Optimal Sensor Placement of Acoustic Sensor for Compressor Blade Crack Detection based on Multi-objective Optimization

Abstract

Nowadays, acoustic sensors have been widely applied for crack detection of compressor blades. As the accuracy is mainly affected by signal quality, the optimal sensor placement (OSP) is significant for crack detection. To search the OSP for reliable signals, the multi-objective acoustic sensor optimization method is proposed to detect crack of compressor blade under variable working conditions. First, a multi-objective function is constructed based on comprehensive consideration of signal quality and sensor cost. Furtherly, the placement and number of acoustic sensors are optimized by multi-objective genetic algorithm. Finally, the long-short term memory network is utilized to fuse the reliable acoustic signals on feature-level, which can detect the crack under different working conditions. The compressor experiments are implemented to test the proposed method. After multi-objective optimization, two acoustic sensors at optimal placement are used to detect crack of five lengths. It can reach average accuracy of 97.74% under four working conditions. Comparing with other number and placement of acoustic sensors, the advantage of the proposed method is validated for crack detection with high accuracy and low sensor cost.