基于改进麻雀搜索算法-BP神经网络的电缆接头线芯温度间接测量方法

电缆接头线芯温度实时监测对提升电缆线路载流量和安全运行有重要意义。针对目前测温方法适用性不强、精度低且抗干扰能力弱的问题，提出了一种改进麻雀搜索算法(improved sparrow search algorithm, ISSA)优化反向传播神经网络(back propagation neural network, BPNN)的温度反演间接测量方法。首先引入帐篷(Tent)混沌映射、自适应T分布变异、生产者数量和搜索空间动态调整混合策略对SSA进行改进，然后用改进后的SSA优化BP神经网络超参数。最后通过不同工况下的接头仿真与试验数据，结合自回归滑动平均模型(auto regressive moving average model, ARMA)对表面测温数据进行降噪，基于线路负荷及表面温度对接头线芯温度进行反演，并与粒子群优化算法(particle swarm optimization, PSO)-BP、SSA-BP、灰狼优化算法(grey wolf optimizer, GWO)-BP反演效果进行对比。结果表明改进模型的平均绝对误差不超过0.5℃,反演精度更高，能够实现对电缆接头运行...

Indirect measurement method of cable joint core temperature based on BP neural network optimized by improved SSA

The real-time monitoring of the core temperature of the cable joint is of great significance for improving the current carrying capacity and safe operation of the cable line. Aiming at the problems of poor applicability, low accuracy and weak anti-interference ability of current temperature measurement methods, an indirect measurement method of temperature inversion based on BP neural network optimized by Improved Sparrow Search Algorithm (ISSA) was proposed. Firstly, the Tent chaotic map, adaptive T distribution variation, the number of producers and the dynamic adjustment of the search space are introduced to improve the SSA, and then the hyperparameters of the BP neural network are optimized with the improved SSA. Finally, through the joint simulation test data under different working conditions, combined with ARMA to denoise the surface temperature measurement data, the joint core temperature is inverted based on the line load and surface temperature, and compare the inversion effect with PSO-BP, SSA-BP, GWO-BP. The results show that the MAE of the improved model does not exceed 0.5°C, the inversion accuracy is higher, and the real-time and effective monitoring of the operating state of the cable joint can be realized.