基于改进WOA-BP神经网络的电气火灾预警算法  

Electrical fire warning algorithm based on improved WOA⁃BP neural network

作　　者：颜磊 王国兵[2] 翁旭峰 刘雪莹 江友华 YAN Lei;WANG Guobing;WENG Xufeng;LIU Xueying;JIANG Youhua(School of Electronic and Information Engineering,Shanghai University of Electric Power,Shanghai 200090,China;Shanghai Huajian Power Equipment Co.,Ltd.,Shanghai 201314,China;Shanghai Power Transmission and Transformation Engineering Co.,Ltd.,Shanghai 200000,China)

机构地区：[1]上海电力大学电子与信息工程学院,上海200090 [2]上海华建电力设备股份有限公司,上海201314 [3]上海送变电工程有限公司,上海200000

出　　处：《电子设计工程》2025年第1期21-26,共6页Electronic Design Engineering

基　　金：上海市自然科学基金(21ZR1424800)。

摘　　要：电气火灾是一种严重危害人员安全和财产损失的事件,因此增强对电气火灾的早期预测和预警至关重要。基于提高电气火灾预测准确性的目的,采用了改进鲸鱼算法优化BP神经网络的方法,构建了电气火灾预警模型。使用剩余电流、工作电流电压和线缆温度作为神经网络的输入特征,结合上述改进方法对权值和阈值进行优化。优化后的参数作为初始参数进行模型训练,用于输出电气火灾的概率。采用电气柜中回路数据进行试验,将预测概率与剩余电流异常持续时间进行模糊化处理,得出火灾决策。研究结果表明,所提模型相关系数达到0.97,相较于传统方法提高了0.08,具有更高的准确性和可靠性。Electrical fires constitute a severe threat to both personal safety and property,emphasizing the critical need for enhanced early prediction and warning systems.With the aim of improving the accuracy of electrical fire prediction,a predictive model has been constructed by employing an enhanced whale algorithm to optimize a BP neural network.This model utilizes residual current,working current voltage,and cable temperature as input features for the neural network,in conjunction with the aforementioned enhancement method for optimizing weights and thresholds.The optimized parameters are utilized as initial settings for model training,enabling the prediction of electrical fire probabilities.Experimental data from electrical cabinet circuits are employed,with the predicted probabilities subject to fuzzy processing relative to the duration of abnormal residual current.Consequently,this approach leads to the formulation of fire⁃related decisions.Research findings indicate that the proposed model achieves a correlation coefficient of 0.97,representing an improvement of 0.08 compared to traditional methods,showcasing higher accuracy and reliability.

关 键 词：电气火灾预警 鲸鱼优化算法 BP神经网络 模糊化 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]