基于粒子群优化算法优化BP神经网络模型的间接空冷散热器性能监测

Performance Monitoring of Indirect Air-cooled Radiators Based on Particle Swarm Optimization BP Neural Network Model

为监测间接空冷散热器的换热性能,提出了监测间接空冷塔出水温度的方法。根据间接空冷系统散热器传热量计算和热平衡方程,分析了间接空冷塔出水温度的影响因素,建立了以环境温度、环境风速、大气压力、间接空冷塔循环水进水温度、循环水进水压力、出水压力和百叶窗开度7个主要参数为输入,出水温度为输出的BP神经网络模型。为避免该模型陷入局部最优,采用非线性动态惯性权重的粒子群优化(PSO)算法对BP神经网络模型的初始权值和阈值进行了优化,构建了PSO-BP神经网络预测模型,并根据某660MW间接空冷机组的运行数据对该模型进行了训练和验证。结果表明:采用PSO算法优化的BP神经网络模型具有较强泛化能力,预测精度高于单纯的BP神经网络模型,预测平均绝对百分比误差为0.55%。

To monitor the heat transfer performance of indirect air-cooled radiators,a method was put forward for monitoring the outlet temperature of related indirect air-cooling towers.Based on heat transfer calculation and heat balance equations,an analysis was conducted on the factors influencing the outlet temperature of the indirect air-cooling towers,and subsequently a BP neural network model was established by taking the ambient temperature,wind speed,atmospheric pressure,inlet circulating water temperature of indirect air cooling tower,inlet pressure of circulating water,outlet water pressure and the shutter opening as the input variables,and outlet water temperature as the outlet variable.To avoid obtaining partial optimal solutions only,the particle swarm optimization(PSO)based on nonlinear dynamic inertia weight was used to optimize the initial weight and threshold of the BP neural network model,and then a neural network prediction model of PSO-BP was set up,which was trained and verified using the operation data of a 660 MW unit indirect air cooling system.Results show that the BP neural network model optimized by PSO algorithm has strong generalization ability,with higher prediction accuracy than the pure BP neural network model,and the mean error predicted is 0.55%.