热电厂锅炉燃烧系统建模及优化研究

Research on modeling and optimization for the power plants boiler combustion system

锅炉的燃烧系统是一个复杂的动态机理系统,各控制回路和变量之间存在非线性、强耦合以及大滞后等特点,利用传统的机理建模和常规的比例、微分、积分(PID)控制策略,并不能有效实现锅炉燃烧系统的优化.从基础的锅炉运行数据出发,结合锅炉燃烧系统非线性的特性,提出一种基于径向基(RBF)神经网络建模和序列二次规划法(SQP)优化算法的智能控制策略.即结合锅炉的燃烧特性和主要影响变量,建立一个4输入、3输出的RBF神经网络模型,在建立精确模型的基础上,利用SQP算法对锅炉燃烧过程的主要影响变量进行寻优操作,从而达到锅炉燃烧优化的目的.试验仿真和实际运行结果表明,RBF神经网络建模及SQP优化算法方法优于传统的机理建模和PID控制的策略.

The boiler combustion system is a complex dynamic mechanism system, which has the characteristics of nonlinear and strong coupling and large delay between the control cir-cuit and the variable. It cannot optimize the boiler combustion by using the traditional mech-anism modeling and conventional proportional integral differential （PID） control strategies. This article embarks from the data base of the boiler. Combined with the nonlinear charac-teristics of boiler combustion system, it proposes an intelligent control strategy which is based on the RBF neural network model and the sequence of two quadratic programming method （SQP） optimization algorithm. According to the combustion characteristics and main effect of boiler variables,it establishes a RBF neural network model with 4 inputs and 3 out-puts, and takes the optimization operation of the main effect of boiler combustion process variables using the SQP algorithm based on the accuracy of the model. Through these operations, the purpose of combustion optimization is achieved. Experimental simulation and practical operation results show that the proposed RBF neural network modeling and SQP optimization algorithm are superior to the traditional mechanism modeling and PID control strategies.