73 / 2025-03-30 23:31:40

A Fault Diagnosis Method for the TE Process Based on the Improved SPBO Algorithm to Optimize - AE - BiGRU

Autoencoder,Fault Diagnosis,Tennessee Eastman Process,Bidirectional Gated Recurrent Unit,Improved Student Psychology Based Optimization Algorithm

Aiming at the problems of chemical engineering data, such as nonlinear distribution, diverse categories, large scale, and difficult-to-distinguish fault characteristics, this paper proposes an improved combination fault diagnosis model of the Student Psychology Based Optimization Algorithm - Autoencoder - Bidirectional Gated Recurrent Unit (AE-BiGRU) to improve the accuracy and reliability of the fault diagnosis of the Tennessee Eastman (TE) process. Firstly, the Autoencoder (AE) is used to extract the features of the TE process data, which effectively reduces the data dimension and retains the key feature information. The Bidirectional Gated Recurrent Unit  has the ability to capture sequence data bidirectionally, and it can better handle the time series characteristics in the chemical process data, thus improving the accuracy of fault diagnosis. In order to further optimize the parameters of the AE-BiGRU model, the Student Psychology Based Optimization Algorithm is introduced. The Tent mapping and Cauchy mutation strategies are used for improvement, and then the optimal solution is quickly searched in the search space to obtain the optimal parameter combination and improve the performance of the model. The experimental results show that compared with other single fault diagnosis methods and combined fault diagnosis methods, the proposed method has significant advantages in terms of fault recognition accuracy, precision, recall rate, and F1 value, and it can accurately and quickly diagnose a variety of fault types.