基于PSO-BP的变电站交直流电源系统可靠性评价研究

Research on Reliability Evaluation of AC/DC Power Supply System in Substation Based on PSO-BP

变电站的交直流电源系统是电力系统的重要组成部分,它们负责为变电站内的各种设备提供稳定可靠的电力。变电站交直流电源系统是一个复杂的系统,包含多个组件和子系统,如直流电源系统、交流电源系统、蓄电池组、充电机、绝缘监察装置等,这些组件之间的相互作用和依赖关系使得系统的可靠性评价变得复杂,为此提出了一种基于PSO-BP的变电站交直流电源系统可靠性评价方法。分析影响交直流电源系统工作性能的各项因素,建立系统可靠性评价指标体系。明确BP神经网络结构和自学习模式,结合粒子群算法优化网络权值、阈值参数,构建基于PSO-BP的可靠性评价模型,对各项指标数据展开学习,得到可靠性综合评价值,依托于两级递进式划分等级模式确定电源系统可靠性级别,从而得出精准的变电站交直流电源系统可靠性评价结果。实验结果表明:新研究方法得出的可靠性评价结果F1值超过了0.9,实现了对电源系统工作状态的准确评估,实际应用效果好。

The Ac/DC power supply system of a substation is an important component of the power system,responsible for providing stable and reliable power to various equipment inside the substation.The AC/DC power supply system of a substation is a complex system that includes multiple components and subsystems,such as DC power supply system,AC power supply system,battery pack,charger,insulation monitoring device,etc.The interactions and dependencies between these components make the reliability evaluation of the system complex.Therefore,a PSo-BP based reliability evaluation method for Ac/DC power supply systems in substations is proposed.Analyze the various factors that affect the performance of Ac/DC power supply systems and establish a system reliability evaluation index system.Clarify the structure and self-learning mode of the BP neural network,optimize network weights and threshold parameters using particle swarm optimization algorithm,construct a reliability evaluation model based on Pso-BP,learn various indicator data,obtain comprehensive reliability evaluation values,and determine the reliability level of the power system based on a two-level progressive classification mode,thereby obtaining accurate reliability evaluation results of the substation Ac/DC power system.The experimental results show that the reliability evaluation results obtained by the new research method have an F1 value exceeding 0.9,which achieves accurate evaluation of the working status of the power system and has good practical application effects.