Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.785183
Title: Model-based parameter estimation for fault detection in process systems using multiparametric programming
Author: Che Mid, Ernie Binti
ISNI:       0000 0004 7970 7251
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Fault detection (FD) has become increasingly important for improving the reliability and safety of process systems. This work presents a model-based FD technique for nonlinear process systems using parameter estimation. For a system described by nonlinear ordinary differential equations (ODEs), estimation of model parameters requires solving an optimisation problem such that the residual between the measurements and model predicted values of state variables is minimised. However, solving an optimisation problem online can be computationally expensive and the solution may not converge in a reasonable time. Thus, a method for parameter estimation for FD using multiparametric programming (MPP) is proposed. In this technique, the nonlinear ODEs model is discretised by using explicit Euler's method to obtain algebraic equations. Then, a square system of parametric nonlinear algebraic equations is obtained by formulating optimality condition. These equations are then solved symbolically to obtain model parameters as an explicit function of the measurements. This allows computation of parameter estimates by simple function evaluation. The detection of fault is carried out by monitoring the changes in the residual between model parameter estimates and 'true' value. The application of the proposed technique for FD is demonstrated on evaporator, tank, heat exchanger, fermentation and wastewater treatment systems. In a single-stage evaporator, changes in heat transfer coefficient and composition of feed are obtained and estimated for FD. In a quadruple-tank system, tank leakage is investigated by estimating the cross-section of outlet holes. Fouling in heat exchanger is detected where the overall heat transfer coefficient is estimated and the fouling resistance is monitored. In demonstrating the technique in relation to the fermentation and the wastewater treatment systems, kinetic model parameters are estimated for FD. The proposed work successfully estimates the model parameters and detects the faults through simple function evaluations of explicit functions. This demonstrates the advantages of MPP for FD using parameter estimation to detect faults quickly and accurately. In addition, a comparison of the implicit Euler's method and explicit Euler's method for discretisation of nonlinear ODEs model for parameter estimation using MPP is presented. Complexity of implicit parametric functions, accuracy of parameter estimates and effect of step size are discussed.
Supervisor: Dua, V. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.785183  DOI: Not available
Share: