H(X) refers to the error record generated when the target electrical equipment was predicted for faults in the past. These error data are crucial for optimizing the prediction model and improving the prediction accuracy. When predicting faults, real-time data is obtained, and the next working condition is predicted through the model. Specifically, it can be the predicted data of the next working condition. The difference between the predicted data and the actual data of the next working condition is judged, and the fault is predicted in combination with the data change trend. The difference between each prediction and the actual data is recorded as the prediction error as the historical prediction error. The real-time data information X is updated in real time to ensure that the prediction model is based on the latest data for prediction. At the same time, the historical prediction error H(X) can be set to be updated regularly or in real time as needed to reflect changes in model performance.

[0027] In one embodiment, a storage query database is constructed using MySQL technology. The construction method of the storage query database includes: Step 1: Design the data model.

[0028] The data model includes a device information table, a real-time data table, a historical prediction error table, and a model parameter table. The device information table stores basic information about the device, including device identification, device category, and electrical equipment. The device category may be, for example, a sensor category, a motor category, and a power system category. Electrical equipment is an independent electrical device under each device category, such as a temperature sensor and a position sensor. The real-time data table stores real-time data of the device, such as sensor readings, motor current, etc., and supports fast writing and querying. The historical prediction error table stores the historical prediction errors of the device. The model parameter table stores the parameters of the prediction model, such as the weight parameter W. s , adaptive correction term λ s .

[0029] Step 2: partition or shard the data.

[0030] Data partitioning is to divide a large table into multiple smaller sub-tables (i.e., partitions) according to certain rules such as time, region, etc. Each partition can be stored and queried independently. In one embodiment, time data is partitioned, for example, the real-time data table and the historical prediction error table are partitioned by time, such as by day or month. In this way, the amount of data for a single query can be reduced, that is, when querying data for a certain time period, only the corresponding partition needs to be scanned instead of the entire table. In addition, after partitioning, the storage query database can process query requests for multiple partitions in parallel.

[0031] Data sharding is to distribute the data of a large table to multiple physical storage nodes, with each node storing only part of the data. In one embodiment, the device data is sharded, for example, the data is stored in shards according to the device identifier, and the data of different devices are distributed to different physical storage nodes. In this way, the query concurrency capability can be improved, that is, the queries of different devices can be distributed to different nodes, avoiding a single node from becoming a performance bottleneck. In addition, horizontal expansion can also be supported, that is, by adding storage nodes, the storage and computing capabilities of the storage query database can be easily expanded.

[0032] Step three: adopt a tiered storage architecture and use automated strategies to achieve dynamic migration between hot data and cold data.

[0033] High-frequency query data, such as the latest real-time data and historical prediction errors of electrical equipment, are stored in Redis, while cold data, such as historical data, is stored in MySQL. Through automated strategies such as TTL mechanism and cache preheating, dynamic migration between hot data and cold data is achieved to ensure that high-frequency access data can respond quickly while reducing the pressure on persistent storage. In this architecture, Redis is used as an in-memory storage query database to store and quickly access hot data such as the latest real-time numbers, while MySQL is used as a disk storage query database to store cold data such as historical numbers. Through a reasonable data migration strategy, the system can optimize the use of storage resources while ensuring query efficiency.

[0034] Step 4: Optimize the performance of the storage query database.

[0035] The performance optimization design of the storage query database includes index optimization, query cache, batch writing and asynchronous processing. In the index optimization design, composite indexes are created for high-frequency query fields to speed up queries, query logs are analyzed regularly, index strategies are adjusted dynamically, and indexes for low-frequency query fields are deleted to reduce storage overhead. In the query cache design, Redis is used to cache high-frequency query results, such as the latest real-time data and historical prediction errors of the device, and a reasonable cache expiration time is set to ensure the timeliness of the cached data. In the batch writing and asynchronous processing design, batch processing is used for the write operation of real-time data to reduce the write frequency of the storage query database, and the update operation of the historical prediction error is asynchronous to avoid blocking the writing of real-time data.

