Energy router life prediction method based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent power grids, in particular to an energy router service life prediction method based on artificial intelligence, which realizes accurate prediction of the residual service life of an energy router by integrating and analyzing data information from different sources and combining an advanced artificial intelligence algorithm. The implementation of the invention not only improves the operation and maintenance efficiency and reliability of the power grid system, but also provides powerful support for the intelligent upgrading of the smart power grid.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to an energy router life prediction method based on artificial intelligence. Background Art

[0002] As the core equipment of smart grid, the stable operation of energy router is crucial to ensure the safety of the grid. However, most traditional energy router life prediction methods rely on a single data source and simple statistical models, which are difficult to fully reflect the real state of the equipment in a complex operating environment, resulting in inaccurate prediction results. With the continuous development of artificial intelligence technology and the increasing maturity of multi-source data fusion technology, new solutions are provided for energy router life prediction. Summary of the invention

[0003] The purpose of the present invention is to provide an energy router life prediction method based on artificial intelligence to solve the problems raised in the background technology and facilitate promotion.

[0004] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0005] An energy router life prediction method based on artificial intelligence mainly includes the following steps:

[0006] Step 1: Multi-source data collection and integration: collect data from multiple sources including the energy router’s monitoring system, environmental monitoring system, and historical maintenance records, and pre-process the collected multi-source data to ensure data quality and consistency;

[0007] Step 2: Feature extraction and fusion. Using feature extraction technology in artificial intelligence, key features related to the life of the energy router are extracted from the preprocessed data. Multi-source data fusion technology is used to integrate feature information from different sources to form a more comprehensive feature set to more accurately reflect the actual status of the equipment.

[0008] Step 3: Build an intelligent prediction model, select a machine learning or deep learning algorithm, build an energy router life prediction model, use the fused feature set as input, train and optimize the model, so that it can learn the complex mapping relationship between features and life;

[0009] Step 4: Model verification and tuning: verify and tune the prediction model through cross-validation, grid search and hyperparameter optimization to improve the prediction performance and generalization ability of the model;

[0010] Step 5: Life prediction and decision support: Apply the trained prediction model to the energy routers in actual operation to perform real-time life prediction and provide decision support for operation and maintenance personnel based on the prediction results.

[0011] Furthermore, the multi-source data collection and integration includes the following steps:

[0012] Step 1: Multi-source data collection, which includes data source selection and data collection method. The data source selection obtains real-time operation data from the monitoring system of the energy router, including current, voltage and power factor, obtains temperature and humidity parameters from the environmental monitoring system, and obtains fault records and repair records from historical maintenance records. The data collection method uses sensors and monitoring equipment to automatically collect data to ensure the real-time and accuracy of the data, and extracts historical data from the database for model training and verification;

[0013] Step 2: Data integration, the data integration includes data cleaning, data format unification, data fusion and data quality control. The data cleaning removes duplicate data and invalid data to ensure the uniqueness and validity of the data and corrects the erroneous data. The data format unification converts the data from different sources into a unified format for subsequent processing and analysis, and standardizes or normalizes the data to ensure the comparability of the data in different dimensions. The data fusion fuses the data from different sources to form a more comprehensive data set, and uses multi-source data fusion technology to extract key features related to the life of the energy router. The data quality control performs quality checks on the integrated data to ensure the accuracy, completeness and consistency of the data.

[0014] Furthermore, the feature extraction and fusion includes the following steps:

[0015] Step 1: feature extraction, the feature extraction includes preliminary screening, feature selection and feature conversion, the preliminary screening selects features related to the life of the energy router, the feature selection uses a feature selection algorithm to select the most representative and most helpful features for prediction from the features initially screened, and the feature conversion converts the selected features to extract deeper information;

[0016] Step 2: Feature fusion, the feature fusion includes data alignment, feature concatenation, feature weighting and feature dimensionality reduction. The data alignment ensures that the features from different sources are aligned in time or space for subsequent processing. The feature concatenation concatenates different features to form a new feature vector. The feature weighting is based on the machine learning algorithm or deep learning algorithm. Different weights are assigned to different features according to their importance to the prediction results. The feature dimensionality reduction is to prevent the feature vector from being too large, which leads to increased computational complexity and overfitting problems. Therefore, feature dimensionality reduction is performed to reduce the number of features and retain the most important information.

