To address the issues of simplicity and weak generalization in current fault diagnosis models,an improved CNN-Bi-LSTM model is employed for fault diagnosis in hydrometallurgical processes.Based on the diagnostic results,an enhanced random forest model is utilized to evaluate the entire hydrometallurgical process.The results indicate that the fault diagnosis accuracy can reach 90.7%,significantly surpassing accuracy of the existing rule-based diagnostic system at the factory(78.4%).Additionally,the fault detection response time is maintained within 2 seconds,ensuring real-time monitoring and rapid response during the process.

CNN-Bi-LSTM  /  random forest  /  numerical simulation  /  empirical study  /  fault diagnosis