Typical Faults Recognizing for Seal Rings of Hydro-Dynamic Shaft Seal of Reactor Coolant Pump Based on ANN and Experimental Research

The hydro-dynamic shaft seal of the main coolant pump operates under strict conditions and serves as a critical barrier of the pressure water reactor. In this paper, in order to study the possible defects and their impact on the performance of the hydro-dynamic shaft seal of a million-kilowatt nuclear reactor main coolant pump during long-term operation, the experiments have been carried on by a self-build full-scale mechanical seal test rig of nuclear main coolant pump, and multiple performance parameters such as sealing performance parameters (leakage flow), thickness variation of liquid film and vibration response are measured. To simulate the defects on seal surface, laser etching was used. The Artificial Neural Networks (ANN) was first introduced into the recognition of shaft seal fault modes, and the comprehensive recognition rate was 87.5% through experimental verification. This work provides a theoretical and practical basis for service performance predicting and reusability evaluating of the hydro-dynamic shaft seal.