[0036] When setting up the database, a dual database system is used, including a cloud database and a local database. The cloud database is used to collect data from scattered electrical equipment, and the local database can interact with the cloud database. Through the cloud database, local electrical equipment can call data from other electrical equipment. When establishing and optimizing the model, a large amount of data in the cloud database can ensure the accuracy of the model, while the data in the local database can make the target optimization module more accurate in adjusting the specific electrical equipment model. In this way, a large amount of historical data of the same type of electrical equipment can be borrowed, so that the initial fusion model can have a relatively high compliance from the beginning, and when the number of specific electrical devices in use is large, the unit cost of use is greatly reduced.

[0037] In step S200, each device category has different operating characteristics and failure modes, so it is necessary to adjust and optimize the prediction model for different device categories so that the model can more accurately reflect the devices of that category, that is, to add some specific adjustments under the general model. Each device category corresponds to a target optimization module, which is responsible for dynamically optimizing the prediction model of the device of that category. The target optimization module adjusts the model parameters such as γ, β, , W s And the corresponding initial fusion model F base (X) to adapt to the operating status of different devices. However, regardless of the type of device, the same formula can be used to output the target prediction fusion model F opt (X).

[0038] It can be understood that the initial fusion model F base (X) is the basic prediction model trained based on historical data, and the F of different device categories base (X) is usually different because different classes of equipment have different operating characteristics. -β|H(X)| is a dynamic correction term based on the historical forecast error H(X), γ and β are used to control the amplitude and decay rate of the correction term, e -β|H(X)| It means that the influence of error on the correction term decreases as the absolute value of error increases. Adaptive correction term λ s It can be dynamically adjusted according to the characteristics of the equipment category and the changes in real-time data to further adjust the model output. By introducing historical prediction errors H(X) and real-time data information X, the model output is dynamically adjusted so that the model can adapt to changes in equipment status. Different initial fusion models F are designed for different equipment categories. base (X) and optimize parameters to improve the pertinence of the model. Use the historical forecast error H(X) to correct the model's deviation and ensure that the model is more accurate in future forecasts.

[0039] Figure 2 The initial fusion model F according to one embodiment of the present invention is shown. base Schematic flow chart of the method for determining (X). Figure 2 As shown, the initial fusion model F base Methods for determining (X) include: Step S201, classifying the electrical equipment of the robot arm into categories, thereby obtaining a plurality of equipment categories; Step S202, for each device category, collect historical data and perform feature extraction; Step S203, selecting at least two basic prediction models matching the feature type from the model database according to the extracted feature type; Step S204: Use the historical data of the equipment category to train the selected basic prediction model, calculate the error index, and select the optimal combination mode of the equipment category as the initial fusion model F of the equipment category. base (X).

[0040] In step S201, the device categories include sensor categories, motor categories, and driver categories, etc. The initial fusion model of the electrical devices in each category can be the same, using the same model parameters. Further, γ, β, λ s and W s Settings can be made for each specific electrical device. In this solution, different specific electrical devices can be distinguished, and the prediction will be more detailed and accurate. However, in general, the correlation coefficient and other specific data for a device category can fully meet the requirements.

[0041] In step S202, historical operation data of the electrical equipment is collected, such as sensor data, motor data, and driver data. Among them, the sensor data may include temperature, position, pressure, current, voltage, etc. The motor data may include, for example, current, voltage, speed, torque, etc. The driver data may include, for example, voltage, current, and power, etc. The key features extracted from the historical data may include, for example, statistical features, time domain features, and frequency domain features. Among them, the statistical features include mean, variance, maximum value, and minimum value, etc. The time domain features include peak value and waveform factor, etc. The frequency domain features include spectrum energy and main frequency, etc.

