基于BP神经网络的燃煤锅炉温度分布预测

Prediction Model of Temperature Distribution in Combustion Zone of Coal-fired Boiler based on BP Neural Network

燃煤锅炉燃烧运行调整的主要方法是基于尾部烟道参数的负反馈调节法,为解决该方法调节灵敏度低、精度差且滞后严重的问题,利用Fluent数值模拟计算得到炉膛内部温度分布,并通过现场热态实验数据对模拟结果进行验证;结合BP神经网络算法搭建燃烧区域温度预测模型,根据预测结果绘制炉内燃烧区域温度分布云图,实现运行参数改变时炉膛燃烧区域温度分布快速、高精度及可视化预测。结果表明:该预测模型基于对已知工况模拟结果的训练,能够对未知工况温度场进行准确预测,预测结果相对偏差为4.11%。

At present,the main method of burner adjustment during the operation of coal-fired boilers is based on the negative feedback adjustment method of the parameters of the tail flue system.However,this method has low sensitivity,poor accuracy and severe hysteresis.Therefore,through the temperature data calculated by numerical simulation software and actual temperature test verification,a calculation model of combustion zone temperature based on neural network was proposed.Then,according to the calculation results,the temperature field of the combustion zone was built to reflect the actual combustion inside the furnace as much as possible.Fast,high-precision and visual prediction of temperature distribution in the combustion zone of the furnace under unknown operating conditions was achieved,which can provide guiding suggestions for feed forward adjustment of the burner.The results show the prediction model based on the training of simulation results of known working conditions can accurately predict the temperature field of unknown working conditions.The relative deviation of the forecast results is 4.11%.