PROCESS ESTIMATION AND SOFT SENSOR Chinese Journal of Chemical Engineering, 20(6) 1206—1212 (2012) Improved Disturbance Observer (DOB) Based Advanced Feedback Control for Optimal Operation of a Mineral Grinding Process * ZHOU Ping (周平) 1, ** , XIANG Bo (向波) 2 and CHAI Tianyou (柴天佑) 1 1 State Key Laboratory of Synthetical Automation for Process Industries, Northeastern University, Shenyang 110819, China 2 Automation Department, Tangshan College, Tangshan 063020, China Abstract Advanced feedback control for optimal operation of mineral grinding process is usually based on the model predictive control (MPC) dynamic optimization. Since the MPC does not handle disturbances directly by controller design, it cannot achieve satisfactory effects in controlling complex grinding processes in the presence of strong disturbances and large uncertainties. In this paper, an improved disturbance observer (DOB) based MPC ad- vanced feedback control is proposed to control the multivariable grinding operation. The improved DOB is based on the optimal achievable H 2 performance and can deal with disturbance observation for the nonminimum-phase delay systems. In this DOB-MPC advanced feedback control, the higher-level optimizer computes the optimal op- eration points by maximize the profit function and passes them to the MPC level. The MPC acts as a presetting controller and is employed to generate proper pre-setpoint for the lower-level basic feedback control system. The DOB acts as a compensator and improves the operation performance by dynamically compensating the setpoints for the basic control system according to the observed various disturbances and plant uncertainties. Several simulations are performed to demonstrate the proposed control method for grinding process operation. Keywords disturbance observer, model predictive control, advanced feedback control, grinding process, steady-state optimization, disturbance rejection 1 INTRODUCTION In mineral industry, grinding circuit is used to reduce the particle size of the ore such that valuable mineral constituents are exposed and can be recovered in subsequent operations. In the operation, the product particle size and circulating load, which represent the product quality and grinding production efficiency, respectively, should be controlled effectively to achieve the optimal operation for the grinding process [1, 2]. Nowadays, model predictive control (MPC) is widely recognized as the dominant operational control technology in process industries [3-10]. To realize op- timal process operation, MPC generally is implemented as a part of multilevel hierarchy of control functions [3, 4, 7, 8-10], as shown in Fig. 1. The optimal steady-state settings from the upper plant-wide optimizer are di- rected to the local real-time steady-state optimization layer. Then, the local optimization system computes an optimal operation point and passes it to the MPC level. The MPC layer will move the plant from one steady-state to a more profitable operation point by changing the setpoints to the basic feedback control layer by solving a specified optimization problem. The basic control system is responsible for basic safety of dynamic processes, and capable of enforcing the set- points determined by the MPC level [5, 8]. The process operation of the mineral grinding circuit is sensitive to various disturbances, such as the variations of ore hardness and feed particle size, model mismatches, and coupling effects. These disturbances and plant perturbations have a great influence on the operation performance of grinding processes. It should be pointed out that MPC, as well as many other ad- vanced control schemes, usually cannot achieve satis- factory effects in controlling complex processes in the presence of strong disturbances or uncertainties. The reason is that they do not handle these disturbances directly by controller design. These single feedback based control schemes usually cannot directly and promptly reject disturbances and compensate plant uncertainties, though the control system can asymp- totically suppress them through feedback regulation in a relatively slow way [5, 8]. To improve the process operation performance in such cases, the effective method is to use disturbance observer (DOB) technique, which estimates disturbances Received 2012-05-26, accepted 2012-07-25. * Supported by the National Natural Science Foundation of China (61104084, 61290323) and the Guangdong Education Univer- sity-Industry Cooperation Projects (2010B090400410). ** To whom correspondence should be addressed. E-mail: zhouping@mail.neu.edu.cn Figure 1 The popular advanced feedback control struc- ture with MPC dynamic optimization