基于组合优化模型的配网开关柜温度预测与预警机制优选

Optimization of Temperature Prediction and Early Warning Mechanism of Distribution Network Switchgear Based on Combinatorial Optimization Model

针对配网开关柜温度存在误报警、漏报警,提出可用于在线监测的组合优化温度预测模型以及对应的预警机制优选方案。使用Savitzky-Golay算法将原温度序列分解成线性和非线性温度序列,对线性温度序列利用移动平均差分自回归(ARIMA)进行预测分析,对非线性温度序列使用极限学习机(ELM)进行预测,利用海洋捕食者算法对ELM的关键参数进行自动寻优,预测结果显示均方根误差为1.172℃。依托组合预测模型输出的高精度温度预测值并结合传统预警机制,形成预测温度评价指标,通过独立性权系数法对预测温度评价指标进行权值排序,筛选出最贴近实际工况的预警机制。

In view of the shortcomings of false alarm and missing alarm in the temperature alarm method of distribution network switchgear,a temperature prediction model is proposed based on combinatorial optimization model and the corresponding optimization scheme of early warning mechanism.The model can be used to online monitor.Firstly,the Savitzky Golay algorithm is used to decompose the original temperature series into linear temperature series and nonlinear temperature series,then the linear temperature series are predicted and analyzed using the moving average absolute deviation autoregression(ARIMA)model,the nonlinear temperature series are predicted using the extreme learning machine(ELM)model,and the marine predator algorithm is used to automatically optimize the key parameters of ELM.The prediction results show that the root-mean-square deviation is 1.172℃.Based on the high-precision temperature prediction values output by the combination prediction model and combined with traditional warning mechanisms,a predictive temperature evaluation index is formed.Finally,the independent weight coefficient method is used to sort the weight of the predictive temperature evaluation index and select the warning mechanism that is closest to the actual working conditions.