Materials management business supervision methods, devices, equipment and media

Abstract

The present application relates to the field of information technology, and provides a material management business supervision method, device, equipment and medium, including: constructing a multi-dimensional material attribute database based on the storage parameters of the stored power grid materials; collecting the corresponding operating status data in real time for the in-service power grid materials, and judging the health status of the power grid materials in combination with the material health assessment expert knowledge base; when the difference between the actual remaining operating life of the power grid materials and the estimated remaining life of the materials is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid materials are marked as predicted waste materials; for the predicted waste materials, searching for alternative replacement material information with similar models and specifications, and generating alternative replacement material plans; using genetic algorithms and particle swarm optimization algorithms, based on the alternative replacement material plans, generating the optimal replacement plan that minimizes the full life cycle cost of power grid materials. This plan realizes the intelligent and refined management of the full life cycle of power grid materials.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a material management business supervision method, device, equipment and medium. Background Art

[0002] The management of power grid materials throughout their life cycle faces the complex challenge of accurately predicting the remaining life of materials, assessing their health status, and formulating optimal replacement plans. Existing management methods make it difficult to fully grasp the multi-dimensional attribute data and dynamic operating status of materials, resulting in an inability to accurately grasp the actual service life and performance degradation process of materials. At the same time, material replacement decisions lack systematic considerations, often focusing only on short-term costs while ignoring the benefits of the entire life cycle. This makes power grid companies face both resource waste caused by premature scrapping and safety hazards caused by delayed updates in material management. How to build a data collection and analysis system covering the entire life cycle of materials, achieve accurate perception and scientific prediction of material status, and quickly find the optimal replacement strategy that balances safety, economy, and reliability among a large number of alternative options is of great significance to improving the level of power grid material management and reducing the cost of the entire life cycle. Summary of the invention

[0003] The present invention provides a material management business supervision method, which mainly includes:

[0004] For each piece of grid material entering the warehouse, obtain the corresponding entry parameters and build a multi-dimensional material attribute database;

[0005] Based on the historical data in the material attribute database, a support vector machine or a random forest algorithm is used to train a material remaining life prediction model, and the material service life, operating load and failure frequency of the in-service power grid materials are input into the material remaining life prediction model to output the estimated remaining life of the materials;

[0006] For in-service power grid materials, the corresponding operating status data is collected in real time, and the current health status of power grid materials is judged in combination with the material health assessment expert knowledge base;

[0007] When the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid material is marked as a predicted waste material; for the predicted waste material, search for information on alternative replacement materials with similar models and specifications, and generate an alternative replacement material plan;

[0008] Genetic algorithm and particle swarm optimization algorithm are used to generate the optimal replacement plan that minimizes the life cycle cost of power grid materials based on alternative replacement material plans;

[0009] Record the full-process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full-process data, formulate a power grid material management update plan and perform a visual display.

[0010] The present invention provides a material management business supervision device, which mainly includes:

[0011] The acquisition module is used to obtain the corresponding storage parameters for each power grid material entering the warehouse and build a multi-dimensional material attribute database;

[0012] A training module is used to train a material remaining life prediction model based on historical data in a material attribute database by using a support vector machine or a random forest algorithm, and input the material service life, operating load and failure frequency of in-service power grid materials based on the material remaining life prediction model to output the estimated remaining life of the materials;

[0013] The first acquisition module is used to collect the corresponding operating status data of the in-service power grid materials in real time, and judge the current health status of the power grid materials in combination with the material health assessment expert knowledge base;

[0014] A marking module, used for marking the corresponding power grid material as a predicted waste material when the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than a first time threshold or the health status shows a continuous downward trend within a preset period;

[0015] The first generation module is used to search for alternative replacement material information with similar models and specifications for predicted waste materials, and generate alternative replacement material plans;

[0016] The second generation module is used to generate an optimal replacement plan that minimizes the full life cycle cost of power grid materials based on alternative replacement material plans by using a genetic algorithm and a particle swarm optimization algorithm;

[0017] The optimization module is used to record the full process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and perform a visual display.

[0018] The present invention further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0019] The present invention also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any of the above methods are implemented. The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0020] The present invention discloses a method for intelligent management of power grid materials throughout their life cycle. The method constructs a multi-dimensional material attribute database, trains a remaining life prediction model, evaluates the health status of materials in real time, automatically marks predicted waste materials, and generates an optimal replacement plan. In view of the problem that it is difficult to grasp the timing of scrapping of power grid materials and it is difficult to optimize replacement plans, the present invention uses a machine learning algorithm to predict the remaining life, combines an expert knowledge base to evaluate the health status, and uses an intelligent optimization algorithm to generate a replacement plan that minimizes the cost of the entire life cycle. By recording the entire process data and continuously optimizing the management strategy, the intelligent and refined management of the entire life cycle of power grid materials is achieved, effectively extending the service life of materials, reducing operation and maintenance costs, and improving asset utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The present invention is a flow chart of a material management business supervision method.

[0022] Figure 2 The present invention is a schematic structural diagram of a material management business supervision device.

