Decentralized Coordinated Optimization Of Whole Ironmaking Process

As one of the principles that must be followed in economy development of our country, reducing energy consumption is of great importance in modern society which is increasingly short of energy supply. However, in the present, steel enterprises in China commonly apply the isolated mode of optimization and control to sub-processes of iron-making flow, and when raw material content fluctuates and various operation disturbances take place, the overall control effect on iron-making flow may not be so satisfied, with unnecessary energy cost and non-smooth running. To solve the above problem, this thesis proposed a coordinated optimization algorithm oriented at multi-time-scale inter-connected large systems and applied it in optimization and control of the simulated iron-making flow system. Results show that this algorithm has the advantage of fine real-time computation ability and good combination property.The contents of this thesis are described as follows:1. A decentralized predictive control algorithm targeted at multi-time-scale inter-connected large systems was proposed. This algorithm operates in rolling-horizon optimization form, and implements feedback for model correction. In each rolling horizon, original reduced state-space decomposition algorithm is used to realize decentralized control of inter-connected large system, and the size of optimization problem is decreased through the decomposition of interior and exterior states, thus the real-time computation ability is greatly improved while optimization performance hardly deteriorates.2. Based on this, a decentralized predictive control realization oriented at multi-time-scale system was proposed. The problem that resulted from multi-time scale is solved by virtually changing the sampling period of fast sub-system, and the length of prediction horizon is determined by the slowest sub-system, in order to solve the problem of bad real-time computation ability. Besides, a filter rule is enforced upon control variables to reduce frequent regulation of actuators, thus to confirm the smoothness of production.3. Using partial actual production data of Hangzhou Steel and Iron Corporation, simulated experiment was carried out on the sub-system models of iron-making flow to test the effectiveness of algorithm of alterable horizon-decentralized predictive control, and contrast simulations with existed algorithms were made which included integrated predictive control algorithm, isolated decentralized predictive control algorithm and normal distributed predictive control algorithm, and the optimization effects were compared in indices of computation time, object function and responses. Results show that algorithm of alterable horizon decentralized predictive control deals extraordinarily well with performance index and real-time computation ability. Besides, the integral coordinated optimization solution to Hangzhou Steel and Iron Corporation ironmaking process is given in this thesis.