基于数据预测与自适应优化的继电保护算法OA
Relay Protection Algorithm Based on Data Prediction and Adaptive Optimization
针对配电自动化终端运维消缺作业时普通备用保护装置接线复杂且保护准确率低的问题,本文提出一种基于历史运行数据预测与自适应优化的配电自动化终端临时继电保护动态定值计算方法.该方法在保留原有主保护的基础上,增设一套临时保护单元,构成可选双重化配置.核心在于利用采集的系统全景历史数据(电压、电流、功率等),经自适应飞蛾火焰优化算法(adaptive moth-flame optimization,AMFO)优化的先进梯度提升决策树算法(extreme gradient boosting,XGBoost)回归模型,对未来最优保护定值进行精准预测,并据此动态设定备用保护的动作阈值.主、备保护可设置单"或"或"与"逻辑运行,有效防止单一保护误动,同时确保在真实故障时具备双重防护能力.仿真试验结果验证,该算法能显著提升保护系统的自适应性与可靠性,有效降低拒动与误动风险.
To address the issues of complex wiring and low protection accuracy in conventional backup protection devices during the operation and maintenance of distribution automation terminals,it proposes a dynamic relay protection setting method based on histori-cal data prediction and adaptive optimization.This method retains the original main protection while adding a set of temporary protec-tion units,forming an optional dual configuration.The core lies in the utilization of the collected panoramic historical data of the sys-tem(voltage,current,power,etc.).The regression model of the advanced extreme gradient boosting(XGBoost)decision tree algo-rithm optimized by the adaptive moth-flame optimization(AMFO)algorithm accurately predict the future optimal protection setting val-ue and dynamically set the action threshold of the backup protection accordingly.The main and backup protections utilize a singal"or"or"and"logic,effectively preventing false activations from a single protection system,while ensuring dual protection capabili-ties in the event of a real fault.Simulation results confirm that this algorithm enhances the adaptability and reliability of the protection system,effectively reducing the risks of both false activations and missed activations.
李斌;徐峥;刘健;刘诚;陈卉雯
国网辽宁省电力有限公司鞍山供电公司,辽宁 鞍山 114002国网辽宁省电力有限公司鞍山供电公司,辽宁 鞍山 114002国网辽宁省电力有限公司鞍山供电公司,辽宁 鞍山 114002国网辽宁省电力有限公司鞍山供电公司,辽宁 鞍山 114002国网辽宁省电力有限公司鞍山供电公司,辽宁 鞍山 114002
信息技术与安全科学
继电保护飞蛾火焰优化算法XGBoost回归模型
relay protectionAMFOXGBoost regression model
《东北电力技术》 2026 (9)
28-32,5
评论