基于机器学习和可解释人工智能的分布式智能电网可解释预测

The concept of a decentralized smart grid has emerged as a viable approach for efficiently managing and distributing electrical energy. Ensuring the stability and reliability of the grid, particularly with the integration of renewable energy sources and the increase in the number of prosumers, is a primary challenge in this domain. This study addresses this challenge by leveraging machine learning (ML) models and explainable artificial intelligence (XAI) techniques to predict the stability of a decentralized smart grid. A four-node star network implementing the decentralized smart grid control (DSGC) concept was investigated, and a dataset based on simulations of this network was used. Ten ML models, including Adaptive Boosting (AdaBoost), Artificial Neural Network (ANN), Gradient Boosting (GBoost), k-Nearest Neighbors (k-NN), Logistic Regression (LR), Naïve Bayes (NB), Random Forest (RF), Stochastic Gradient Descent (SGD), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost), were compared for their performance in predicting the grid stability. Models were evaluated using various metrics, and XAI methods, specifically SHapley Additive exPlanations (SHAP) and Individual Conditional Expectation (ICE) plots, were employed to enhance the interpretability of the models. The results demonstrate that the ANN model outperformed the other ML algorithms, achieving an area under the curve (AUC) of 99.4% and showing commendable performance in terms of cumulative accuracy (CA), precision, recall, and F1-score, all of which reached 96.2%. The SHAP and ICE analyses provided a comprehensive understanding of the model’s behavior, highlighting the critical influence of parameters such as reaction time, nominal power, and price elasticity on the stability of the decentralized smart grid. The findings contribute to the development of more reliable and interpretable predictive models for decentralized smart grid stability, which can aid in the design and optimization of effective control mechanisms.