改进BP神经网络的数据融合方法在智能灯光控制系统中的应用  

Improved BP Neural Network Data Fusion Method Application in Intelligent Lighting Control System

作　　者：郭瞻 肖祖铭 GUO Zhan;XIAO Zu-ming(School of Mechanical and Electronic Engineering,Jingdezhen University,Jiangxi Jingdezhen 333000,China)

机构地区：[1]景德镇学院机械电子工程学院,江西景德镇333000

出　　处：《机械设计与制造》2022年第9期88-92,共5页Machinery Design & Manufacture

基　　金：2019年江西省教育厅科技项目(GJJ191175);2018年市科技局项目(20182GYZD012-02)。

摘　　要：针对现有智能灯光控制系统的高能耗以及无法准确确定人体的静止状态等问题,在多传感器采集的基础上,提出一种将BP神经网络与改进遗传算法相结合的多信息融合算法用于智能灯光控制系统中。通过改进遗传算法获得一组次优解,用作训练BP神经网络初始权值和阈值。通过仿真将这里算法与遗传算法优化的BP神经网络算法、BP神经网络、遗传算法进行比较,以验证融合算法的优越性。仿真结果表明,该算法在收敛性、网络能耗和网络时延等方面都有较大改善,平均收敛时间为4.11s,检测精度为100%,具有一定的实用性。这项研究为智能灯光控制系统的发展提供了一定的参考。In view of the high energy consumption of the existing intelligent light control system and the inability to accurately de⁃termine the static state of human body,a multi-information fusion algorithm combining BP neural network and improved genet⁃ic algorithm is proposed for the intelligent light control system based on the multi-sensor acquisition.A set of suboptimal solutions are obtained by improved genetic algorithm and are used as the initial weights and thresholds of BP neural network algorithm training.Through simulation,the algorithm in this paper is compared with BP neural network algorithm,BP neural network al⁃gorithm and genetic algorithm optimized by genetic algorithm to verify the superiority of the fusion algorithm in here.The simula⁃tion results show that the algorithm has great improvement in convergence,network energy consumption,network delay,etc.The average convergence time is 4.11 seconds,the detection accuracy is 100%,and the algorithm has certain practicability.This study provides a reference for the development of intelligent light control system.

关 键 词：智能灯光控制系统 多传感器 多信息融合 BP神经网络 遗传算法 

分 类 号：TH16[机械工程—机械制造及自动化]