Research on Shaft Deformation Identification Based on Multi-Sensor Data Fusion of Hoisting Conveyance

Abstract

To validate the feasibility of SDI using HCOC and lay a theoretical foundation for SD monitoring, the SDI research is conducted based on HCOC. Firstly, a noise reduction algorithm for the AITQWPT is proposed based on the TQWT, the high-frequency components above a certain decomposition level are iteratively processed by a filter bank and finely decomposed. The particle swarm optimization algorithm is employed to automatically optimize quality factor parameters, and its fitness function is constructed based on kurtosis and correlation coefficients, which can perform frequency band selection and reconstruction and achieve noise reduction of HCOC under SD. Its denoising performance is evaluated through simulation models and test bench data. Secondly, on the basis of AITQWPT and residual convolutional neural network, a multi-sensor data fusion model for SDI is constructed according to the HCOC under SD. The normalized data from AITQWPT after denoising is input into the residual convolutional neural network, where the convolutional layer extracts shallow features. the pooling layer achieves dimensionality reduction, and the concatenation layer finishes the feature-level fusion. The fused features sequentially pass through residual modules for deep feature extraction, then input into the fully connected layer, and the SDI is performed through SoftMax function and the effectiveness is verified. The analysis results indicate that the proposed denoising method can effectively suppresses the interference components, highlight the SD characteristics, and has strong robustness for HCOC under SD. The multi-sensor data fusion model achieves the complementary of SD feature, reduces the number of layers and complexity of residual convolutional networks, verifies the feasibility of SDI through HCOC, and enhances the SDI accuracy. The aforementioned research revolutionizes the paradigm of SD monitoring, provides scientific guidance for SDI through HCOC and theoretical support for accelerating its industrial application.