Intelligent fault diagnosis of distillation column system based on PCA and multiple ANFIS

This paper proposes a novel method based on multiple adaptive neuro-fuzzy in combination of statistic method to detect and diagnose the faults occurring in complex dynamical systems. The basic idea is to use PCA to extract the features for reducing the complexity of the data achieved from a process. The most superior features are fed into multiple ANFIS to identify different faulty conditions in order to prevent the system from serious system failure and possible shutdowns. Each ANFIS has employed to diagnose one of the faults in order to make a decision about the abnormal cases. Ability, and at the same time simplicity and rapidity has significantly enhanced. Moreover, therepsilas no need to have information about the model or the structure, which is the best advantage of using this approach. Using multiple ANFIS units significantly reduces the scale and complexity of the system, speeds up the diagnosis, and simplifies the training of the network. As an example, the proposed algorithm has applied to fault diagnosis of a simulated nonlinear MIMO distillation column. Results confirm the effectiveness of this method comparing to single ANFIS. The presented procedure is applicable to a variety of industrial applications in which continuous on-line monitoring and diagnosis is needed.