The accuracy of wire rope damage identification is of great importance for ensuring engineering safety. To address the problems of low recognition accuracy and the tendency of existing wire rope damage signal recognition methods to fall into local optima, a CWT-IWOA-CNN model based on Continuous Wavelet Transform (CWT) and an Improved Whale Optimization Algorithm (IWOA)-optimized Convolutional Neural Network (CNN) is proposed in this study. First, wavelet denoising was employed to preprocess the original signals, and the effects of different wavelet bases, decomposition levels, and soft and hard threshold functions on denoising performance were systematically analyzed. Subsequently, the denoised signals were transformed into time–frequency images using CWT and used as input features for the neural network. To improve the training performance of the model, an improved whale optimization algorithm was introduced. The proposed IWOA enhances the search capability and convergence stability of the algorithm through chaotic initialization, a multi-stage nonlinear convergence factor, and a weighted update strategy. Based on this approach, an original CNN model and six CNN models optimized by IWOA, WOA, GWO, PSO, DE, and GA were constructed to investigate the effects of different intelligent optimization algorithms on CNN parameter tuning and classification performance. Experimental results demonstrate that the IWOA-CNN model achieved competitive performance in terms of accuracy, precision, recall, and F1-score. After hyperparameter optimization using IWOA, the CNN model achieved an average test accuracy of 94.73% over 30 independent runs. The results indicate that the proposed method can effectively improve the performance and stability of wire rope damage identification and provides a promising technical approach for intelligent wire rope damage detection.
Wang et al. (Tue,) studied this question.