Improved Particle Swarm Optimization With Redundant Solution Constraints for Robust Data Reconciliation

In the alumina evaporation production process, accurate and reliable process data is crucial for the effectiveness of the soft measurement models used to estimate the concentration of the outlet mother liquor. However, process data is often compromised by harsh industrial conditions such as high temperature, pressure, and alkalinity, leading to measurement errors, missing data, and anomalies. To address these issues, we propose a robust data reconciliation method that utilizes redundant solution constraints and an improved particle swarm optimization (PSO) algorithm with chaotic mutation to correct erroneous measurements and estimate unmeasured variables. This method leverages redundant process data and integrates statistical analysis with process mechanisms to enhance data accuracy. The proposed method was tested on actual industrial data from a multi-effect evaporation system. Results demonstrate that our approach significantly improves the reliability of the reconciled data, providing a robust foundation for soft measurement models. This work presents a novel solution to the challenges of data reconciliation in complex industrial processes, ensuring better accuracy and reliability in the presence of harsh operational conditions.