思维进化算法和神经网络结合的PSD非线性优化  

PSD Nonlinear Optimization Combined with Mind Evolutionary Algorithm and Neural Network

作　　者：李壮 辛志杰[1,2] 周进节 崔帅[3] LI Zhuang;XIN Zhi-jie;ZHOU Jin-jie;CUI Shuai(School of Mechanical Engineering,North University of China,Taiyuan 030051,China;Deep Hole Processing Engineering and Technology Research Center of Shanxi Province,Taiyuan 030051,China;Shanxi Aerospace Qinghua Equipment Co.,Ltd.,Changzhi 046600,China)

机构地区：[1]中北大学机械工程学院,山西太原030051 [2]山西省深孔加工工程技术研究中心,山西太原030051 [3]山西航天清华装备有限责任公司,山西长治046600

出　　处：《仪表技术与传感器》2023年第3期110-113,121,共5页Instrument Technique and Sensor

摘　　要：针对位置敏感探测器(PSD)的非线性输出误差导致激光对中仪整体精度较差问题,建立思维进化算法(MEA)和LM-BP神经网络相结合的模型对其进行优化。该方法利用MEA较强的全局寻优能力,优化LM-BP网络的初始权阈值,找到最优个体构建MEA-LM-BP神经网络。使用MATLAB软件编程构建BP神经网络,对实验获得的原始数据进行分析处理。仿真结果表明,该方法可以有效提高LM-BP法的泛化能力,同时具有更高的优化精度,可以极好地提高传感器的测量精度以及仪器整体精度。Aiming at the problem of poor overall accuracy of the laser alignment instrument caused by the nonlinear output error of the position sensitive detector(PSD),a model combining the mind evolutionary algorithm(MEA)and the LM-BP neural network was established to optimize it.This method utilized the strong global optimization ability of MEA,optimized the initial weight threshold of LM-BP network,and found the optimal individual to construct MEA-LM-BP neural network.The BP neural network was constructed by MATLAB software programming,and the original data obtained by the experiment were analyzed and processed.The simulation results show that this method can effectively improve the generalization ability of the LM-BP method,and at the same time has a higher optimization accuracy,which can greatly improve the measurement accuracy of the sensor and the overall accuracy of the instrument.

关 键 词：PSD 非线性 思维进化算法 MEA-LM-BP法 神经网络 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]