Expert system based on CBR and grey BP for Vannamei disease diagnosis

This paper proposes a novel inference method that is a combination of CBR (Case-based reasoning), grey theory and BP-ANN (Back propagation Artificial Neural Network). There are four basic steps in a typical CBR process: retrieving, reusing, revising and retaining. BP-ANN and grey theory are applied in retrieving to replace the traditional matching of similar cases. Before using the method to solve an actual issue, historical cases obtained from the case base will be converted to learning samples in the form of numerical values, then BPANN will be trained by these processed samples. The method has been successfully applied in the Vannamei expert system for disease diagnosis and treatments. Symptoms from users are converted to numerical values that will be inputted into BP-ANN, and outcomes of BP-ANN will be processed by defuzzification and graying. After that, the best therapeutic scheme of a similar disease from the case base will be sent back to the users. Meanwhile, the new case will be appended to the case base. As is found in experiment and in practical application, this system is theoretically feasible and practically effective.