[0042] In step S203, the model database stores various models, such as linear regression models, support vector machines, random forests, neural networks, etc. For different equipment categories, the selected basic prediction models may be different or the same. In order to improve efficiency, it is preferred to use the same basic prediction model. For example, for sensor categories, the extracted features are, for example, mean, variance, maximum value, minimum value, peak value, waveform factor, spectrum energy, main frequency, skewness, kurtosis, etc., and the basic models that can be selected are linear regression models, support vector machines, and random forests. For motor categories, the extracted features are, for example, current, speed, torque, vibration amplitude, vibration frequency, motor temperature, ambient temperature, etc., and the basic models that can be selected are linear regression models, support vector machines, and random forests. For power supply categories, the extracted features are, for example, voltage, current, power, voltage fluctuation, current fluctuation, power supply temperature, radiator temperature, etc., and the basic models that can be selected are linear regression models, support vector machines, and neural networks.

[0043] Figure 3 Shows Figure 2 The initial fusion model F of step S204 is shown base (X) is a schematic flow chart of the method for obtaining. Figure 3 As shown, step S204 includes: Step S2041, dividing the historical data of each device category into segments according to time periods, and using the historical data of each time period to train the selected basic model; Step S2042, calculate the error of each basic model in each time period according to the following formula: ; in, represents the jth basic model f j In time period T i The error within Indicates time period T i The number of samples within Indicates the actual value, represents the predicted value; Step S2043, for each time period, selecting the basic model with the smallest error as the optimal model in the time period; Step S2044: Count the frequencies of each basic model being selected as the optimal model in all time periods according to the following formula: , ; in, represents the jth basic model, For time period T i The optimal model within represents the indicator function; Step S2045, selecting at least two basic models with the highest frequencies to be fused as target basic models; if the frequencies of multiple basic models are the same, selecting at least two basic models with the smallest errors as target basic models; Step S2046: weighted fusion is performed on the selected target basic models to obtain an initial fusion model F base (X).

[0044] In step S4021, it is assumed that the divided time periods are T1, T2, ..., T m , the basic model is f1,f2,...,f n , calculate the error in each time period by the formula in step S2042, and obtain E 11 、E 21 、E 31 ,...,E m1 , E 12 ,E 22 ,E 32 ,...,E mn, compare these errors, determine the basic model corresponding to the minimum error value, and use the basic model with the minimum error value as the optimal model in this time period. On this basis, count the frequencies of each basic model being the optimal model in all time periods. For example, if the hth and kth models are the optimal models in the time period, then f h 、f k As the target basic model, we can further convert the model f h 、f k Perform weighted fusion, and select the corresponding weight of each model in a weighted fusion manner. Usually, two optimal models are selected for weighted fusion, and the number of models used for weighted fusion can also be increased as needed. Furthermore, a secondary judgment is made on the data obtained at different times. If the optimal model has a regular distribution in the divided time periods, the corresponding model with a smaller error is selected in different time periods. For example, different models are selected for use in the startup phase, normal operation phase, emergency phase, and before the end of the equipment.

[0045] In order to further optimize the accuracy of the initial fusion model, this application further refines the weighted fusion method, and makes the model more accurate by introducing more possible models. The specific weighted fusion in this solution includes the following steps: For each target base model, calculate its average error over all time periods; According to the target basic model and the error, the i-th basic model is calculated according to the following formula: Weight , ; Among them, S represents the set of target basic models, ε represents the adjustment coefficient, Represents the i-th basic model The average error of j is the number of all basic models to be used; get the initial fusion model F base The formula for (X) is ; in, , k represents the number of selected target basic models, and the initial fusion model F is obtained base (X) In this way, the weight of the basic model to be used can be obtained according to the average error. By allocating different weights, multiple models can be taken into account, further reducing the impact of the error.

[0046] In order to improve the model's ability to continuously adapt to changes in equipment components, such as replacement of specific electrical equipment, normal changes during the use cycle, etc., and eliminate the impact of these normal changes, the initial fusion model F baseAfter the formula of (X) is determined, the average error of each basic model on the latest historical data is calculated at preset intervals. When the average error of the latest historical data exceeds the threshold error, the model update is triggered to reselect the optimal basic model or adjust the weight of the target basic model, thereby obtaining an updated initial fusion model. In this scenario, since the data used by the basic model is the data information collected in the most recent period of time, it will be closer to the actual situation, so that the data information with reduced relevance in the historical data previously used to form the model can be eliminated, thereby obtaining a prediction model that is more in line with the existing device.