[0017] Furthermore, the construction of the intelligent prediction model includes the following steps:

[0018] Step 1: Select a machine learning or deep learning algorithm. Select an appropriate machine learning or deep learning algorithm to build a prediction model. The algorithms include support vector machine (SVM), random forest (RF) and neural network (NN). When selecting an algorithm, you need to consider the characteristics of the data, the accuracy of the prediction, the computational complexity and the interpretability of the algorithm.

[0019] Step 2: Model architecture design. Design the model architecture based on the selected algorithm. For deep learning algorithms such as neural networks, it is necessary to determine the number of network layers, the number of neurons in each layer, and the activation function. For traditional machine learning algorithms such as support vector machines, it is necessary to determine the kernel function and regularization parameters.

[0020] Step 3: Model training and optimization. Use the preprocessed data and extracted features to train the model. During the training process, the model parameters need to be continuously adjusted to minimize the prediction error. This is achieved through gradient descent and stochastic gradient descent. At the same time, cross-validation is also required to evaluate the performance of the model and prevent overfitting.

[0021] Step 4: Model verification and evaluation. After the model training is completed, an independent test data set is needed to verify the performance of the model.

[0022] Step 5: Model deployment and integration. After verification and evaluation, the model is deployed to the actual production environment. The model needs to be integrated into the existing monitoring system to obtain data in real time and make predictions. At the same time, the model update and maintenance issues need to be considered to ensure that the model can continue to provide accurate prediction results.

[0023] Step 6: Model monitoring and tuning. During the actual operation, the performance of the model is continuously monitored and optimized according to the actual situation. This is achieved by collecting new data, updating model parameters, and adjusting feature selection. At the same time, attention should also be paid to the stability and reliability of the model to ensure its long-term stable operation.

[0024] Furthermore, the model verification and tuning includes the following steps:

[0025] Step 1: Model validation, which includes cross-validation, performance index evaluation, overfitting and underfitting check and feature importance evaluation. The cross-validation is used to evaluate the generalization ability of the model. The data set is divided into multiple subsets, one of which is used as the test set and the other subsets are used as the training set in turn. Multiple training and testing are performed to evaluate the performance of the model on different data sets. The performance index evaluation is used to evaluate the performance of the model, including prediction accuracy, recall rate, F1 score and mean square error (MSE). The feature importance evaluation is used to evaluate the influence of different features on the model prediction results, which helps to further optimize feature selection and feature engineering.

[0026] Step 2: Model tuning, the model tuning includes parameter adjustment, feature selection and transformation, integration method, grid search and random search and feedback loop. The parameter adjustment adjusts the parameters of the model according to the results of model verification, including regularization strength, learning rate and batch size, to improve the performance of the model. The feature selection and transformation selects features with strong correlation with the target variable according to the results of feature importance evaluation, or transforms the features to reduce variance and improve the stability of the model. The integration method is used to improve the generalization ability and robustness of the model, and multiple prediction models are combined to obtain the final prediction result by voting or weighted averaging. The grid search and random search are used to find the optimal parameter combination, which can systematically search for the optimal solution in a given parameter space, so as to find the parameter combination that makes the model perform best. The feedback loop continuously optimizes the prediction performance by continuously collecting new data, updating model parameters and feature selection, and retraining the model, which helps to ensure that the model can adapt to the changing environment and data distribution.

