A fast impact load identification method based on kernel extreme learning machine

Recently, artificial neural networks (ANNs) have shown strong performance in impact load identification. However, model training and load identification remain time-intensive. To solve this, we propose a method based on kernel extreme learning machine (KELM) for impact load identification and optimize the key parameters of KELM using sparrow search algorithm (SSA). Compared to deep learning models like convolutional neural networks (CNNs) and recurrent neural networks (RNNs), KELM significantly enhances training and identification speeds while maintaining high accuracy. We validated the method with simulation data from a three-degree-of-freedom (3DOF) system and experimental data from a double-clamped beam. Results indicate that KELM offers higher accuracy in identifying impact loads compared to CNN and RNN. Meanwhile, this method also significantly improves model training and load identification speeds. Further validation with data from a pumped liquid rocket engine confirms its effectiveness for complex load paths in intricate structures, making it a promising tool for real-time impact load identification in engineering applications.