基于改进麻雀搜索算法优化BP神经网络的照明控制优化  

Lighting Control Optimization of BP Neural Network Based on Improved Sparrow Search Algorithm

作　　者：徐劲 吴全玉 孙健 潘玲佼 陶为戈 XU Jin;WU Quanyu;SUN Jian;PAN Lingjiao;TAO Weige(School of Electrical and Information Engineering,Jiangsu University of Technology,Changzhou 213001,China)

机构地区：[1]江苏理工学院电气信息工程学院,江苏常州213001

出　　处：《常州工学院学报》2024年第6期39-48,共10页Journal of Changzhou Institute of Technology

基　　金：国家自然科学基金青年项目(62001196);常州市社会发展项目(CE20225045)。

摘　　要：针对智能照明控制算法易陷入局部最优的问题,提出采用改进麻雀搜索算法(improved sparrow search algorithm,ISSA)对灯具调光系数进行最优化求解。该算法通过对麻雀搜索算法引入Tent混沌种群初始化、柯西变异和种群历史最优个体的策略来增强寻优能力,基于BP神经网络的照度模型来预测灯具调光系数,利用ISSA提取神经网络中的最优权值和阈值进行网络优化来提高预测精度。然后,通过基准测试函数对ISSA进行性能测试,并对多个优化算法进行调光寻优实验。仿真实验结果证实,相较于粒子群算法和麻雀搜索算法,基于ISSA-BP的照明控制方法,能够更加精确地找到最优调光比的组合,且在满足平均照度、照度均匀度和眩光值的前提下,实现最大化节能的要求。This study addresses the challenge of intelligent lighting control algorithms prone to local optima by designing an improved sparrow search algorithm(ISSA).ISSA enhances optimization by integrating Tent chaotic population initialization,Cauchy variation,and historical memory function.The research introduces a BP neural network-based illumination model to predict lamp dimming coefficients.ISSA optimizes this model by refining neural network weights and thresholds,thereby boosting prediction accuracy.Benchmark tests of ISSA’s performance and dimming optimization experiments across multiple algorithms were conducted.Compared to particle swarm optimization and the standard SSA,the ISSA-BP based ligh-ting control demonstrates superior precision in determining optimal dimming ratios.At the same time,it fulfills average illuminance,uniformity,and glare standards,significantly enhancing energy efficiency.

关 键 词：智能照明 麻雀搜索算法 混沌种群初始化 柯西变异 种群历史最优个体 

分 类 号：TU133.6[建筑科学—建筑理论]