Prediction control is a new type of computer control algorithm that has developed in recent years. Since it uses multi-step test, rolling optimization and feedback correction control strategy, the control effect is good, suitable for controlling industrial production processes that are not easy to establish exact digital models and more complex industrial production, so it is attached to the domestic and foreign engineering projects. It has been successful in the control system of industrial sectors such as oil, chemical, electricity, metallurgy, and machinery. The process of industrial production is complex, and the model we build is also imperfect. It is the very complex modern control theory of theory. The effect of its control is often not satisfactory, and even in some aspects, it is still not as traditional PID control. In the 1970s, in addition to strengthening the research on the modeling, system identification, adaptive control of the production process, people began to break the concept of traditional control ideas, and tried to develop a low requirements for various models and online calculations. Convenient, new type of algorithm for controlling comprehensive effect. In this context, a predictive control is, that is, the model algorith control (Mac-Model Algorithmic Control) is first applied in industrial control in France. Therefore, predictive control is not a product of a unified theory, but has gradually developed in industrial practice. At the same time, the development of computer technology also provides a substance foundation for the implementation of algorithms. Nowadays, the most popular algorithms include: Model Algorithm Control (MAC); Dynamic Matrix Control (DMC); Generalized Predictive Control (GPC); Generalized Prediction Point (GPP) Control; Inner Mode Control (IMC); Reasoning Control (IC), etc. Wait. The following is controlled as an example to illustrate the basic principles of prediction control. Model Algorithm (MAC) control mainly includes several parts such as internal model, feedback correction, rolling optimization, and parameter input trajectory. It uses a pulsed corresponding non-parameter model as an internal model, used in the past and future input and output states, and predicts the future output status according to the internal model. After the feedback correction is performed with the model output error, the secondary performance index is compared, and the secondary performance index is used to scroll, optimize, and then the current time is added to the system, complete the entire action loop. The basic characteristics of predict control include the establishment of a predictive model convenient; use rolling optimization strategy; model error feedback correction. These features reflect the essence of predictive control and is the difference between this control algorithm and other algorithms. The predictive control is accompanied by the development of the industry, so predictive control has a close combination of industrial production. The steel ball grinding machine for thermal power plants is a multivariate, large lag, strong coupling control object, and its mathematical model is difficult to establish. At present, most of the controllers installed in the domestic thermal power plant are PID controllers. Because the system variables are coupted, the PID controller is difficult to adapt, causing the steel ball mills that cannot be turned automatically. With the 8051 microcontroller, add the A / D8 interface and its interface circuit, plus the control key and the display, constitute a prediction controller. After using the MAC algorithm, it is possible to make up for the shortcomings of the PID controller. Since predictive control has the characteristics of adapting to complex production process, predictive control has powerful vitality. It can be predicted that with the continuous development and improvement of the forecast control in both theory and application, it will have a growing role in industrial production, showing a broad application prospect.
