| The iron and steel smelting production mainly consists of two processes:iron making-the extraction of usable metal from oxidized iron ores, and steel making-the further refining of the steel products. Within both processes, a great amount of energy consumption happens along with numerous chemical reactions, resulting in increasing temperature and heat and even worse production environment. Therefore, the production technique and control are too complicated for identifying the inside smelting state. Nevertheless, to realize precise control and optimization of the production processes of the blast furnace and converter in real practice, and thus improve the product quality, it is of vital importance to get the accurate information about the inside smelting state. In this respect, this dissertation would take the two representative processes:burden distribution of blast furnace and converter steel making as the objects. After making in-depth analytics into the operating principle of the two production processes, a series of new problems of operation analytics and optimization are proposed, as well as the operation analytics and optimization methods of certain values in practical terms. The details are as follows:(1) The operation optimization problems during the burden distribution production process of blast furnace are proposed, since the material layers with perfect O/C are hard to obtain manually due to the combined influence of multiple reasons. Through the operation analysis method, the perfect O/C to ensure the successful operation currently is available. Taking the deviation between the O/C related to the burden surface and the perfect O/C as the objective function, the burden weight, the opening of material flow control valve, chute rotation speed and dip angle as variables, and considering the radial alignment constraint of layer thickness and the operation time interval, etc., the operation optimization model for the burden distribution process is built. To meet the real-time requirement of the industrial sites, the dimensionality reduction is proceeded to formulate the mixed algorithm of Scatter Search and Differential Evolution for high-efficiency solutions. The result shows the maximum deviation is fewer than 3%, which speaks for the effectiveness. Based on the operation model and algorithm of the burden distribution of blast furnace, the real burden distribution system is developed, which can well improve the stability of the furnace condition, and meet the requirements of scientific and stable control of O/C and other technical indices at the industrial sites. In the mean time, the system would enable the function of providing burden distribution adjustment program in time when the furnace condition changes. After the practical application, the system is proved to be effective.(2) The operation optimization problems of the converter smelting production process can be classified into two categories due to different control methods:stable state operation optimization and dynamic state operation optimization. How to scientifically and reasonably determine the adding amount of the molten iron and dolomite, etc., is a challenge for stable state operation optimization. To solve this problem, we build the stable state operation optimization model for converter production process, taking the adding amount of various kinds of materials as variables, and the minimum deviation between the production cost and the hit rate of the end-point as the objective function, as well as the energy balance, the mass balance and the dynamic balance within the molten steel and the slag of the various kinds of elements inside the furnace as constraint conditions, and then propose the mixed algorithm of Differential Evolution and quantum-behaved particle swarm optimization algorithm to proceed high-efficiency solutions. The practical production data and site experimental model comparatively confirm the accuracy of predicted amount of the needed raw materials for production. As a matter of fact, the hitting rate of the end-point reaches 87.4%, and the deviation between calculated amount of material and actual amount is less than 4.5%, and the average is under 3.1%.(3) The dynamic operation optimization problem for converter smelting production process requires the dynamic optimization and correction for the blowing and smelting operation parameters based on the collected information, so as to realize the scheduled goal. Nonetheless, the difficulty and limitation to detect the information of the molten steel quality would lead to the incomplete quality information. To solve this problem, a dynamic data analytics method based on LS-SVM is proposed and the dynamic prediction model for the molten steel quality is established. The Improved Particle Swarm Optimization (IPSO) is formed to optimize parameters of the Least Squares Support Vector Machine (LS-SVM). The real production data proved the prediction performance and the effectiveness of the model, which is able to provide accurate prediction for the molten steel quality continuously, with the maximum deviation of carbon content less than 0.237%, and that of temperature under 3.5%.(4) A new kind of dynamic operation optimization problems based on the dynamic prediction of the molten steel quality is proposed to improve the dynamic control of the converter smelting production process. To solve the problem that the converter smelting production is too complicated to conduct online control, the dynamic operation optimization model is built with the temperature increasing and decarburization speed as the control indices (desired value obtained through operation analytics), and the minimumdeviation between the real-time index value and the desired value as the objective, as well as the smoke deviation correction and parameters estimation as constraints. The operation variables include the total inputting oxygen amount, top blown mode, the total bottom-blown gas amount and the weight of the various kinds of additive materials. The improved mutation-based differential evolution algorithm is formulated to provide high-efficiency solutions, and work out the set points of the current operation variables, thus controlling the machine to realize the dynamic control for the converter smelting production process. The collected actual production data proves the effectiveness of the model and the algorithm. The minimum and the average hitting rate of molten steel carbon content are 0.9527% and 0.97% respectively, the maximum and the average deviation of molten steel temperature are 7.52 degree and 3.082 degree respectively.Finally, the dissertation also shares the vision about the highlight issues of the potential research work on the operation optimization for iron-steel smelting process. |
PDF Full Text Request