基于KPCA-DFNN海洋微生物发酵过程软测量建模

Soft Sensor Modeling for the marine microbe fermentation process based on KPCA and DFNN

摘要:

针对海洋微生物发酵过程中关键生物参量（基质浓度、菌体浓度、产物浓度等）在线测量困难，离线化验滞后大，难以实现实时控制的问题，提出了一种基于核主元分析(KPCA)与动态模糊神经网络(DFNN)相结合的软测量方法。以典型的海洋微生物-海洋蛋白酶发酵过程为例，通过KPCA提取输入数据空间中的非线性主元，将提取的主元作为DFNN的输入，基质浓度、菌体浓度、相对酶活作为DFNN的输出，建立了基于KPCA-DFNN的海洋蛋白酶发酵过程生物参量软测量模型。仿真结果表明，KPCA-DFNN模型比DFNN和PCA-DFNN建模的测量精度高，跟踪性能强，能很好地满足发酵过程中生物参量的测量要求。

Abstract:

To overcome the difficulty that crucial biological variables ( such as substrate concentration，biomass concentration，product concentration，etc.) cannot be effectively controlled during the marine microbe fermentation process due to a lack of real-time on-line instrumentation，a soft sensor method is proposed by combining the Kernel Principal Component Analysis ( KPCA) with the Dynamic Fuzzy Neural Network (DFNN).The typical marine microbe fermentation process (the marine protease fermentation process) was taken as an example. Firstly, KPCA was applied to choose the nonlinear principal component of the model input data space. And then its result was taken as input of the DFNN, substrate concentration , biomass concentration and relative enzyme activity were taken as output of the DFNN. Finally, the soft sensor model of biological parameters based on KPCA-DFNN is established in the marine protease fermentation process. Simulation results indicate that the KPCA-DFNN model has a higher accuracy, better tracking performance when compared with DFNN model and the PCA-DFNN model. Therefore, the proposed method can satisfy the requirements of on-line measurement of biological variables in the marine microbe fermentation process.