[0023] Figure 3 The figure is a schematic diagram of the structure of a computer device of the present invention. DETAILED DESCRIPTION

[0024] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0025] like Figure 1 The present embodiment provides a material management business supervision method, which may specifically include steps S101-S107:

[0026] S101. For each piece of grid material entering the warehouse, the corresponding entry parameters are obtained, and a multi-dimensional material attribute database is constructed.

[0027] S102. Based on the historical data in the material attribute database, a support vector machine or a random forest algorithm is used to train a material remaining life prediction model, and based on the material remaining life prediction model, the material service life, operating load and failure frequency of the in-service power grid materials are input, and the estimated remaining life of the materials is output.

[0028] The material attribute data obtained from the material attribute database is used as historical data; training data is formed based on historical data, and the material service life, operating load, and fault frequency data of the power grid materials contained in the training data are normalized to obtain the values ​​A, B, and C, and the material operation ratio D is calculated based on the material service life and the material operation time. A random forest or support vector machine is used to build the model, and a grid search is used based on the values ​​A, B, C, and D to find the optimal value of the random forest algorithm or the support vector machine algorithm hyperparameter to obtain the model parameter configuration data. Based on the training data and model parameter configuration data containing the material service life, operating load, and fault frequency, it is determined whether the number of training data is greater than the preset number threshold. If it is greater than the preset number threshold, machine learning training is performed; if the number of training data is less than or equal to the preset number, the external database is collected as new material attribute data and added to the material attribute database, and the material attribute database is updated by collecting new material attribute data. After the initial prediction model is obtained through machine learning training, the verification data set is input for verification to determine whether the average error is higher than the preset error threshold. If the average error is higher than the preset error threshold, the initial prediction model is used to fuse the K nearest neighbor algorithm to obtain the material remaining life prediction model. If the average error is lower than or equal to the preset error threshold, the initial prediction model is determined as the material remaining life prediction model. The material remaining life prediction model is used to predict the material life of the in-service power grid materials based on the input material service life, operating load and failure frequency of the in-service power grid materials to obtain the predicted value, and the predicted value is denormalized to obtain the estimated remaining life of the material.

[0029] Specifically, the material attribute database is the basis of power grid material management. It contains key information such as the service life of materials, operating load, and fault frequency. In order to effectively predict the remaining life of materials, these data need to be processed by feature engineering. For example, for a transformer, its service life may be 20 years, it has been in operation for 15 years, the operating load is 80%, and the average number of failures per year is 2 times. Through normalization, these data can be converted into standardized values ​​A, B, and C, which are convenient for model training and comparison. The importance of feature engineering is reflected in its ability to extract the most representative features and improve the prediction accuracy of the model. For example, by calculating the material operation ratio D (operating time / service life), the degree of material use can be more intuitively reflected. For the above transformer, its operation ratio D is 0.75, indicating that 75% of the expected life has been used. Random forest and support vector machine are commonly used machine learning algorithms, and they each have their own advantages. Random forest is suitable for processing high-dimensional data, while support vector machine performs well in processing nonlinear problems. Model performance can be optimized by finding the best hyperparameters through grid search. For example, for random forests, it may be found that the model performs best when the number of trees is set to 100 and the maximum depth is 10. The amount of training data directly affects the reliability of the model. If the existing data is insufficient, you can consider collecting supplementary data from external databases. For example, you can obtain usage data of similar materials from other power companies or equipment manufacturers to enrich the training set. This can not only improve the generalization ability of the model, but also enhance the prediction accuracy of material life under different working environments. Model validation is a key step to ensure prediction accuracy. If the average error of the initial model is higher than the preset threshold, for example, the prediction error exceeds 10%, model fusion technology is needed to improve accuracy. The K nearest neighbor algorithm can make predictions based on historical data of similar materials. After combining with the initial model, it can comprehensively consider the individual differences and overall trends of materials to improve the accuracy and stability of predictions. The resulting material remaining life prediction model can provide important decision support for power grid material management. For example, for a batch of cables that are about to reach their expected service life, the model prediction finds that the corresponding actual remaining life is still 5 years, which can avoid unnecessary replacement and save costs. On the contrary, if it is found that the actual life of some equipment is much lower than expected, the maintenance strategy or update plan can be adjusted in time to ensure the safe and stable operation of the power grid. Through this data-driven forecasting method, power grid companies can achieve refined management of the entire life cycle of materials, optimize resource allocation, and improve economic benefits and operational efficiency. At the same time, this method also provides reliable data support for the formulation of scientific procurement strategies and maintenance plans, helping to build a more intelligent and sustainable power grid system.

[0030] S103. For in-service power grid materials, corresponding operating status data is collected in real time, and the current health status of the power grid materials is determined in combination with the material health assessment expert knowledge base.

[0031] According to the equipment type of the power grid materials in service, sensors are deployed to the field equipment to obtain the operating status data of the equipment at different times. It is judged whether the acquisition frequency of the acquired operating status data reaches the preset threshold. If not, the acquisition frequency of the sensor is adjusted until the preset threshold is reached. According to the degree of damage of each device in the operating status data, it is judged whether the preset damage threshold is reached. If the preset damage threshold is reached, the principal component analysis algorithm is used to reduce the dimension of the equipment operating status data, extract the main features in the operating status data, and calculate the influence weight of each principal component on the equipment performance according to the pre-established equipment performance indicators. The historical operating status data and the influence weight are used as samples, combined with the health status labels in the material health assessment expert knowledge base, and the multivariate linear regression model is used for training to establish a health status regression model. The current operating status data of the equipment is obtained at a certain time point, and it is used as the input of the health status regression model. The health status value of the equipment at that time point is calculated by the health status regression model. According to the preset health status threshold interval, the health status of the current equipment, that is, the health status of the power grid materials, is judged.

