再热汽温神经网络预测模型的构建及仿真

在火电机组运行过程中,提高再热汽温预测精度对于提高机组运行经济性、可靠性具有重要意义。影响再热汽温的因素错综复杂,采用传统方法难以建立精确的数学模型,或模型预测精度不高。混沌理论的发展为这一问题的研究提供了新的思路。本文在揭示再热汽温混沌特性的基础上,利用混沌特性处理输入样本及确定神经网络结构,将神经网络与改进型遗传算法结合,构建了基于改进型遗传算法的再热汽温神经网络预测模型。仿真结果表明,该模型相对误差的最大绝对值仅为0.068 12%,收敛速度快。

Modeling and Simulation of Reheated Vapour Temperature Based on Neural Network Predicting Model

It is very important to improve the predictive precision of reheated vapour temperature for enhancing the economy and reliability of the generating unit in power station.However,precise mathematical model is hard to build by traditional methods and model's predictive precision isn't high enough because the reheated vapour temperature is under the influence of many factors.Chaotic theory supplies a new method to solve the problem.In this article,the chaotic dynamic performance of reheated vapour temperature is revealed.Chaotic performance is used to deal with input samples and determine the structure of neural network.Neural network is combined with advanced genetic algorithm to construct a new predicting model of the reheated vapour temperature.Simulation results show that the model has higher precision and faster convergence speed,and the maximum absolute value of the relative error for the model is 0.06812%.