By using the regulating principle of immune genetic system and the recursive least square method

a new type RBF network model has been established.Through information entropy of the calculation characteristic parameters

a knowledge base for determining fault diagnosis has been rapidly and accurately determined.In application of said model and knowledge base in fault diagnosis of flow passage in steam turbines

the model has quick convergent speed and high precision

as well as better generalization ability.The said method has been used in practical diagnosis of a 300 MW unit in one power plant

the fouling in HP cylinder and the scaling in flow passage of the regulating valve of HP cylinder being judged

satisfying the requirements of fault diagnosis in flow passage of steam turbines.