Risk Warning of Rock Burst Based on CSA Optimization

To provide risk warning of rock bursts, this study proposes a method of risk warning of rock bursts based on the crocodile search algorithm (CSA) optimization. Since there is no accurate analytical model for the chaotic nature of rock bursts, a capsule neural network model was developed to optimize the rock burst data series using the classification principle of capsule neural networks. Furthermore, the rock burst risk warning grade and frequency objective functions are formulated based on the shock entropy and risk frequency error. The constraint conditions are determined based on the shock cusp catastrophe potential, microseismic energy, electromagnetic radiation intensity, and stress. CSA is used to optimize the objective function for early warning of rock burst risk. To address the challenges of conflicting optimization of the rock burst objective functions and the slow search for the optimal solution, a shock factor is introduced based on the number of iterations to expedite the identification of rock burst data. Additionally, an explosion mutation strategy is proposed to enhance solution accuracy. Finally, the results demonstrate that the rock burst time series optimized by the capsule neural network data model can effectively characterize the rock burst information. Compared with the original Reptile Search Algorithm, the CSA algorithm improves the accuracy of risk warning of rock burst by optimizing the objective function of risk warning of rock burst.