论文部分内容阅读
青霉素发酵过程具有非线性极强、动态特性变化快等特点,机理模型难以建立,过程监控和控制很困难。本文针对以上特点,基于多方向主元分析法(MPCA),研究了多阶段MPCA法对青霉素发酵过程的建模及监控。该方法通过多阶段线性结构改进传统MPCA法的单模型线性化结构,以分段线性建模逼近非线性的方式,弥补传统MPCA法的不足,提高监控的准确度和速度。本文详细地讨论了多阶段MPCA建模方法、原理,并通过在青霉素发酵仿真过程的应用,验证了该方法的监控效果要好于传统MPCA法,可以更及时地检测到过程故障,从而进一步说明该方法比MPCA法的监控性能更可靠。
The process of penicillin fermentation has the characteristics of extremely strong nonlinearity and rapid change of dynamic characteristics. The mechanism model is difficult to set up and the process monitoring and control is very difficult. In this paper, based on the above characteristics, based on the multi-directional principal component analysis (MPCA), the multi-stage MPCA method of penicillin fermentation process modeling and monitoring. This method improves the single-model linearization structure of the traditional MPCA method through a multi-stage linear structure, makes up the non-linearity by piecewise linear modeling, makes up for the deficiencies of the traditional MPCA method and improves the accuracy and speed of monitoring. This paper discusses in detail the multi-stage MPCA modeling method and principle, and through the application of the penicillin fermentation simulation process, verifies that the monitoring effect of this method is better than the traditional MPCA method, which can detect the process fault more promptly, The method is more reliable than the MPCA method.