[0047] In the specific use process, in order to accurately predict fault-related information, a large amount of electrical equipment data will be used to build a general prediction model. When the relevant model is applied to a specific electrical device, the target electrical equipment will be optimized. As the electrical device runs, it is possible to consider reducing the proportion of other electrical equipment data when the general model was previously constructed, so that the prediction model is more in line with the prediction of the current electronic equipment. Therefore, it is necessary to optimize the data range used. At the same time, it is also necessary to avoid the adverse effects of data fluctuations to achieve a smooth transition. In the technical solution of this application, the above purpose is achieved by using historical data from the latest period of time. In other cases, when electrical equipment is updated, this specific method can also be used to update the data, thereby optimizing the model.

[0048] With the use of the target prediction fusion model, the weight parameters can also be adjusted according to the historical prediction errors of the equipment categories. The initial weight parameters of the corresponding equipment categories are used as the benchmark, and the historical prediction errors are taken into account. The initial fusion model F corresponding to different equipment categories base The weight parameter W in (X) s Satisfies the following formula: ; Among them, W s (0) Represents the initial weight parameters of the corresponding device category. In this scheme, the system can adjust the weight parameters in the target prediction fusion model according to the initial weight parameters, and can optimize the weight parameters as the usage process changes, thereby making the target prediction fusion model more effective.

[0049] In order to further accurately adjust the error of the prediction model, the adaptive correction term λ s The acquisition method of is further refined, including the following steps: First, the system comprehensively collects the historical prediction error data of all electrical equipment in the target equipment category, and pre-sets a basic correction term λ s (0), as the basis for building an adaptive correction mechanism to ensure the accuracy and reliability of subsequent calculations; then, in order to evaluate the overall prediction error of the entire target device category, the system adopts the average error calculation method. Specifically, the average error H of all electrical devices in the target device category is calculated according to the following formula s , ; Where N is the number of electrical devices in the target device category, and H i represents the historical prediction error of the i-th electrical device in the target device category. This step helps to determine the prediction error level of the target device category. Subsequently, in order to fine-tune the correction term, the impact factor α of the target device category is defined. s and the influencing factors of the target electrical equipment , respectively, are calculated as follows: α s =e -k|Hs| , , k and p are adjustment coefficients, which are used to adjust the sensitivity of the above two influencing factors to the average error Hs. According to the formula Calculate the adaptive correction term λ s .

[0050] In the above steps, the historical prediction error data is introduced, and the average error of all electrical equipment is obtained through the average error calculation method. Then, the relationship (sensitivity) between the target electrical equipment and the errors of all electrical equipment is determined. The adaptive correction item in the prediction model of the target equipment is selected through the sensitivity relationship to reduce the difference between the target electrical equipment and all electrical equipment, so that the corrected prediction model is more relevant to the target electrical equipment and the impact of other electrical equipment on the error is reduced.

[0051] Optionally, the adjustment coefficient k is obtained according to the following formula: ; ; The adjustment coefficient p is obtained according to the following formula: ; ; Among them, k0 and p0 are the initial adjustment coefficients, t k and t p All are adjustment parameters. By putting historical data into the above formula, the corresponding adjustment parameters are determined for different target devices. The historical data is used to make the adaptive correction items closer to the actual situation and more accurately determine the sensitivity of different electrical devices relative to all electrical devices.

