Zhengjiang Zhang

Correntropy Based Data Reconciliation and Gross Error Detection and Identification for Nonlinear Dynamic Processes

Abstract Measurement information in dynamic chemical processes is subject to corruption. Although nonlinear dynamic data reconciliation (NDDR) utilizes enhanced simultaneous optimization and solution techniques associated with a finite calculation horizon, it is still affected by different types of gross errors. In this paper, two algorithms of data processing, including correntropy based NDDR (CNDDR) as well as gross error detection and identification (GEDI), are developed to improve the quality of the data measurements. CNDDRs reconciliation and estimation are accurate in spite of the presence of gross errors. In addition to CNDDR, GEDI with a hypothesis testing and a distance–time step criterion identifies types of gross errors in dynamic systems. Through a case study of the free radical polymerization of styrene in a complex nonlinear dynamic chemical process, CNDDR greatly decreases the influence of the gross errors on the reconciled results and GEDI successfully classifies the types of gross errors of the measured data.

Simultaneous data reconciliation and gross error detection for dynamic systems using particle filter and measurement test

Abstract Good dynamic model estimation plays an important role for both feedforward and feedback control, fault detection, and system optimization. Attempts to successfully implement model estimators are often hindered by severe process nonlinearities, complicated state constraints, systematic modeling errors, unmeasurable perturbations, and irregular measurements with possibly abnormal behaviors. Thus, simultaneous data reconciliation and gross error detection (DRGED) for dynamic systems are fundamental and important. In this research, a novel particle filter (PF) algorithm based on the measurement test (MT) is used to solve the dynamic DRGED problem, called PFMT-DRGED. This strategy can effectively solve the DRGED problem through measurements that contain gross errors in the nonlinear dynamic process systems. The performance of PFMT-DRGED is demonstrated through the results of two statistical performance indices in a classical nonlinear dynamic system. The effectiveness of the proposed PFMT-DRGED applied to a nonlinear dynamic system and a large scale polymerization process is illustrated.

Robust particle filter for state estimation using measurements with different types of gross errors

For industrial processes, the state estimation plays a key role in various applications, such as process monitoring and model based control. Although the particle filter (PF) is able to deal with nonlinear and non-Gaussian processes, it rarely considers the influence of measurements with gross errors, such as outliers, biases and drifts. Nevertheless, measurements of dynamical systems are often influenced by different types of gross errors. This paper proposes a robust PF approach, in which gross error identification is used to estimate magnitudes of gross error. The gross errors can be removed or compensated so that a feasible set of particle sampling can contain the true states of the system. The proposed robust PF approach is implemented on a complex nonlinear dynamic system, the free radical polymerization of styrene. The application results show that the proposed approach is an appealing alternative to solving PF estimation problems with measurements containing gross errors.

Valve stiction detection using the bootstrap Hammerstein system identification

Stiction in control valve is one of the long-standing and common problems in the process industries, which accelerates equipment wear and even affects the stability of closed-loop systems. A valve stiction model is proposed to describe the dynamic feature of sticky valve. To detect and quantify valve stiction, a bootstrap Hammerstein system identification procedure is proposed. Through the identified set of the model parameters and operation plant data, the parameter confidence intervals can be predefined and the valve stiction can be easily detected. Numerical examples are provided to illustrate the effectiveness of the proposed method.

Optimal design of H 2 /H ∞ based robust PID controller by constrained extremal optimization and differential evolution

How to design an optimal mixed H2/H∞ robust FID controller for a complex control system is of great practical importance, but it is still an open issue. From the perspective of evolutionary algorithm, this paper formulates this issue firstly as a typical constrained optimization problem by minimizing a weighted objective function consisting of the robust stability performance, disturbance attenuation performance and tracking error, and then presents a novel a novel optimal design method of mixed H2/H∞ robust FID controllers based on a constrained evolutionary algorithm called CEO-DE by combining constrained extremal optimization and differential evolution. The simulation results on two typical multi-variable control systems have demonstrated the proposed CEO-DE based design method performs better than some reported popular methods such as intelligent genetic algorithm and multi-objective particle swarm optimization.

Methodology of multi-group particle filter for robust state estimation in nonlinear dynamic process systems

State estimation plays an important role for both process control and fault detection. In this paper, a methodology of multi-group particle filter is proposed for the uncertainty problem of state initialization in the nonlinear process systems. The measurement test criterion is introduced to indirectly identify whether the state initialization is accurate. According to the result of identification, multi-group particle filter is selected to generate the initial particles under bad state initialization, which can increase the probability of generating correct initial particles. The rectified errors of observed variables are used for the selection of the optimal particles. Finally, reliable state estimation would be derived through iterative particles. The proposed methodology of multi-group particle filter is compared with the generic particle filtering method through two examples of nonlinear dynamic systems. The results demonstrate the effectiveness and robustness of the proposed methodology.