热压混合材料板力学特性PSO-SVR模型预测

摘要

In this paper,in order to predict the mechanical properties of the mixed material board accurately and quickly,reduce production costs and improve resource utilization,taking the process of hot-pressing control as the research object,a support vector machine regression (SVR) model based on particle swarm optimization (PSO) optimization was proposed.Based on the performance test data of finished board and orthogonal experimental design,built the predictive model which took the hot-pressing pressure,hot-pressing temperature,hot-pressing time and moisture content of slab as the argument variables,and the modulus of rupture (MOR),modulus of elasticity (MOE) and internal bonding strength (IB) as the dependent variables.Comparison and analysis of PSO-SVR and SVR prediction results showed that PSO-SVR model could well describe the nonlinear relationship between the hot-pressing control parameters and the mechanical properties of the mixed material board and achieved rapid and accurate prediction according to the independent variables.Compared with SVR,PSO-SVR algorithm model had the advantages of strong robustness,high precision and fast learning speed,which could provide reference for the prediction of mechanical properties of mixed material under different process parameters in hot-pressing process.%精确、快速预测热压过程混合材料板力学特性,可降低生产成本,提高资源利用率.文章以热压过程为研究对象,提出基于粒子群算法(Particle Swarm Optimization,PSO)优化支持向量机回归(Support Vector Regression,SVR)模型.通过正交试验设计,结合混合材料板性能测试数据,以热压压力、热压温度、含水率、热压时间为自变量,预测混合材料板静曲强度、弹性模量、内结合强度.对比分析PSO-SVR与SVR预测结果,结果表明,PSO-SVR预测模型可明确热压参数与混合材料板力学特性间非线性关系,根据自变量预测混合材料板力学特性.与SVR相比,PSO-SVR算法模型具有鲁棒性强、精确度高、泛化能力强等优点.研究结果可为混合材料板力学特性预测及热压控制参数选择提供参考.