MPC Monitoring Based On Economic Performance

With the development of control theory, MPC (model predictive control) is widely used as the representative of the advanced controller because of its good control performance and economic benefits for industrial process, but at the same time, people realized that the controller can meet various performance requirements at the initial operation period, if without regular maintenance, the controller performance would be often degraded over time, so would be the economic benefits. Therefore, to ensure the process is safe, efficient, high-quality, low-power, the control system needs to be monitored. In this paper, we focus on the economic benefits of MPC, and a monitoring method based on economic indicators is proposed. The specific contributions are as follows:1. MPC performance monitoring based on MV (Manipulated Variable) statistical property. A MPC performance monitoring method based on MV is proposed in this paper, and this method could be used to identify the reasons of performance degrading. The mean value and variance of MV in IMC (Internal Model Control) were analyzed. This method was compared with the method based on CV (Controlled Variable). The result shows the feasibility of the method, and also demonstrates that the method based on MV is more sensitive to model mismatch and unmeasured disturbance. A numerical simulation in MPC control system was used to illustrate the accuracy.2. MPC economic performance monitoring. This paper derived a MPC economic performance supervision method based on the historical data, In this method, the economic profits formula is given, and several performance benchmarks based on the historical data are defined. The economic profits formula is used to associate the monitoring variables with economic profits which are measured by monetary units; it can measure the economic benefits of the entire factory in a more intuitive way. What’s more, an optimization variables monitoring method is proposed based on system identification, which can monitor the optimization variables in the process and to estimate whether they are in the optimal state. As an example, we analyzed the economic performance of the thermal power units, the burning unit, the load tracking performance and the denitration unit in a coal-fired power plant, and then put forward several economic performance monitoring methods under different benchmarks.3. The diagnosis system based on economic performance and the monitoring software. This diagnosis method is proposed mainly to diagnose the reasons that cause the economic performance degrading in control system. Furthermore, using correlation analysis, it can locate the control variables which cause the economic performance degrading. At the end of this paper, a piece of monitoring software is developed which contains the whole monitoring scheme mentioned before. As an example, the operation data in a real power plant is used to calculate the economic benefit using this software.