Power Plant Production Equipment Abnormal Sound Perception Method Research based on Machine Hearing

As the frontier construction direction of Smart Power Plants, “Unmanned Monitoring” requires monitoring system to autonomously identify abnormal states of production equipment. So as to further improve the intelligence level of the power plant sound monitoring system, this paper proposes an abnormal sound perception scheme for power plant production equipment based on machine hearing technology. Firstly, the sound defect records and inspection records of the production equipment are investigated and analyzed. Besides, a sound monitoring evaluation system based on the failure risk and inspection experience is proposed. This paper analyzes the abnormal sound characteristics of power plant production equipment, and verifies the effectiveness of time spectrum as a feature learning. Finally, combined with its audio characteristics, the abnormal sound perception scheme is proposed taking the existing machine hearing technology as the main frame.