Advanced Process Control Strategy, Software Design And Its Application

Control systems of modern industrial processes become large and complex recently. It tends to satisfy the more and more rigorous requirements in products quality, production efficiency, system safety as well as the concern with the environmental impact. The traditional PID control strategy is unable to solve the control problems of complex production processes with the multi-variable, highly non-linear and coupled with great time varying and delay. It leads to an imperative need to enhance research and development on new control strategies. The requirements of high control performance of enterprises and fast development of control theory and computer technology have boosted the research and application of advanced control strategies. The development of modern network and software technologies is now initiating a new technological revolution in process control. In the past few decades, a lot of commercialized softwares on advanced control have been developed and applied in industrial processes. However, there exists a big gap in the research and development on advanced control strategies between domestic progress and world advances. Based on detail analysis of achievements in key advanced control technologies made abroad, a distributed component technology is adopted in this paper to improve the design and research of traditional control softwares and the results of research on advanced control strategy with multi-layer distributed component technology is described in more details.Base on the research achievement in advanced control strategy for complex process systems and software development experiences, the soft sensor technology, global identification technology, model prediction control technology and multi-level distributed component technology are adopted to present a design patter of advanced control software, then the related theoretical fundamentals and conducts research on design methodology are also studied. A general soft sensor package, an industry process model identification package and a model prediction control package are developed and have been successfully applied in a complex real industrial process.The main achievements and innovative contributions of this paper are listed as follows: 1. Based on the analysis of advanced control strategies and the status of advanced control softwares, the main trend of development and the theoretical fundamentals of advanced~ III ~//IABSTRACTcontrol strategy are investigated and discussed. The objectives of research are identified and focused on the key technologies of advanced control strategy, including soft sensor technologies, model identification and model prediction control.2. For the soft sensor technology, algorithms of data pre-processing, supplementary variables selection and modeling methods are discussed. The increment method, FIR and first-order filtering are adopted for automatic rejection of outliers and data pre-processing, respectively. The traditional stepwise method is applied for automatic selection of supplementary variables and the final soft sensor model is derived by using the improved BP algorithm, partial least square algorithm as well as fuzzy neural network technology. A first-principle model is built and the problem of modeling and model auto-adjustment is also solved. The obtained results are adopted and proven in an industrial PTA process of a famous chemical engineering company.3. A global identification technology is introduced to build a global model that can precisely estimate the real process based on the input and output data by using the information provided by local identification. The result of the model development can better approximate the actual properties of a practical industrial process and hence has more application value to a greater extent.4. Model prediction control technique is one of the most successful advanced control strategies that is used most widely. The theoretical mechanism and algorithm of model prediction control is analysed in this paper combined with t...