[0027] Furthermore, the lifespan prediction and decision-making includes the following steps:

[0028] Step 1: Life prediction, which includes selecting a prediction method, prediction accuracy, and application of prediction results. The selected prediction method predicts the life of the energy router based on multi-source data fusion and artificial intelligence algorithms. The prediction accuracy verifies the accuracy of the prediction model through cross-validation and performance indicator evaluation methods to ensure that the prediction results can truly reflect the life status of the energy router and provide a reliable basis for subsequent decision support. The prediction results are used to guide the maintenance plan formulation, spare parts inventory management, and risk assessment of the energy router. By predicting the life of the energy router in advance, maintenance time and resources can be reasonably arranged to reduce failure rate and maintenance costs;

[0029] Step 2: Decision support, the decision support includes decision requirements, decision models, decision optimization and decision implementation. The decision requirements determine the maintenance strategy, spare parts procurement plan and risk assessment decision requirements based on the life prediction results of the energy router, aiming to ensure the stable operation of the energy router and reduce the risk of failure and maintenance costs. The decision model constructs a decision model based on the prediction results, and automatically recommends the optimal decision plan based on the prediction results and actual needs. The decision optimization optimizes the decision plan by continuously collecting new data, updating the prediction model and decision model, and considering the uncertainty and risks in the actual operation. The decision implementation puts the optimized decision plan into practice.

[0030] As an improvement, the beneficial effects of the present invention are:

[0031] 1. Improve prediction accuracy: By integrating multi-source data and extracting key features, combined with advanced artificial intelligence algorithms, the accuracy of energy router life prediction is significantly improved.

[0032] 2. Enhanced decision support capabilities: Provides decision support based on prediction results for operation and maintenance personnel, which helps to formulate more reasonable maintenance plans and operation strategies and reduce the risk of equipment failure.

[0033] 3. Improve operation and maintenance efficiency: Through automated and intelligent prediction and decision-making processes, manual intervention and repetitive work are reduced, and operation and maintenance efficiency is improved.

[0034] 4. Promote intelligent upgrading of power grids: The implementation of the present invention helps to promote the intelligent upgrading of smart grids and improve the overall efficiency and reliability of power grid systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a general flow chart of an energy router life prediction method based on artificial intelligence of the present invention;

[0036] Figure 2 A flow chart of an energy router life prediction method A based on artificial intelligence of the present invention;

[0037] Figure 3 This is a flow chart of an energy router life prediction method B based on artificial intelligence of the present invention;

[0038] Figure 4 A flow chart of an energy router life prediction method C based on artificial intelligence of the present invention;

[0039] Figure 5 A flow chart of a method D for predicting the life of an energy router based on artificial intelligence according to the present invention;

[0040] Figure 6This is a flow chart of an energy router life prediction method E based on artificial intelligence of the present invention; DETAILED DESCRIPTION

[0041] In order to make the contents of the present invention more clearly understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0042] An energy router life prediction method based on artificial intelligence mainly includes the following steps:

[0043] Step 1: Multi-source data collection and integration: collect data from multiple sources including the energy router’s monitoring system, environmental monitoring system, and historical maintenance records, and pre-process the collected multi-source data to ensure data quality and consistency;

[0044] Step 2: Feature extraction and fusion. Using feature extraction technology in artificial intelligence, key features related to the life of the energy router are extracted from the preprocessed data. Multi-source data fusion technology is used to integrate feature information from different sources to form a more comprehensive feature set to more accurately reflect the actual status of the equipment.

[0045] Step 3: Build an intelligent prediction model, select a machine learning or deep learning algorithm, build an energy router life prediction model, use the fused feature set as input, train and optimize the model, so that it can learn the complex mapping relationship between features and life;

[0046] Step 4: Model verification and tuning: verify and tune the prediction model through cross-validation, grid search and hyperparameter optimization to improve the prediction performance and generalization ability of the model;

[0047] Step 5: Life prediction and decision support: Apply the trained prediction model to the energy routers in actual operation to perform real-time life prediction and provide decision support for operation and maintenance personnel based on the prediction results.

