The Research On Industrial Pulverized Coal Boiler Combustion Optimal Control Algorithm

With the rapid development of national economy, the demand for electricity resources is rapidly increasing. Thermal electricity production is still occupies a dominant position in our cont, in order to meet the demand for electricity of production and lives, it requires large amounts of coal resources annually, and produce large amounts of NOx and other pollution emissions. Industrial pulverized coal boiler combustion process is very complex, involving combustion, heat transfer, flow and other aspects of the process, the boiler combustion efficiency and pollutant emissions affected by air volume the wind speed, the furnace excess air ratio, coal type, unit load and other factors. Moreover, these factors generally affect each other, being staggered, further increased the complexity of the combustion process, therefore the combustion process is a non-linear, multi-variable complex system. Currently, the problems we need to solve is optimizing the combustion efficiency of the complex system and reducing the industrial emissions. We must use the advanced control and engineering optimization technology in the industrial process, improving combustion efficiency and reducing emissions as the dual goals of boiler combustion system.In the thermal process, there is often have such a process, which the adjustment time is very long, and its operating conditions is constantly changing, especially in the units which utilize automatic generation control technology (AGC Automatic Generation Control), those load changes according to demand on the grid, when the operating conditions changes, some of the optimal operating point will float up, which requires get the optimal operating point in a short period of time. However, the existing online optimization algorithms are essentially based on a static model, it cannot get an effective data until the system is stable, we can take the dynamic characteristics of the process into consideration to resolve this contradiction. This paper analyzes the power station boiler efficiency, which as a function of load and flue gas oxygen content. Considering the NO emissions, convert its punishment into the amount of consumption of coal to correct boiler efficiency, and the result as the optimization process objective function finally. In each period we identify the optimization objective function model, and determine the steady-state model of optimization objective function based on the recognition result, and enable the system to follow the changes of working conditions and re-determine the optimal operating point quickly.In this article, we introduce Hammerstein model structure as amended boiler efficiency model structure, adopting HOE_AMLS algorithm processing online parameter identification. Then, considering the steady-state model of each cycle and using BFGS quasi-Newton algorithm to optimize. So we can get the optimum of oxygen set point, which can make correct boiler efficiency be the maximum under different loads. The main Research and works of this paper as follows:1.Improving and optimizing On-line adaptive algorithm which proposed in paper[3], firstly, for the online identification model, adopting HOE-AMLS method instead of fading memory generalized least squares algorithm to enhances processing capabilities for real-time data in the system;2.For the multi-variable optimization algorithm, we use BFGS quasi-Newton algorithm to ensure that the searching direction is the decline direction of the objective function in the iteration point, quasi-Newton method can overcome the defects when Hess matrix is not singular, also can avoid huge computation problems of Hess matrix, and under certain conditions, such algorithms still has super-linear convergence rate.3.We study several main factors that affect the efficiency of the boiler deeply, and analyze the impact degree of those factors. By analyzing oxygen settings effect on the various heat losses, we convert the influence of the boiler heat loss on the thermal efficiency into oxygen affect, and establish the general model structure and verify its practicality and versatility by simulation.4.Combining combustion optimization in the monitoring layer line with oxygen control in process control layer, completing the online pulverized coal boiler combustion optimization algorithms and proved by simulation.