[0032] Specifically, the monitoring of the health status of power grid materials is an important part of ensuring the safe and stable operation of the power system. First, sensors are deployed according to the type of equipment to collect operating status data. For example, sensors such as temperature, vibration, and partial discharge can be installed on transformers to monitor their operating parameters in real time. The acquisition frequency must reach a preset threshold, such as once every 5 minutes, to ensure the timeliness and continuity of the data. The acquired operating status data is used to determine the degree of equipment damage. Taking the circuit breaker as an example, indicators such as the number of operations and the degree of contact wear can be monitored. When an indicator exceeds the preset threshold, such as the number of operations exceeds 5,000 times, the judgment rule in the expert knowledge base is triggered. The knowledge base contains the maintenance standards of various types of equipment, and combines historical data to comprehensively evaluate the necessity of maintenance. In order to improve the efficiency of data analysis, the principal component analysis method is used for dimensionality reduction processing. Taking the transmission line as an example, multidimensional data such as voltage, current, and power can be reduced to 2-3 principal components. According to the pre-established performance indicators, such as line transmission efficiency, the influence weight of each principal component is calculated. This method not only retains key information, but also simplifies the subsequent modeling process. The multivariate linear regression model is trained using the reduced-dimensional historical data and influence weights. The model can reflect the relationship between equipment performance and operating status. For example, for distribution transformers, the model input may include load rate, oil temperature, winding temperature, etc., and the output is the health status value of the transformer. The advantage of this method is that it can quickly evaluate the equipment status and facilitate the timely discovery of potential problems. In practical applications, the real-time environmental parameters of the equipment are regularly obtained as model inputs. For example, for a main transformer in a substation, the current load rate is 80%, the oil temperature is 55°C, the winding temperature is 65°C, and the model output health status value is 0.85. This value is compared with the preset health status threshold range, such as 0.9-1.0 is good, 0.7-0.9 is normal, 0.5-0.7 is attention, and less than 0.5 is a warning. Based on this, it is judged that the transformer is in a normal state, but close to the attention range, and needs to be monitored more closely. This data-driven health status monitoring method can timely discover potential equipment problems, optimize maintenance strategies, and improve the reliability of power grid operation. By establishing a mathematical model, complex equipment status is quantified into intuitive health indicators, which facilitates managers to make quick decisions. At the same time, the continuously accumulated data also provides strong support for the full life cycle management of equipment, which helps to formulate more accurate renewal and transformation plans.

[0033] S104: When the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid material is marked as predicted waste material.

[0034] Identify the design service life of the current power grid materials, and obtain the actual remaining operating life of the power grid materials based on the difference between the design service life and the operating life of the power grid materials; determine whether the difference between the actual remaining operating life and the estimated remaining life of the materials is greater than a first time threshold; obtain the health status of the power grid materials within the preset period, and determine whether the health status of the power grid materials within the preset period shows a continuous downward trend; if the difference between the actual remaining operating life and the estimated remaining life of the materials is greater than the first time threshold or the health status of the power grid materials within the preset period shows a continuous downward trend, mark the corresponding power grid materials as predicted waste materials.

[0035] Specifically, the management of power grid materials is an important part of ensuring the safe and stable operation of the power system. Identifying the design service life of power grid materials is the basis for evaluating their life status. For example, for a transformer, the corresponding design service life is usually 30 years. This information can be obtained by consulting the equipment archives or the technical specifications provided by the manufacturer. The estimated remaining life is a scientific prediction based on the current status of the material and historical operating data. Taking the transformer as an example, its actual status can be evaluated by methods such as insulating oil analysis and partial discharge detection. Combined with load history, environmental factors, etc., the service life prediction model calculates the estimated remaining life. Assuming that a transformer has been in operation for 25 years and its actual remaining operating life is 5 years, its remaining life is estimated to be 2 years through comprehensive evaluation. The setting of the first time threshold is to identify materials that are about to reach their service life in advance. It can usually be set to 1-2 years. In the above example, the difference between the design service life and the estimated remaining life is 3 years, which is greater than the threshold of 2 years, indicating that the corresponding equipment health status has obvious loss, so the transformer will be marked as an object that needs special attention. Continuous evaluation of health status is crucial to timely detect material performance degradation. Taking distribution transformers as an example, their health status can be assessed by regularly measuring parameters such as insulation resistance and dielectric loss tangent. Assume that in the past 6 months, the insulation resistance of a transformer has dropped from 500MΩ to 300MΩ, and the dielectric loss tangent has increased from 0.5% to 0.8%, showing a trend of continuous deterioration. Marking materials as pre-scrap status helps to make renewal plans in advance. For the above two situations, that is, materials approaching the design service life or with a continuously declining health status, timely marking as pre-scrap can trigger subsequent renewal preparations. This includes formulating renewal plans, applying for budgets, arranging power outages and maintenance, etc., so as to ensure that the replacement is completed before the materials fail and avoid power outages caused by sudden failures. With this method, power companies can achieve active management of the entire life cycle of materials, improve equipment utilization, reduce operation and maintenance costs, and ensure the safe and reliable operation of the power grid. This data-driven decision-making method can not only optimize asset management strategies, but also provide important basis for power grid planning and investment decisions, and ultimately maximize the value of power grid assets.