[0048] In this embodiment, multi-source data collection and integration includes the following steps:

[0049] Step 1: Multi-source data collection. Multi-source data collection includes data source selection and data collection methods. Data source selection obtains real-time operation data from the monitoring system of the energy router, including current, voltage and power factor, obtains temperature and humidity parameters from the environmental monitoring system, and obtains fault records and repair records from historical maintenance records. The data collection method uses sensors and monitoring equipment to automatically collect data to ensure the real-time and accuracy of the data. At the same time, historical data is extracted from the database for model training and verification;

[0050] Step 2: Data integration. Data integration includes data cleaning, data format unification, data fusion and data quality control. Data cleaning removes duplicate data and invalid data to ensure the uniqueness and validity of the data, and corrects erroneous data. Data format unification converts data from different sources into a unified format for subsequent processing and analysis. The data is standardized or normalized to ensure the comparability of the data in different dimensions. Data fusion fuses data from different sources to form a more comprehensive data set. Multi-source data fusion technology is used to extract key features related to the life of the energy router. Data quality control performs quality checks on the integrated data to ensure the accuracy, completeness and consistency of the data.

[0051] Through the above-mentioned technical solution, through multi-source data collection and integration, the present invention can obtain a more comprehensive and accurate data set, providing strong support for subsequent feature extraction, model training and life prediction.

[0052] In this embodiment, feature extraction and fusion include the following steps:

[0053] Step 1: Feature extraction. Feature extraction includes preliminary screening, feature selection and feature conversion. The preliminary screening selects the features related to the life of the energy router. The feature selection uses the feature selection algorithm to select the most representative and most helpful features for prediction from the preliminary screened features. The feature conversion converts the selected features to extract deeper information.

[0054] Step 2: Feature fusion. Feature fusion includes data alignment, feature concatenation, feature weighting and feature dimensionality reduction. Data alignment ensures that features from different sources are aligned in time or space for subsequent processing. Feature concatenation concatenates different features to form a new feature vector. Feature weighting is based on the machine learning algorithm or deep learning algorithm. Different features are assigned different weights based on their importance to the prediction results. Feature dimensionality reduction is performed to prevent the feature vector from being too large, which will increase the computational complexity and overfitting problems. Therefore, feature dimensionality reduction is performed to reduce the number of features and retain the most important information.

[0055] Through the above technical solution, through feature extraction and fusion, the present invention can extract more comprehensive and accurate key features from multi-source data, providing strong support for subsequent intelligent prediction and decision-making. These features not only reflect the current state of the device, but also contain historical information of the device and environmental conditions and other information, thereby improving the accuracy and reliability of the prediction.

[0056] In this embodiment, building an intelligent prediction model includes the following steps:

[0057] Step 1: Select a machine learning or deep learning algorithm. Select an appropriate machine learning or deep learning algorithm to build a prediction model. The algorithms include support vector machine (SVM), random forest (RF) and neural network (NN). When selecting an algorithm, you need to consider the characteristics of the data, the accuracy of the prediction, the computational complexity and the interpretability of the algorithm.

[0058] Step 2: Model architecture design. Design the model architecture based on the selected algorithm. For deep learning algorithms such as neural networks, it is necessary to determine the number of network layers, the number of neurons in each layer, and the activation function. For traditional machine learning algorithms such as support vector machines, it is necessary to determine the kernel function and regularization parameters.

[0059] Step 3: Model training and optimization. Use the preprocessed data and extracted features to train the model. During the training process, the model parameters need to be continuously adjusted to minimize the prediction error. This is achieved through gradient descent and stochastic gradient descent. At the same time, cross-validation is also required to evaluate the performance of the model and prevent overfitting.

[0060] Step 4: Model verification and evaluation. After the model training is completed, an independent test data set is needed to verify the performance of the model.

[0061] Step 5: Model deployment and integration. After verification and evaluation, the model is deployed to the actual production environment. The model needs to be integrated into the existing monitoring system to obtain data in real time and make predictions. At the same time, the model update and maintenance issues need to be considered to ensure that the model can continue to provide accurate prediction results.