[0052] In order to more accurately reflect the historical prediction error, the data at different times are smoothed and the change trend is adjusted. The following method is used to obtain the historical prediction error H(X) of the target electrical equipment, which specifically includes the following steps: Obtain the prediction error H of the target electrical equipment at time t-1 t-1 (X), prediction error H at time t-2 t-2 (X) and the mean prediction error H of the M moments before time t-1 m (X); Based on the weighted moving average method, the first historical prediction error H is calculated according to the following formula EWMA,t-1 (X), ; Wherein, q represents the smoothing factor, and 0<q<1; The second historical prediction error T is calculated according to the following formula t (X), T t-1 (X)=r×(H t-1 (X)-H t-2 (X)); Where r represents the trend adjustment coefficient, and 0.5<r<1.5; According to the formula H(X)=θ1×H EWMA,t-1 (X)+θ2×T t-1 (X) Calculate the historical prediction error H(X), where θ1 and θ2 are both constants, and the sum of the two is equal to 1. Through the above steps, it is possible to eliminate the impact of the change trend that should appear in some design situations according to the change trend of the equipment data. Specifically, in some cases, there will be normal and reasonable change trends in the use process. If the relevant change trends are not considered, the expansion or contraction trends within the normal range will be ignored, which will bring unrealistic deviations to the prediction model. On the basis of introducing the smoothing factor and the trend adjustment coefficient, if there is no change, the smoothing factor can be selected as 0.5 and the trend adjustment coefficient can be selected as 1. When there is a change trend, the balance factor starts to change from 0.5 and the trend adjustment coefficient starts to change from 1. According to historical data, the most accurate and practical balance factors and trend adjustment coefficients for different electrical equipment are determined.

[0053] In step S300, the prediction fusion model of the target electrical equipment is modified, and the target electrical equipment is predicted to fail or the probability of failure according to the equipment identification and real-time data information. Specifically, the possible data parameters of the next working condition of the target electrical equipment are obtained according to the prediction fusion model, and the failure is judged according to the obtained possible data parameters of the next working condition. The relevant information is warned and provided to the user. In some specific applications, the output result of the model can be directly converted into the failure probability for display.

[0054] In addition, the present application also provides an electrical equipment fault prediction system, which includes at least one processor, and at least one processor is used to execute computer programs or instructions to perform operations corresponding to the aforementioned electrical equipment fault prediction method. The entire system is also provided with a database, and the processor has a relevant model. In the specific use process, the device identification, real-time data information and historical prediction error of the target electrical equipment are obtained, the target device category is determined according to the device identification, and the target optimization module corresponding to the target device category is called, and the target optimization module is dynamically optimized based on the real-time data information and the historical prediction error to output the target prediction fusion model, and based on the device identification and real-time data information, the target prediction fusion model outputs the prediction result of the target electrical equipment.

[0055] The prediction system further includes a database and a communication unit. The database adopts a local database and a cloud database. The cloud database is designed as a core hub for data collection and sharing. It can widely collect data from scattered electrical equipment in various places. These data cover multi-dimensional information such as the operating status, performance parameters, and environmental conditions of the equipment. The distributed storage and powerful processing capabilities of the cloud database enable it to efficiently process massive amounts of data, providing a basis for subsequent data analysis, model building, and optimization. Through the cloud database, electrical equipment in different locations can break through geographical restrictions and achieve data interconnection. Local electrical equipment can therefore easily call data from other electrical equipment, promoting the sharing and utilization of information resources.

[0056] The local database serves as a supplement to the cloud database, storing real-time data of local electrical equipment and being able to obtain required data from the cloud database as needed. This allows the local database to maintain data independence while making full use of the resources in the cloud database, providing strong support for model optimization and fault diagnosis of local electrical equipment.

[0057] In the process of model establishment and optimization, the large amount of data in the cloud database plays an important role. These data cover a wide range of operating experience of the same type of electrical equipment, providing rich samples for model training and verification. By deeply mining the laws and characteristics of these data, a more accurate and reliable electrical equipment model can be established. The data in the local database enables the target optimization module to adjust the model more accurately for specific electrical equipment. This combination of global data and local data not only ensures the universality of the model, but also takes into account the particularity of the model, improving the adaptability and accuracy of the model.

[0058] Through the setting of this dual database system, a large amount of historical data of the same type of electrical equipment can be borrowed at the beginning of model establishment, so that the initial fusion model can have relatively high compliance and accuracy, which not only reduces the time and cost required for model establishment, but also improves the effect of the model in practical application. At the same time, when the number of specific electrical devices in use is large, due to the synergy of the cloud database and the local database, the unit cost of use is greatly reduced, and the scale effect can further highlight the economy and practicality.

