On the precision evaluation in non-linear sensor network design

Abstract

The aim of this work is to provide a tool for evaluating the precision of process variable estimates in the context of the optimal design of a sensor network in chemical plants. One of the possible formulations for the optimal design of an instrumentation system for monitoring tasks is the solution of nonlinear optimization problems with constraints, where the objective function is the cost of the instrument and the constraints are the observability and global precision associated with a sensor placement. When a metaheuristic approach is used to solve this problem, a methodology for computing the constraints is needed to evaluate the quality of a proposed solution. A simulation technique has been selected to solve the precision associated with a set of measurements. The simulator requires a variable classification methodology and a data reconciliation function that consists of solving another non-linear optimization. The proposed strategies have been applied to a continuous stirred tank reactor, a nonlinear problem including flows, compositions, and temperatures related by mass and energy balances. Results demonstrating the performance of the proposed metaheuristics are presented.