基于混沌果蝇优化最小二乘支持向量机的秸秆发酵过程软测量建模

Soft Sensor Model for Straw Fermentation Process Based on Least Squares Support Vector Machine Optimized by Chaos Fruit Fly Algorithm

针对秸秆发酵制取燃料乙醇过程的关键参量乙醇质量浓度难以用传统物理传感器实时在线测量,给发酵过程的监测与控制带来困难这一难题,采用混沌果蝇优化算法(chaos fruit fly optimization algorithm,CFOA)优化最小二乘支持向量机(least square support vector machine,LSSVM)的关键参数,避免了普通交叉验证法选取参数的耗时性和盲目性,建立混沌果蝇最小二乘支持向量机软测量模型,实现乙醇质量浓度的在线实时测量.实验仿真表明:基于CFOA的LSSVM模型平均误差为4.55%,明显优于LSSVM模型,表明该软测量建模方法预测能力强,测量精度高.

It is difficult to directly measure the product concentration by using traditional physical sensors during the straw fermentation process, which makes the monitoring and real-time control impossible. To resolve this problem, the chaos fruit fly optimization algorithm ( CFOA) is introduced to least square support vector machine ( LSSVM) to optimize some key parameters, which overcomes some shortcomings of the cross validation method such as time consuming and blindness in parameter selection. Using this way, the CFOA-LSSVM soft sensor model is built for the straw fermentation process, which realizes the real-time measure of product concentration in this process. The simulation shows that the average measurement error of the proposed CFOA-LSSVM soft sensor is 4. 5 5 % , which is smaller than the traditional LSSVM model. The proposed CFOA-LSSVM soft sensor model has strong forecasting capability and high accuracy.