基于AW-GA-BP算法的配电网设备运行环境相对湿度的预测方法及应用

Prediction Method and Application of Relative Humidity of Operating Environment of Distribution Network Equipment with AW-GA-BP Algorithm

【目的】准确监控和预测配电网设备所处环境相对湿度的状态和变化趋势。【方法】通过分析影响相对湿度的相关因素,提出了一种权重调整(Weight of adjust,AW)和遗传算法(Genetic algorithms,GA)相结合的BP算法(AW-GA-BP算法),在此算法的基础上,建立了对配电网设备所处环境相对湿度变化的神经网络预测模型,并将此算法应用到项目组为云南省某供电局开发的配电网运行环境智能化监测系统上,利用该监测系统所采集到的数据,将不同采样试验数据分别作为训练样本和验证样本,对比研究了AW-GA-BP算法、GA-BP算法和标准BP算法的预测误差。【结果】基于AW-GABP算法预测得到的相对湿度百分误差平均值是4.28%,基于GA-BP算法预测得到的相对湿度百分误差平均值是8.87%,基于标准BP算法预测得到的百分误差平均值是14.64%。【结论】基于AW-GA-BP算法所建模型的相对湿度预测平均误差更小,为预测配电网设备所处环境相对湿度提供了一种更为准确的方法。

[Purposes]To accurately monitor and predict the status and trends of the relative humidity of the distribution network equipment.[Methods]With factors affecting relative humidity accounted,it proposed a BP algorithm(AW-GA-BP algorithm)by considering both the Weight of Adjust(AW)and the Genetic Algorithms(GA),and constructed a neural network model that could predict the change of environmental humidity where distribution network equipment is located.The algorithm was applied to the intelligent monitoring system of the distribution network operation environment of a power supply bureau in Yunnan Province.The collected data were used as training samples and verification samples respectively,and the prediction accuracy and error of AW-GABP algorithm and GA-BP algorithm and standard BP algorithm were compared.[Findings]Experiments showed that the average error of relative humidity predicted by AW-GA-BP algorithm was 4.28%,the average error of relative humidity predicted by GA-BP algorithm was 8.87%,the average error of relative humidity predicted by BP algorithm was 14.64%.[Conclusions]The model based on the AW-GA-BP algorithm would provide a more accurate method for predicting the relative humidity of the environment where the distribution network equipment is located.