基于多状态量感知技术的继电保护设备状态在线监测方法

Multi-status-variable Perception-based Online Monitoring of Protective Relays

针对现有监测方法存在的故障节点判定不佳、设备状态监测准确程度低的问题,提出基于多状态量感知技术的继电保护设备状态在线监测方法。通过继电保护状态信息之间的约束关系,计算不同节点的采样值分向量,获得设备状态信息。在校准环境下,通过提供外部阻尼来消除振荡。比较振幅中的细节系数得到最终去噪后的状态信号量。建立SSD监测模型生成不同层尺度的先验框,运用损失函数进行故障图像先验框匹配,得出最优先验预测框,锁定继电保护设备状态的检测结果,从而完成监测。实验结果表明,当空间相关度指标连续低于阈值的时间窗口时,判定测试的节点为故障节点;实验组的监测准确程度始终在95%以上,表现出较好的配准性能,实现对异常问题的精准监测。

Due to inferiorities such as incorrect determination of fault nodes and low monitoring accuracy of equipment status with respect to the currently prevailing monitoring methods,a multi-status-variable perception-based online monitoring method for protective relays was designed.Through the constraint relationship between relay status information,the sampling value component vectors of different nodes were calculated to obtain equipment status information.In a calibrated environment,oscillations were eliminated by providing external damping,and the factors of detail in the amplitude were compared to obtain the final denoised status signal.The SSD monitoring model was established to generate the prior frames of different layer scales,and the loss function was used to match the prior frames of the fault images.The prediction frame of the highest priority was obtained,and the detection results of relay status were locked,so as to complete the monitoring.The experimental results showed that when the spatial correlation index was constantly lower than the time window of the threshold,the tested node was determined to be a faulty node.The monitoring accuracy of the experimental group was always above 95%,showing satisfactory selectivity and accuracy in monitoring abnormalities.