[0036] S105. For the predicted waste materials, search for information on alternative replacement materials with similar models and specifications, and generate alternative replacement material plans.

[0037] According to the model and specification information of the predicted waste materials, a multi-dimensional material attribute vector is constructed; natural language processing technology is used to perform word segmentation, part-of-speech tagging and semantic analysis on the text description information of power grid materials to extract key attribute information; based on the key attribute information, the materials are grouped through a clustering algorithm, and power grid materials with similar attributes are classified into the same category to form a material classification system; for each predicted waste material, the top N power grid materials with the highest attribute similarity in the same category are searched as potential alternative replacement materials, where N is an integer greater than 1; algorithms such as cosine similarity are used to calculate the similarity between the predicted waste materials and the alternative replacement materials, and several materials with the highest similarity are selected from the alternative replacement material schemes as recommended replacement schemes; according to the importance, procurement cycle and inventory quantity of the power grid materials, the recommended replacement schemes are prioritized to generate the final alternative replacement material scheme.

[0038] Specifically, a multi-dimensional material attribute vector is constructed based on the model and specification information of the predicted waste materials. For example, the model and specification information of a certain power grid material includes voltage level, capacity, material, manufacturer, etc. This information is converted into a numerical vector, such as [220kV, 100MVA, aluminum alloy, A company], for subsequent processing. Natural language processing technology is used to perform word segmentation, part-of-speech tagging and semantic analysis on the text description information of power grid materials to extract key attribute information. Assuming that the text description of a certain material is "220kV transformer, capacity 100MVA, aluminum alloy shell, manufactured by A company", the words "220kV", "transformer", "capacity", "100MVA", "aluminum alloy", "shell", "A company", "manufacturing" and other words are obtained through word segmentation. Then, part-of-speech tagging and semantic analysis are performed to extract key attributes such as voltage level, capacity, material, and manufacturer. Based on the key attribute information, the materials are grouped through the clustering algorithm, and the power grid materials with similar attributes are classified into the same category to form a material classification system. For example, using the K-means clustering algorithm, transformers with similar voltage levels, capacities, and materials are grouped into one category to form the category of "220kV-100MVA-aluminum alloy transformer". For each predicted waste material, search for power grid materials with high attribute similarity in the same category as potential alternative replacement materials. Assuming that a predicted waste material is "220kV-100MVA-aluminum alloy transformer", in the category of "220kV-100MVA-aluminum alloy transformer", search for materials with similar attributes, such as "220kV-105MVA-aluminum alloy transformer", as alternative replacement materials. Using algorithms such as cosine similarity, calculate the similarity between the predicted waste material and the alternative replacement materials, and select several materials with the highest similarity as recommended replacement solutions. For example, calculate the cosine similarity of the attribute vectors of "220kV-100MVA-aluminum alloy transformer" and "220kV-105MVA-aluminum alloy transformer". If the similarity is high, it will be listed as a recommended replacement solution. According to the importance, procurement cycle and inventory quantity of power grid materials, the recommended replacement plans are prioritized to generate the final alternative replacement material plan. Assuming that a certain material is a key equipment with a long procurement cycle and insufficient inventory, the alternative materials with sufficient inventory and short procurement cycle are recommended first. For example, "220kV-105MVA-aluminum alloy transformer" has sufficient inventory, short procurement cycle, high priority, and is listed as the preferred plan. Constructing a multi-dimensional material attribute vector can fully reflect the characteristics of the material and facilitate subsequent analysis and matching. Natural language processing technology extracts key attribute information to improve the accuracy and efficiency of information processing. The clustering algorithm groups materials to form a classification system, which is convenient for quickly locating similar materials. The cosine similarity algorithm quantifies the similarity of materials to ensure the rationality of the recommended plan. Priority sorting comprehensively considers the importance, procurement cycle and inventory quantity to ensure the feasibility and efficiency of the replacement plan.For example, a power grid company needs to replace an old "220kV-100MVA-aluminum alloy transformer". Through the above steps, first construct its attribute vector, then extract key attributes through natural language processing, cluster and group, find "220kV-105MVA-aluminum alloy transformer" and other alternative materials in the same category, calculate the similarity, and find that their similarity is high, so they are listed as recommended solutions. Considering that the transformer is a key equipment, the inventory is insufficient, and the procurement cycle is long, the "220kV-105MVA-aluminum alloy transformer" with sufficient inventory and short procurement cycle is recommended first, and finally a replacement solution is generated. Through this method, the power grid company can quickly and accurately find suitable replacement materials, reduce replacement costs, and improve the efficiency, safety and reliability of power grid operation. Constructing multi-dimensional attribute vectors, natural language processing, clustering grouping, similarity calculation and priority sorting are linked together to form a strict technical chain to ensure the scientificity and practicality of the replacement solution.

