基于IBA-LSSVM的光合细菌发酵软测量

Soft sensor of photosynthetic bacteria fermentationbased on IBA-LSSVM

摘要:

针对光合细菌(PSB)发酵过程活菌浓度难以在线检测,离线测量又存在很大延时及易染菌的问题,提出一种基于改进蝙蝠-最小二乘支持向量机(IBA-LSSVM)的软测量建模方法。本文先对BA的速度更新公式进行改进,且将差分进化算法(DE)的变异机制引入BA,增加了种群多样性,进而提升了BA算法的全局及局部搜索能力,然后构建了活菌浓度的IBA-LSSVM软测量模型,并与BA-LSSVM软测量模型进行对比。仿真结果表明,改进的模型相较于BA-LSSVM模型有着更好的学习能力和预测性能,测量误差为0.1358,可为光合细菌发酵过程的优化控制提供准确有效的指导,有一定的实际应用价值。

Abstract:

In view of the living cell concentration is difficult to measure on line in photosynthetic bacteria(PSB)fermentation process and the off-line measurement is accompanied by large time-delay error and easy to stain bacteria , a soft sensor model based on IBA-LSSVM was proposed. The velocity equation of the BA algorithm was improved and the variance mechanism of DE algorithm was introduced in the BA algorithm. Thus, the diversity of the population can be increased and the global and local searching ability of the BA algorithm can be enhanced. Furthermore, the IBA-LSSVM soft sensor model was established for the living cell concentration and compared with BA-LSSVM soft sensor model. Finally, the simulation results showed that the improved model was better learning ability and prediction performance than BA-LSSVM, the measurement error is 0.1358. The improved model could provide accurate guidance for the photosynthetic bacteria fermentation control optimization. This model has certain practical value.