基于LSSA-LSSVM的蚕茧解舒质量预测模型  

Prediction model of cocoon reeling quality based on LSSA-LSSVM

作　　者：邵铁锋 黄程卓[1,2] 孙卫红 梁曼 赵卫章 杨华 SHAO Tiefeng;HUANG Chengzhuo;SUN Weihong;LIANG Man;ZHAO Weizhang;YANG Hua(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China;Cocoon and Silk Quality Inspection Technology Institute,China Jiliang University,Hangzhou 310018,China;Huaian Fiber Inspection Institute,Huaian 223001,China;Shandong Fiber Inspection Institute,Jinan 250021,China)

机构地区：[1]中国计量大学机电工程学院,浙江杭州310018 [2]中国计量大学茧丝绸质量检测技术研究所,浙江杭州310018 [3]淮安市纤维检验所,江苏淮安223001 [4]山东省纤维质量监测中心,山东济南250021

出　　处：《中国测试》2023年第7期48-53,共6页China Measurement & Test

基　　金：市场监管总局科技计划项目(S2021MK0217);浙江省公益技术应用研究项目(LGG20E050014)。

摘　　要：针对煮茧工艺优化需根据解舒质量反复人工试煮而造成生产效率低、原料浪费等问题,基于LSSA-LSSVM算法,提出一种面向纤检机构真空减压煮茧工艺的蚕茧解舒质量预测模型。首先,提取蚕茧质量特性、真空减压煮茧工艺参数与解舒质量变量作为最小二乘支持向量机(LSSVM)的输入与输出变量。其次,引入拉丁超立方抽样方法(LHS)与Levy飞行策略优化原始麻雀搜索算法的初始化方式与位置更新方式,获得改进的麻雀搜索算法(LSSA)。最后,利用LSSA得到LSSVM的最优超参数组合(γ*,σ2*),建立解舒质量预测模型。实验结果表明,该模型预测准确率均值可达94.75%,预测时间均值为0.15 s,满足煮茧工艺精度与实时性要求,可用于煮茧工艺参数仿真优化,进而减少试煮次数,提高生产效率,该方法同时可推广至缫丝企业。Aiming at the problem of low production efficiency and waste of raw materials caused by repeated manual trial cooking according to the reeling quality in cocoon cooking process optimization,based on the LSSA-LSSVM algorithm,a prediction model of cocoon reeling quality for the vacuum decompression cocoon cooking process of fiber inspection institutions is proposed.Firstly,the cocoon quality characteristics,the vacuum decompression cocoon cooking process parameters and the reeling quality variables were screened as the input and output variables of the least squares support vector machine(LSSVM).Secondly,improved sparrow search algorithm(LSSA)was acquired by introducing Latin Hypercube Sampling(LHS)and Levy Flight Strategy to optimize the initialization and location update of the original sparrow search algorithm.Finally,The optimal super parameter combination(γ*,σ2*)of LSSVM was acquired by the improved sparrowsearch algorithm(LSSA).And the prediction model of reeling quality was established.The experimental results show that the average prediction accuracy of the model can reach 94.75%,and the average prediction time is 0.15 s,which meets the requirements of cocoon cooking process accuracy and real-time,and can be used for the simulation and optimization of cocoon cooking process parameters,thus reducing the number of trial cooking and improving production efficiency.

关 键 词：煮茧 解舒质量 改进的麻雀搜索算法 最小二乘支持向量机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TS143[自动化与计算机技术—控制科学与工程] TB9[轻工技术与工程—纺织材料与纺织品设计]