[0062] Step 6: Model monitoring and tuning. During the actual operation, the performance of the model is continuously monitored and optimized according to the actual situation. This is achieved by collecting new data, updating model parameters, and adjusting feature selection. At the same time, attention should also be paid to the stability and reliability of the model to ensure its long-term stable operation.

[0063] Through the above technical solutions, an accurate, stable and reliable prediction model can be constructed to provide strong support for the life prediction of energy routers.

[0064] In this embodiment, model verification and tuning include the following steps:

[0065] Step 1: Model validation. Model validation includes cross-validation, performance indicator evaluation, overfitting and underfitting check, and feature importance evaluation. Cross-validation is used to evaluate the generalization ability of the model. The data set is divided into multiple subsets, and one of the subsets is used as the test set and the remaining subsets are used as the training set in turn. Multiple training and testing are performed to evaluate the performance of the model on different data sets. Performance indicator evaluation is used to evaluate the performance of the model, including prediction accuracy, recall rate, F1 score, and mean square error (MSE). Feature importance evaluation is used to evaluate the influence of different features on the model prediction results, which helps to further optimize feature selection and feature engineering.

[0066] Step 2: Model tuning. Model tuning includes parameter adjustment, feature selection and transformation, integration method, grid search and random search, and feedback loop. Parameter adjustment adjusts the model parameters, including regularization strength, learning rate, and batch size, according to the results of model verification to improve the performance of the model. Feature selection and transformation select features with strong correlation with the target variable according to the results of feature importance assessment, or transform the features to reduce variance and improve the stability of the model. Integration method is used to improve the generalization ability and robustness of the model. Multiple prediction models are combined to obtain the final prediction result by voting or weighted averaging. Grid search and random search are used to find the optimal parameter combination. The optimal solution can be systematically searched in a given parameter space to find the parameter combination that makes the model perform best. The feedback loop continuously optimizes the prediction performance by continuously collecting new data, updating model parameters and feature selection, and retraining the model, which helps ensure that the model can adapt to the changing environment and data distribution.

[0067] Through the above technical solutions, the performance of the model is verified through cross-validation, performance index evaluation, overfitting and underfitting check, and feature importance evaluation; and the performance of the model is optimized through parameter adjustment, feature selection and transformation, integration method, grid search and random search, and feedback loop. The implementation of these steps will help improve the accuracy and reliability of energy router life prediction.

[0068] In this embodiment, life prediction and decision-making include the following steps:

[0069] Step 1: Life prediction. Life prediction includes the selection of prediction methods, prediction accuracy and application of prediction results. The prediction method is selected based on multi-source data fusion and artificial intelligence algorithm to predict the life of the energy router. The prediction accuracy is verified by cross-validation and performance indicator evaluation methods to ensure that the prediction results can truly reflect the life status of the energy router and provide a reliable basis for subsequent decision support. The prediction results are used to guide the maintenance plan formulation, spare parts inventory management and risk assessment of the energy router. By predicting the life of the energy router in advance, maintenance time and resources can be reasonably arranged to reduce failure rate and maintenance costs.

[0070] Step 2: Decision support. Decision support includes decision requirements, decision models, decision optimization and decision implementation. Decision requirements determine maintenance strategies, spare parts procurement plans and risk assessment decision requirements based on the life prediction results of the energy router, aiming to ensure the stable operation of the energy router and reduce failure risks and maintenance costs. The decision model builds a decision model based on the prediction results. According to the prediction results and actual needs, it automatically recommends the optimal decision plan. Decision optimization optimizes the decision plan by continuously collecting new data, updating prediction models and decision models, and considering the uncertainties and risks in actual operation. Decision implementation puts the optimized decision plan into practice.

[0071] Through the above technical solutions, accurate life prediction and scientific decision support, the stable operation of the energy router can be ensured, the risk of failure and maintenance costs can be reduced, and the overall operational efficiency can be improved.