[0039] S106. Genetic algorithm and particle swarm optimization algorithm are used to generate an optimal replacement plan that minimizes the full life cycle cost of power grid materials based on alternative replacement material plans.

[0040] For the predicted waste power grid materials, relevant attribute parameters are extracted from the alternative replacement material schemes as the initial population of the optimization algorithm. A genetic algorithm is used to minimize the full life cycle cost as the optimization goal, and the initial population is iteratively optimized through operations such as selection, crossover, and mutation. In each round of iteration, some individuals are eliminated according to the fitness value of the current population, and new individuals are added from the alternative replacement material scheme to maintain population diversity. If the iteration reaches the preset convergence condition or the maximum number of iterations, the current optimal individual is used as the final optimal replacement scheme; otherwise, iterative optimization continues. Initialize a particle swarm based on the alternative replacement material scheme, wherein each particle represents an alternative replacement material scheme, wherein the position of the particle represents various parameters, and the speed represents the trend of parameter changes; update the position and speed of the particle based on the optimal position of the particle. Set the convergence condition or the maximum number of iterations, stop the iteration when the condition is met, and obtain the optimization result; compare the optimal replacement scheme obtained by the genetic algorithm with the optimization result of the particle swarm algorithm, and select the scheme with lower full life cycle cost as the final scheme, that is, the optimal replacement scheme.

[0041] Specifically, a combination of genetic algorithm and particle swarm algorithm can be used to select the optimal replacement scheme for pre-scrapped power grid materials. First, extract relevant attribute parameters from the alternative replacement material scheme as the initial population. These attribute parameters may include material specifications, performance indicators, service life, etc. For example, for a pre-scrapped transformer, its rated capacity, voltage level, loss parameters, etc. can be extracted as the genes of the initial population. The genetic algorithm takes minimizing the full life cycle cost as the optimization goal. The full life cycle cost includes procurement cost, installation cost, operation and maintenance cost, and scrapping cost. Through selection, crossover and mutation operations, the algorithm continuously optimizes the population. In the selection operation, the roulette selection method can be used, and individuals with high fitness are more likely to be selected. The crossover operation can use single-point crossover or multi-point crossover, such as exchanging the capacity and voltage level of two transformer schemes. The mutation operation may randomly change a certain attribute parameter, such as slightly adjusting the loss parameter of the transformer. In each round of iteration, some individuals are eliminated according to the fitness value of the current population. The fitness value can be obtained by calculating the inverse of the full life cycle cost. The lower the cost, the higher the fitness. In order to maintain the diversity of the population, new individuals can be added from the alternative solutions. For example, some transformer solutions with different manufacturers or new technologies can be introduced to increase the breadth of the search space. During the iteration process, it is necessary to set the convergence condition or the maximum number of iterations. The convergence condition can be that the change amplitude of the optimal solution for multiple consecutive generations is less than a certain threshold, such as the change of the life cycle cost does not exceed 0.1%. The maximum number of iterations can be set according to the problem scale and computing resources, such as setting it to 1000 generations. At the same time, the particle swarm algorithm can also be used for optimization. The particle swarm algorithm simulates the intelligent behavior of the group, and each particle represents a possible replacement solution. The position of the particle represents the various parameters of the solution, and the speed represents the change trend of the parameters. By continuously updating the position and speed of the particles, the algorithm gradually converges to the optimal solution. Finally, the optimal replacement solutions obtained by the genetic algorithm and the particle swarm algorithm are compared, and the solution with lower life cycle cost is selected as the final solution. This method of combining multiple algorithms can make full use of the advantages of each algorithm and improve the reliability of the optimization results. For example, the genetic algorithm may be better at handling discrete parameter optimization, while the particle swarm algorithm performs better in the continuous parameter space. By comparing the results of the two algorithms, a more comprehensive and reliable optimal replacement solution can be obtained.

[0042] S107. Record the full-process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full-process data, formulate a power grid material management update plan and perform a visual display.

[0043] Obtain data from all aspects of the entire life cycle of power grid materials and store them in a database. Preprocess the data from each link, clean abnormal data, and convert the data into a format for application modeling. Use a clustering algorithm to classify power grid materials according to material type and operating condition parameters to obtain characteristic parameters of different types of power grid materials. For each type of power grid material, establish the relationship between material usage time and characteristic parameters based on a regression algorithm to obtain an optimal scrapping time prediction model. Input the full process data into the optimal scrapping time prediction model to obtain the optimal scrapping time for each type of power grid material under different operating conditions, and generate a material update plan. Use data visualization technology to display material life cycle management information and material update plans, and update them regularly.