[0059] In summary, the present invention proposes an electrical equipment fault prediction method and system thereof, which improves the accuracy and timeliness of fault prediction through a dynamic optimization model. First, the equipment identification, real-time data information and historical prediction error of the target electrical equipment are obtained. After obtaining the necessary data, the present invention determines the target equipment category according to the equipment identification, and calls the corresponding target optimization module. The target optimization module dynamically optimizes the preset initial fusion model according to a specific formula based on the real-time data information and the historical prediction error, and outputs the target prediction fusion model, taking into account the influence of the historical prediction error, and dynamically adjusting the adaptive correction term and weight parameter so that the model can adapt to the fault characteristics and prediction requirements of different equipment categories.

[0060] The method for determining the initial fusion model is also one of the innovations of the present invention. By classifying the electrical equipment of the robot arm, collecting historical data and extracting features, a basic prediction model matching the feature type is selected from the model database. The selected basic prediction model is trained using historical data, and the error index is calculated. Finally, the optimal combination is selected as the initial fusion model to ensure that the initial fusion model has high accuracy and generalization ability.

[0061] In order to further optimize the accuracy of the model, the present invention also proposes a refinement of the weighted fusion method, by introducing more possible models and calculating the weights of each model based on the average error, so as to obtain a more accurate initial fusion model. At the same time, the present invention also considers the problem of continuous adaptation and updating of the model, by regularly calculating the average error of each basic model on the latest historical data, and triggering the model update when the error exceeds the threshold, so as to ensure the timeliness and accuracy of the model.

[0062] The present invention applies the target prediction fusion model to actual fault prediction, and outputs the fault prediction results of the target electrical equipment based on equipment identification and real-time data information, which can provide strong support for the maintenance and management of the equipment and reduce the possibility and impact of faults.

[0063] In addition, the present invention also provides an electrical equipment fault prediction system, which includes at least one processor for executing a computer program or instruction to implement the above electrical equipment fault prediction method. The entire system provides a comprehensive, accurate and real-time solution for electrical equipment fault prediction through efficient data management and model optimization strategies.

[0064] The above is only a specific implementation of the present application, and any improvements made based on the concept of the present application are deemed to be within the protection scope of the present application.

Claims

1. A method for predicting electrical equipment failure, characterized in that: The method is used to predict the fault of the electrical equipment of the robot manipulator, including the following steps: Obtaining the equipment identification, real-time data information X and historical prediction error H(X) of the target electrical equipment; Determine the target device category according to the device identifier, call the target optimization module corresponding to the target device category, and make the target optimization module perform the preset initial fusion model F according to the following formula based on the real-time data information X and the historical prediction error H(X): base (X) Perform dynamic optimization to output the target prediction fusion model F opt (X), F opt (X)=F base (X)+(γ×e -β|H(X)| +λ s )×W s ×X; Among them, γ and β are coefficients, λ s represents the adaptive correction term, W s Represents the weight parameter of the target electrical equipment. One target optimization module corresponds to one equipment category. Based on the device identification and the real-time data information, the target prediction fusion model outputs a prediction result of the target electrical device.

2. The electrical equipment fault prediction method according to claim 1, characterized in that: The adaptive correction term λ s The method for obtaining includes the following steps: Obtain the historical prediction errors and preset basic correction items λ of all electrical equipment in the target equipment category s (0) ; Calculate the average error H of all electrical devices in the target device category according to the following formula: s , ； Where N is the number of electrical devices in the target device category, and H i represents the historical prediction error of the i-th electrical equipment in the target equipment category; Define the impact factor α of the target device category s and the influencing factors of the target electrical equipment , α s =e -k|Hs| , , k and p both represent adjustment coefficients; According to the formula Calculate the adaptive correction term λ s .

3. The electrical equipment fault prediction method according to claim 2, characterized in that: The adjustment coefficient k is obtained according to the following formula: ； ； The adjustment coefficient p is obtained according to the following formula: ； ； Among them, k0 and p0 are the initial adjustment coefficients, t k and t p These are adjustment parameters.