[0072] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the life of an energy router based on artificial intelligence, which mainly includes the following steps: Step 1: Multi-source data collection and integration: collect data from multiple sources including the energy router’s monitoring system, environmental monitoring system, and historical maintenance records, and pre-process the collected multi-source data to ensure data quality and consistency; Step 2: Feature extraction and fusion. Using feature extraction technology in artificial intelligence, key features related to the life of the energy router are extracted from the preprocessed data. Multi-source data fusion technology is used to integrate feature information from different sources to form a more comprehensive feature set to more accurately reflect the actual status of the equipment. Step 3: Build an intelligent prediction model, select a machine learning or deep learning algorithm, build an energy router life prediction model, use the fused feature set as input, train and optimize the model, so that it can learn the complex mapping relationship between features and life; Step 4: Model verification and tuning: verify and tune the prediction model through cross-validation, grid search and hyperparameter optimization to improve the prediction performance and generalization ability of the model; Step 5: Life prediction and decision support: Apply the trained prediction model to the energy routers in actual operation to perform real-time life prediction and provide decision support for operation and maintenance personnel based on the prediction results.

2. According to the artificial intelligence-based energy router life prediction method of claim 1, it is characterized in that: The multi-source data collection and integration includes the following steps: Step 1: Multi-source data collection, which includes data source selection and data collection method. The data source selection obtains real-time operation data from the monitoring system of the energy router, including current, voltage and power factor, obtains temperature and humidity parameters from the environmental monitoring system, and obtains fault records and repair records from historical maintenance records. The data collection method uses sensors and monitoring equipment to automatically collect data to ensure the real-time and accuracy of the data, and extracts historical data from the database for model training and verification; Step 2: Data integration. The data integration includes data cleaning, data format unification, data fusion and data quality control. The data cleaning removes duplicate data and invalid data to ensure the uniqueness and validity of the data and corrects erroneous data. The data format unification converts data from different sources into a unified format for subsequent processing and analysis. The data is standardized or normalized to ensure the comparability of the data in different dimensions. The data fusion fuses data from different sources to form a more comprehensive data set. Multi-source data fusion technology is used to extract key features related to the life of the energy router. The data quality control performs quality checks on the integrated data to ensure the accuracy, completeness and consistency of the data.

3. According to the artificial intelligence-based energy router life prediction method of claim 1, it is characterized in that: The feature extraction and fusion comprises the following steps: Step 1: feature extraction, the feature extraction includes preliminary screening, feature selection and feature conversion, the preliminary screening selects features related to the life of the energy router, the feature selection uses a feature selection algorithm to select the most representative and most helpful features for prediction from the features initially screened, and the feature conversion converts the selected features to extract deeper information; Step 2: Feature fusion, the feature fusion includes data alignment, feature concatenation, feature weighting and feature dimensionality reduction. The data alignment ensures that the features from different sources are aligned in time or space for subsequent processing. The feature concatenation concatenates different features to form a new feature vector. The feature weighting is based on the machine learning algorithm or deep learning algorithm. Different weights are assigned to different features according to their importance to the prediction results. The feature dimensionality reduction is to prevent the feature vector from being too large, which leads to increased computational complexity and overfitting problems. Therefore, feature dimensionality reduction is performed to reduce the number of features and retain the most important information.