[0044] Specifically, data from all aspects of the entire life cycle of power grid materials are obtained and stored in the database. For example, a power grid company collects production, use, maintenance and scrapping data of materials such as transformers and cables in real time through sensors and IoT devices. These data include the production batch, service life, operating temperature, maintenance times, etc. of the materials. The data is automatically transmitted to the database through the API interface to ensure the integrity and real-time nature of the data. Preprocess the data of each link, clean the abnormal data, and convert the data into a format for application modeling. Taking the transformer as an example, it is found that the operating temperature of a certain record is abnormally high, which may be a data error caused by sensor failure. By setting the temperature threshold, the abnormal data that exceeds the reasonable range is eliminated. After that, the data is standardized, and parameters of different dimensions such as service life (years) and operating temperature (degrees Celsius) are uniformly converted to values ​​between 0 and 1, which is convenient for subsequent modeling and analysis. According to the material type and operating condition parameters, the clustering algorithm is used to classify the power grid materials. Assuming there are 1,000 transformers, they are divided into 5 categories using the K-means clustering algorithm according to the corresponding parameters such as service life, operating temperature, maintenance times, etc. By calculating the center points of each type, the characteristic parameters of each type of transformer are obtained. For example, the average service life of the first type of transformer is 10 years, and the average operating temperature is 60 degrees Celsius. For each type of power grid material, a linear regression model between the material use time and the characteristic parameters is established using a regression algorithm to represent the relationship between the material use time and the characteristic parameters. Taking the first type of transformer as an example, a linear regression model is used to analyze the relationship between the service life, operating temperature, and maintenance times. The model shows that with the increase of service life, the number of maintenance times increases by 1 for every 5 degrees Celsius increase in operating temperature, and the probability of transformer scrapping increases by 10%. The generated full process data is input into the prediction model to obtain the optimal scrapping time of various types of power grid materials under different working conditions. The full process data includes historical data in the material attribute database and operating status data during use. It is assumed that the model predicts that the optimal scrapping time of the first type of transformer is 12 years when the operating temperature exceeds 70 degrees Celsius and the number of maintenance times exceeds 5 times. Based on this prediction, a material update plan is generated, and it is planned to replace 50 transformers close to the scrap standard in the next two years. Data visualization technology is used to display the material life cycle management information and material update plan. Use tools such as Tableau to create dashboards that display the age distribution, operating temperature trends, and maintenance statistics of various transformers. Dynamic charts can be used to visually display which areas of transformers need to be replaced first, facilitating management decision-making. Regular updates are performed to ensure the timeliness and accuracy of data. The latest data is obtained from sensors and IoT devices on a regular basis every month, the database is updated, and the forecast model is re-run to adjust the material update plan. For example, if the operating temperature of a transformer in a certain area is found to be abnormally high in a certain month, the plan can be adjusted in time to replace the transformer in that area in advance to avoid potential failures.Through the above steps, power grid companies can not only accurately predict the scrapping time of materials and optimize material renewal plans, but also improve management efficiency and decision-making quality through data visualization technology. This kind of refined management and prediction based on data helps to reduce operation and maintenance costs, extend the service life of materials, and ensure the safe and stable operation of the power grid.

[0045] Reference Figure 2 , In one embodiment, the present application further provides a material management business supervision device, including: an acquisition module 100, for acquiring corresponding storage parameters for each power grid material entering the warehouse, and constructing a multi-dimensional material attribute database;

[0046] The training module 200 is used to train a material remaining life prediction model based on historical data in the material attribute database using a support vector machine or a random forest algorithm, and input the service life, operating load and failure frequency of power grid materials based on the material remaining life prediction model to output the estimated remaining life of the materials;

[0047] The first collection module 300 is used to collect corresponding operation status data of in-service power grid materials in real time, and judge the current health status of power grid materials in combination with the material health assessment expert knowledge base;

[0048] The marking module 400 is used to mark the corresponding power grid material as a predicted waste material when the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than a first time threshold or the health status shows a continuous downward trend within a preset period;

[0049] The first generation module 500 is used to search for alternative replacement material information with similar models and specifications for predicted waste materials, and generate alternative replacement material solutions;

[0050] The second generation module 600 is used to generate an optimal replacement plan that minimizes the full life cycle cost of power grid materials based on the alternative replacement material plans by using a genetic algorithm and a particle swarm optimization algorithm;

[0051] The optimization module 700 is used to record the full process data of each power grid material, continuously optimize the full life cycle management strategy of the power grid material, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and perform a visual display.

[0052] Reference Figure 3 In an embodiment of the present application, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, an internal memory, a storage medium (non-volatile storage medium) and a network interface connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes the above-mentioned storage medium (non-volatile storage medium) and the internal memory. The storage medium (non-volatile storage medium) stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the storage medium (non-volatile storage medium). The database of the computer device is used to store usage data, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. Furthermore, the above-mentioned computer device can also be provided with an input device and a display screen, etc. When the above computer program is executed by a processor, a material management business supervision method is implemented, including the following steps: for each piece of grid material entering the warehouse, the corresponding storage parameters are obtained, and a multi-dimensional material attribute database is constructed; based on the historical data in the material attribute database, a support vector machine or a random forest algorithm is used to train a material remaining life prediction model, and based on the material remaining life prediction model, the service life, operating load and failure frequency of the grid material are input, and the estimated remaining life of the material is output; for the grid materials in service, the corresponding operating status data is collected in real time, and the current health status of the grid materials is judged in combination with the material health assessment expert knowledge base; when the actual remaining operating life of the grid material is different from the When the difference between the estimated remaining life of materials is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid materials will be marked as predicted waste materials; for predicted waste materials, search for information on alternative replacement materials with similar models and specifications, and generate alternative replacement material plans; use genetic algorithms and particle swarm optimization algorithms to generate the optimal replacement plan that minimizes the full life cycle cost of power grid materials based on the alternative replacement material plans; record the full process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, and predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and display it visually. Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0053] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, a visual early warning analysis method for power equipment damage is implemented, including the following steps: for each piece of grid material entering the warehouse, the corresponding storage parameters are obtained, and a multi-dimensional material attribute database is constructed; based on the historical data in the material attribute database, a support vector machine or a random forest algorithm is used to train a material remaining life prediction model, and based on the material remaining life prediction model, the service life, operating load and failure frequency of the grid material are input, and the estimated remaining life of the material is output; for the grid materials in service, the corresponding operating status data is collected in real time, and the current health status of the grid material is judged in combination with the material health assessment expert knowledge base. ; When the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid material is marked as predicted waste material; for the predicted waste material, search for alternative replacement material information with similar models and specifications to generate alternative replacement material solutions; use genetic algorithms and particle swarm optimization algorithms to generate the optimal replacement solution that minimizes the full life cycle cost of power grid materials based on the alternative replacement material solutions; record the full process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and display it visually. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0054] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0055] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device system, article or method 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, device system, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, device system, article or method including the element.

