Optimization approach of faults samples for heavy automobile engine

Nowadays, the faults diagnosis approach based on artificial neural networks has achieved great progress [1, 2]. However, in the maintenance work, it is found the accuracy rate of faults diagnosis of heavy automobile engine is low [3, 4]. To eliminate the work faults, it is necessary for technical workers to check and compare faults symptoms carefully. Therefore, the diagnosis eect greatly depends on the experience and professional knowledge of technical workers.