4. The method for predicting the life of an energy router based on artificial intelligence according to claim 1 is characterized in that: The construction of the intelligent prediction model comprises the following steps: Step 1: Select a machine learning or deep learning algorithm. Select an appropriate machine learning or deep learning algorithm to build a prediction model. The algorithms include support vector machine (SVM), random forest (RF) and neural network (NN). When selecting an algorithm, you need to consider the characteristics of the data, the accuracy of the prediction, the computational complexity and the interpretability of the algorithm. Step 2: Model architecture design. Design the model architecture based on the selected algorithm. For deep learning algorithms such as neural networks, it is necessary to determine the number of network layers, the number of neurons in each layer, and the activation function. For traditional machine learning algorithms such as support vector machines, it is necessary to determine the kernel function and regularization parameters. Step 3: Model training and optimization. Use the preprocessed data and extracted features to train the model. During the training process, the model parameters need to be continuously adjusted to minimize the prediction error. This is achieved through gradient descent and stochastic gradient descent. At the same time, cross-validation is also required to evaluate the performance of the model and prevent overfitting. Step 4: Model verification and evaluation. After the model training is completed, an independent test data set is needed to verify the performance of the model. Step 5: Model deployment and integration. After verification and evaluation, the model is deployed to the actual production environment. The model needs to be integrated into the existing monitoring system to obtain data in real time and make predictions. At the same time, the model update and maintenance issues need to be considered to ensure that the model can continue to provide accurate prediction results. Step 6: Model monitoring and tuning. During the actual operation, the performance of the model is continuously monitored and optimized according to the actual situation. This is achieved by collecting new data, updating model parameters, and adjusting feature selection. At the same time, attention should also be paid to the stability and reliability of the model to ensure its long-term stable operation.

5. The method for predicting the life of an energy router based on artificial intelligence according to claim 1 is characterized in that: The model verification and tuning includes the following steps: Step 1: Model validation, which includes cross-validation, performance index evaluation, overfitting and underfitting check and feature importance evaluation. The cross-validation is used to evaluate the generalization ability of the model. The data set is divided into multiple subsets, one of which is used as the test set and the other subsets are used as the training set in turn. Multiple training and testing are performed to evaluate the performance of the model on different data sets. The performance index evaluation is used to evaluate the performance of the model, including prediction accuracy, recall rate, F1 score and mean square error (MSE). The feature importance evaluation is used to evaluate the influence of different features on the model prediction results, which helps to further optimize feature selection and feature engineering. Step 2: Model tuning, the model tuning includes parameter adjustment, feature selection and transformation, integration method, grid search and random search and feedback loop. The parameter adjustment adjusts the parameters of the model according to the results of model verification, including regularization strength, learning rate and batch size, to improve the performance of the model. The feature selection and transformation selects features with strong correlation with the target variable according to the results of feature importance evaluation, or transforms the features to reduce variance and improve the stability of the model. The integration method is used to improve the generalization ability and robustness of the model, and multiple prediction models are combined to obtain the final prediction result by voting or weighted averaging. The grid search and random search are used to find the optimal parameter combination, which can systematically search for the optimal solution in a given parameter space, so as to find the parameter combination that makes the model perform best. The feedback loop continuously optimizes the prediction performance by continuously collecting new data, updating model parameters and feature selection, and retraining the model, which helps to ensure that the model can adapt to the changing environment and data distribution.

6. The method for predicting the life of an energy router based on artificial intelligence according to claim 1 is characterized in that: The life prediction and decision-making process includes the following steps: Step 1: Life prediction, which includes selecting a prediction method, prediction accuracy, and application of prediction results. The selected prediction method predicts the life of the energy router based on multi-source data fusion and artificial intelligence algorithms. The prediction accuracy verifies the accuracy of the prediction model through cross-validation and performance indicator evaluation methods to ensure that the prediction results can truly reflect the life status of the energy router and provide a reliable basis for subsequent decision support. The prediction results are used to guide the maintenance plan formulation, spare parts inventory management, and risk assessment of the energy router. By predicting the life of the energy router in advance, maintenance time and resources can be reasonably arranged to reduce failure rate and maintenance costs; Step 2: Decision support, the decision support includes decision requirements, decision models, decision optimization and decision implementation. The decision requirements determine the maintenance strategy, spare parts procurement plan and risk assessment decision requirements based on the life prediction results of the energy router, aiming to ensure the stable operation of the energy router and reduce the risk of failure and maintenance costs. The decision model constructs a decision model based on the prediction results, and automatically recommends the optimal decision plan based on the prediction results and actual needs. The decision optimization optimizes the decision plan by continuously collecting new data, updating the prediction model and decision model, and considering the uncertainty and risks in the actual operation. The decision implementation puts the optimized decision plan into practice.

Images