[0056] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A material management business supervision method, characterized in that: The method comprises: For each piece of grid material entering the warehouse, obtain the corresponding entry parameters and build a multi-dimensional material attribute database; Based on the historical data in the material attribute database, a support vector machine or a random forest algorithm is used to train a material remaining life prediction model, and the material service life, operating load and failure frequency of the in-service power grid materials are input into the material remaining life prediction model to output the estimated remaining life of the materials; For in-service power grid materials, the corresponding operating status data is collected in real time, and the current health status of power grid materials is judged in combination with the material health assessment expert knowledge base; When the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, the corresponding power grid material is marked as a predicted waste material; For predicted waste materials, search for alternative replacement materials with similar models and specifications, and generate alternative replacement material plans; Genetic algorithm and particle swarm optimization algorithm are used to generate the optimal replacement plan that minimizes the life cycle cost of power grid materials based on alternative replacement material plans; Record the full process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and perform visual display; The method adopts the genetic algorithm and the particle swarm optimization algorithm to generate the optimal replacement scheme that minimizes the full life cycle cost of power grid materials based on the alternative replacement material scheme, including: For the predicted waste power grid materials, relevant attribute parameters are extracted from the alternative replacement material solutions as the initial population of the optimization algorithm; A genetic algorithm is used to iteratively optimize the initial population by selecting, crossing over and varying the population, with the goal of minimizing the total life cycle cost; In each round of iteration, some individuals are eliminated according to the fitness value of the current population, and new individuals are added from the alternative replacement material plan to maintain population diversity; If the iteration reaches the preset convergence condition or the maximum number of iterations, the current optimal individual is used as the best replacement solution; Otherwise, continue iterative optimization; Initializing a particle swarm based on an alternative material replacement scheme, wherein each particle represents an alternative material replacement scheme, wherein the position of the particle represents various parameters, and the speed represents a parameter change trend; Update the particle's position and velocity based on the particle's best position; Set the convergence condition or the maximum number of iterations, stop the iteration when the condition is met, and get the optimization result; The best replacement solution obtained by the genetic algorithm is compared with the optimization result of the particle swarm algorithm, and the solution with lower life cycle cost is selected as the final solution, that is, the optimal replacement solution; The full process data of each power grid material is recorded, the full life cycle management strategy of the power grid material is continuously optimized, the optimal scrapping time of different types of power grid materials under various usage conditions is predicted based on the full process data, and the power grid material management update plan is formulated and visualized, including: Obtain data from all aspects of the entire life cycle of power grid materials and store them in the database; Preprocess the data at each stage, clean abnormal data, and convert the data into a format suitable for application modeling; According to the material type and working condition parameters, the clustering algorithm is used to classify the power grid materials and obtain the characteristic parameters of different types of power grid materials; For each type of power grid material, the relationship between material usage time and characteristic parameters is established based on the regression algorithm to obtain the optimal scrap time prediction model; Input the full process data into the optimal scrapping time prediction model to obtain the optimal scrapping time of various types of power grid materials under different working conditions, and generate a material update plan; Data visualization technology is used to display the full life cycle management information and material update plan of materials, and updates them regularly.

2. The method according to claim 1, characterized in that The method uses a support vector machine or a random forest algorithm to train a material remaining life prediction model based on historical data in a material attribute database, and inputs the material service life, operating load and failure frequency of in-service power grid materials into the material remaining life prediction model to output the estimated remaining life of the materials, including: The material attribute data obtained from the material attribute database is used as historical data; Based on historical data, training data is formed. The service life, operating load, and fault frequency data of power grid materials contained in the training data are normalized to obtain values ​​A, B, and C. The material operation ratio D is calculated based on the service life of the materials and the material operation time. Use random forest or support vector machine to build the model, and use grid search based on values ​​A, B, C and D to find the best value of random forest algorithm or support vector machine algorithm hyperparameter to obtain model parameter configuration data; Based on the training data including the service life of materials, operating load, and failure frequency and the model parameter configuration data, it is determined whether the amount of training data is greater than a preset quantity threshold. If it is greater than the preset quantity threshold, machine learning training is performed; If the amount of training data is less than or equal to a preset amount threshold, collect the external database as new material attribute data and add it to the material attribute database, and update the material attribute database by collecting the new material attribute data; After the initial prediction model is obtained through machine learning training, the validation data set is input for verification to determine whether the average error is higher than the preset error threshold; If the average error is higher than the preset error threshold, the initial prediction model is combined with the K nearest neighbor algorithm to obtain the material remaining life prediction model; If the average error is lower than or equal to the preset error threshold, the initial prediction model is determined as the material remaining life prediction model; The material remaining life prediction model is used to predict the material life of the in-service power grid materials based on the input material service life, operating load and failure frequency of the in-service power grid materials to obtain the predicted value, and the predicted value is denormalized to obtain the estimated remaining life of the material.

