| In this paper , the fault diagnostic problem of middle storage pulverizing system for ChongQing power plant is developed , By mixed knowledge representation ,BP neural network ,expert system based on knowledge technology etc are integrated in this system. According to the operating and fault characteristic of the pulverizing system , the general frame of fault diagnosis system for pulverizing system is presented firstly , then sample compiling module,fault style updating module,condition monitoring and dealing module,fault diagnosis and explaining module,data querying and analyzing module etc are established . Meanwhile this system is programmed with DELPHI5.0 and its interface is easy for operating .During the course of compiling fault sample , multi-method of knowledge representation , as produce rule ,frame ,neural network representation etc are synthesized to organize and express the fault sample of pulverizing system . Before being putted into the database , the sample data is disposed to ensure sample data integrity and no redundancy .During the course of develop fault diagnostic method , the influence to the training circle number with network structure,learning rate,initial weight value & door value etc are discussed. By comprehensive analyses and comparing , the comparatively rational value is adopted to be network's eigenvalue . In addition to mend self-learning rate BP neural network , one parameter called error grads amending coefficient is found , the result indicates that the fault diagnosis system could make error reach to the aim value quickly and efficaciously for all pulverizing system fault samples . If there are plenty of samples , this system can diagnose pulverizing system fault exactly , and the system also includes explaining mechanism which can provide fault reason and fault dissolving method . It is obvious that the system can help to prevent pulverizing system fault happening and ensure the pulverizing system run safely and reliably .
|
PDF Full Text Request