Design of instrumentation sensor networks for non-linear dynamic processes using extended Kalman filter

This paper presents a methodology for design of instrumentation sensor networks in non-linear chemical plants. The method utilizes a robust extended Kalman filter approach to provide an efficient dynamic data reconciliation. A weighted objective function has been introduced to enable the designer to incorporate each individual process variable with its own operational importance. To enhance the evaluation accuracy of the weighted objective function, a true relative standard deviation measure has been employed in the presented formulation. A Genetic Algorithm (GA) has been used to solve the resulting constrained optimization problem due to cost-optimal and performance-optimal design objectives. The proposed method has been tested on a non-linear continuous-stirred tank reactor (CSTR) benchmark plant, illustrating its effective design capabilities.