4. The electrical equipment fault prediction method according to any one of claims 1 to 3, characterized in that: The method for obtaining the historical prediction error H(X) of the target electrical equipment comprises the following steps: Obtain the prediction error H of the target electrical equipment at time t-1 t-1 (X), prediction error H at time t-2 t-2 (X) and the mean prediction error H of the M moments before time t-1 m (X); Based on the weighted moving average method, the first historical forecast error H is calculated according to the following formula: EWMA,t-1 (X), ； Wherein, q represents the smoothing factor, and 0<q<1; The second historical prediction error T is calculated according to the following formula t (X), T t-1 (X)=r×(H t-1 (X)-H t-2 (X)); Where r represents the trend adjustment coefficient, and 0.5<r<1.5; According to the formula H(X)=θ1×H EWMA,t-1 (X)+θ2×T t-1 (X) Calculate the historical prediction error H(X), where θ1 and θ2 are both constants, and their sum is equal to 1.

5. The electrical equipment fault prediction method according to claim 4, characterized in that: The initial fusion model F base The method for determining (X) comprises the following steps: Classify the electrical equipment of the robot manipulator into categories, thereby obtaining multiple equipment categories; For each device category, historical data is collected and feature extraction is performed; Selecting at least two basic prediction models matching the feature type from a model database according to the extracted feature type; The selected basic prediction model is trained using the historical data of the equipment category, and the error index is calculated to select the optimal combination mode of the equipment category as the initial fusion model F of the equipment category. base (X).

6. The electrical equipment fault prediction method according to claim 5, characterized in that: The selected basic prediction model is trained by using the historical data of the equipment category, and the error index is calculated to select the optimal combination mode of the equipment category as the initial fusion model F of the equipment category. base (X), comprising the following steps: The historical data of each device category is segmented according to time periods, and the selected basic model is trained using the historical data of each time period; The error of each basic model in each time period is calculated according to the following formula: ； in, represents the i-th basic model f i In time period T i The error within Indicates time period T i The number of samples within Indicates the actual value, represents the predicted value; For each time period, the basic model with the smallest error is selected as the optimal model in that time period; The frequency of each basic model being selected as the optimal model in all time periods is counted according to the following formula , ； in, represents the jth basic model, For time period T i The optimal model within represents the indicator function; Select at least two basic models with the highest frequency to be fused as the target basic model. If the frequencies of multiple basic models are the same, select at least two basic models with the smallest error as the target basic model. Perform weighted fusion on the selected target basic models to obtain the initial fusion model F base (X).

7. The electrical equipment fault prediction method according to claim 6, characterized in that: The selected target basic models are weightedly fused to obtain the initial fusion model F base (X), comprising the following steps: For each target base model, calculate its average error over all time periods; According to the target basic model and the error, the i-th basic model is calculated according to the following formula: Weight , ； Among them, S represents the set of target basic models, ε represents the adjustment coefficient, Represents the i-th basic model The average error of Get the initial fusion model F base The formula for (X) is ； in, , k represents the number of selected target base models.

8. The electrical equipment fault prediction method according to claim 7, characterized in that: The initial fusion model F base After the formula of (X) is determined, the following steps are also included: At every preset time interval, calculate the average error of each basic model on the latest historical data; After the average error of the latest historical data exceeds the threshold error, the model update is triggered to reselect the optimal basic model or adjust the weight of the target basic model to obtain an updated initial fusion model.

9. The method for predicting electrical equipment failure according to any one of claims 1-3 and 5-8, characterized in that: Initial fusion model F corresponding to different device categories base The weight parameter W in (X) s Satisfies the following formula: ； Among them, W s (0) Indicates the initial weight parameter of the corresponding device category.

10. An electrical equipment fault prediction system, characterized in that: The method comprises at least one processor, wherein the at least one processor is used to execute computer programs or instructions to perform operations corresponding to the method for predicting faults of electrical equipment according to any one of claims 1 to 8.

Images