Programming Of Intelligent Hybrid System Based On Rough Sets And Neural Networks For Fault Diagnosis

The current intelligent fault diagnosis technology tends to combine allkinds of intelligent technologies. If the training data and diagnostic dataused in neural network diagnosis have redundant information, it will affectthe diagnostic ability. The paper's major task is to combine rough settheory and neural network, compile certain practical program applied tofault diagnosis system to realize data reduction and adopt neural networkto train and diagnose.The paper proposes an intelligent diagnosis system based on rough setand neural network. Taking centrifugal compressor as an example, itdesigns two modules: data module and network module. For two data typesgets from three different input modes, it compiles corresponding programand makes a series of treatments including discretization of SOM neuralnetwork and training sample's reduction based on rough set, whichsimplifies training sample and meets the request of BP network diagnosis,reduces the complicated degree of neural network, makes diagnostic data simplified and meet the request of neural network diagnosis.The paper also introduces some theories of software realization ofestablish intelligent diagnosis system based on rough set and neuralnetwork. On the basis of this, it mainly adopts MATLAB and its toolbox tocompile application program of some modules, establishes VB systeminterface platform, and achieve their seamless connection by reasonablemethod, realizes the data processing function. Meanwhile, it presents theconcrete code.