Optimization for Data-Driven Preventive Control Using Model Interpretation and Augmented Dataset

Abstract

Transient stability preventive control (TSPC) ensures that power systems have a sufficient stability margin by adjusting power flow before faults occur. The generation of TSPC measures requires accuracy and efficiency. In this paper, a novel model interpretation-based multi-fault coordinated data-driven preventive control optimization strategy is proposed. First, an augmented dataset covering the fault information is constructed, enabling the transient stability assessment (TSA) model to discriminate the system stability under different fault scenarios. Then, the adaptive synthetic sampling (ADASYN) method is implemented to deal with the imbalanced instances of power systems. Next, an instance-based machine model interpretation tool, Shapley additive explanations (SHAP), is embedded to explain the TSA model’s predictions and to find out the most effective control objects, thus narrowing the number of control objects. Finally, differential evolution is deployed to optimize the generation of TSPC measures, taking into account the security and economy of TSPC. The proposed method’s efficiency and robustness are verified on the New England 39-bus system and the IEEE 54-machine 118-bus system.

Data-driven active corrective control in power systems: an interpretable deep reinforcement learning approach

“…Reference Liu et al (2023) adopts the method of feature importance to explain and analyze the transient stability assessment process of power systems based on machine learning. Reference Ren et al (2021) proposes using feature importance method to explain the decision results of transient stability preventive control model and identify the most effective control objects to reduce the number of control objects. Reference Yang et al (2022) can predict the static voltage stability index online by correcting the control model, and calculate the approximate value of each characteristic sensitivity to the voltage stability index under any operating mode.…”

Giant Trevally Optimization Approach for Probabilistic Optimal Power Flow of Power Systems Including Renewable Energy Systems Uncertainty

“…The deterministic OPF (D-OPF) was initially addressed using conventional mathematical programming methods, which proved to be effective in demonstrating their feasibility [5]. Despite being commonly used to effectively solve the D-OPF issue, conventional optimization techniques, like Newton's technique [6], gradient projection technique [7], linear programming approach, and interior point approach [8], are reported to be associated with various challenges [9]. As optimization problems continue to evolve, new techniques encompassing artificial intelligence, as well as the metaheuristic search-based optimization approaches were designed to tackle the D-OPF problem.…”

Optimization Method for Multiple Measures to Mitigate Line Overloads in Power Systems

“…When the optimization methods are adopted, the control measures are usually simplified and represented by continuous constraints. In reference [14], a data-driven preventive control optimization strategy was introduced, which deployed differential evolution to obtain preventive control measures for transient stabilization of the power system. Reference [15] provided an overview of intelligent optimization algorithms that can be used for continuous load shedding in power systems and discussed the advantages and limitations of different algorithms.…”