关联向量机在微生物发酵传感器故障诊断中的应用

微生物发酵过程具有严重的非线性和时变性,有许多过程参数需要监控,因此发酵罐的传感器故障诊断显得尤为重要。为此提出了一种微生物发酵的故障诊断新方法,即两个关联向量机分别作为观测器和分类器。观测器用于估计某传感器所测得的参数值(文中以二氧化碳释放率为例)以便得到残差序列,分类器用于对残差序列进行分类。仿真结果表明这种方法是可以有效地诊断传感器的故障的。

The Application of Relevance Vector Machines to Microbiological Fermentation Sensor Fault Diagnosis

Microbiological fermentation is a complicated batch process with serious nonlinearity and time dependence. In order to make the microbiological fermentation process successful, many process parameters need to be monitored and controlled. Therefore the fault diagnose is important for microbiological fermentation sensor. Here a new method for fault diagnosis of microbiological fermentation sensor was provided, i.e. two relevance vector machines were used as observer and classifier respectively. The observer was applied to estimate the parameter values measured by the sensor to gain residual sequence of that parameter. The classifier was applied to classify the residual sequence. The result of a fermentation simulation demonstrated this method can effectively diagnose sensor fault.