3. The method according to claim 1, characterized in that The above-mentioned real-time collection of corresponding operating status data of in-service power grid materials and the determination of the current health status of power grid materials in combination with the material health assessment expert knowledge base include: Deploy sensors to on-site equipment based on the types of power grid equipment in service to obtain equipment operating status data at different times; Determine whether the acquisition frequency of the acquired operating status data reaches a preset threshold, and if not, adjust the acquisition frequency of the sensor until it reaches the preset threshold; According to the damage degree of each device in the operation status data, determine whether the preset damage threshold is reached; If the preset damage threshold is reached, the principal component analysis algorithm is used to reduce the dimension of the equipment operation status data, extract the main features in the operation status data, and calculate the weight of each principal component on the equipment performance based on the pre-established equipment performance indicators; The historical operation status data and impact weights are used as samples, combined with the health status labels in the material health assessment expert knowledge base, and trained using a multivariate linear regression model to establish a health status regression model. At a certain point in time, the current operating status data of the device is obtained as the input of the health status regression model, and the health status value of the device at that point in time is calculated by the health status regression model; The health status of the current device, that is, the health status of the power grid material, is determined according to a preset health status threshold range.

4. The method according to claim 1, characterized in that When the difference between the actual remaining operating life of the power grid material and the estimated remaining life of the material is greater than the first time threshold or the health status shows a continuous downward trend within a preset period, marking the corresponding power grid material as a predicted waste material includes: Identify the design service life of the current power grid materials, and obtain the actual remaining service life of the power grid materials based on the difference between the design service life and the service life of the power grid materials; Determining whether the difference between the actual remaining operating life and the estimated remaining life of the material is greater than a first time threshold; Obtaining the health status of the power grid materials within the preset period, and determining that the health status of the power grid materials within the preset period shows a continuous downward trend; If the difference between the actual remaining operating life and the estimated remaining life of the material is greater than the first time threshold or the health status of the power grid material shows a continuous downward trend during the preset period, the corresponding power grid material is marked as predicted waste material.

5. The method according to claim 1, characterized in that The method of searching for information on alternative replacement materials with similar models and specifications for predicted waste materials and generating alternative replacement material solutions includes: Construct a multi-dimensional material attribute vector based on the model and specification information of the predicted waste materials; Use natural language processing technology to perform word segmentation, part-of-speech tagging and semantic analysis on the text description information of power grid materials to extract key attribute information; Based on key attribute information, materials are grouped through clustering algorithms, and power grid materials with similar attributes are classified into the same category to form a material classification system; For each predicted waste material, search for the top N power grid materials with the highest attribute similarity in the same category as potential alternative replacement materials, where N is an integer greater than 1; The cosine similarity algorithm is used to calculate the similarity between the predicted waste materials and the alternative replacement materials, and several materials with the highest similarity are selected from the alternative replacement materials as recommended replacement solutions; According to the importance, procurement cycle and inventory quantity of power grid materials, the recommended replacement plans are prioritized to generate the final alternative replacement material plan.

6. A material management business supervision device, used to execute the method according to any one of claims 1 to 5, characterized in that: The device comprises: an acquisition module for acquiring corresponding storage parameters for each piece of power grid material stored in the warehouse and constructing a multi-dimensional material attribute database; A training module is used to train a material remaining life prediction model based on historical data in a material attribute database by using a support vector machine or a random forest algorithm, and input the material service life, operating load and failure frequency of in-service power grid materials based on the material remaining life prediction model to output the estimated remaining life of the materials; The first collection module is used to collect the corresponding operating status data of the in-service power grid materials in real time, and judge the current health status of the power grid materials in combination with the material health assessment expert knowledge base; the marking module is used to mark the corresponding power grid materials as predicted waste materials when the difference between the actual remaining operating life of the power grid materials and the estimated remaining life of the materials is greater than the first time threshold or the health status shows a continuous downward trend within a preset period; The first generation module is used to search for alternative replacement material information with similar models and specifications for predicted waste materials, and generate alternative replacement material plans; The second generation module is used to generate an optimal replacement plan that minimizes the full life cycle cost of power grid materials based on alternative replacement material plans by using a genetic algorithm and a particle swarm optimization algorithm; The optimization module is used to record the full process data of each power grid material, continuously optimize the full life cycle management strategy of power grid materials, predict the optimal scrapping time of different types of power grid materials under various usage conditions based on the full process data, formulate a power grid material management update plan and perform a visual display.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images