基于BP和RBF神经网络的机组能耗特性研究
发布时间：2018-12-06 16:11
【摘要】：火电机组在运行过程中不仅产生大量的历史数据,同时这些边界参数还与热耗率之间存在复杂的非线性关系。针对某电厂的实时数据,首先利用敏感性分析,从大量的机组运行参数中筛选出对机组能耗影响较大的重要参数:负荷、循环水入口温度、主蒸汽温度、再热蒸汽温度、主蒸汽压力、循环水流量。然后,对BP和RBF神经网络在热耗率与机组边界参数的应用进行了对比分析。训练和预测结果表明,BP和RBF神经网络都能对此进行分析研究,但RBF比BP神经网络的训练和预测的相对误差较小些,可以更准确地对机组热耗进行预测。为今后的可控参数优化提供了有效的模型,具有一定指导意义。
[Abstract]:Not only a large amount of historical data are generated during the operation of thermal power units, but also there is a complex nonlinear relationship between these boundary parameters and the heat consumption rate. According to the real-time data of a power plant, firstly, by using sensitivity analysis, the important parameters which have great influence on unit energy consumption are selected from a large number of unit operating parameters: load, inlet temperature of circulating water, main steam temperature, reheat steam temperature, etc. Main steam pressure, circulating water flow. Then, the application of BP neural network and RBF neural network in heat consumption rate and unit boundary parameters are compared and analyzed. The results of training and prediction show that both BP and RBF neural networks can analyze and study this problem, but the relative error of training and prediction of RBF neural network is smaller than that of BP neural network, so it is more accurate to predict unit heat consumption. It provides an effective model for the optimization of controllable parameters in the future and has certain guiding significance.
【作者单位】： 华北电力大学能源动力与机械工程学院;国电怀安热电有限公司;
【基金】：中央高校基本科研业务费专项基金资助项目(12NQ40)
【分类号】：TM621
[Abstract]:Not only a large amount of historical data are generated during the operation of thermal power units, but also there is a complex nonlinear relationship between these boundary parameters and the heat consumption rate. According to the real-time data of a power plant, firstly, by using sensitivity analysis, the important parameters which have great influence on unit energy consumption are selected from a large number of unit operating parameters: load, inlet temperature of circulating water, main steam temperature, reheat steam temperature, etc. Main steam pressure, circulating water flow. Then, the application of BP neural network and RBF neural network in heat consumption rate and unit boundary parameters are compared and analyzed. The results of training and prediction show that both BP and RBF neural networks can analyze and study this problem, but the relative error of training and prediction of RBF neural network is smaller than that of BP neural network, so it is more accurate to predict unit heat consumption. It provides an effective model for the optimization of controllable parameters in the future and has certain guiding significance.
【作者单位】： 华北电力大学能源动力与机械工程学院;国电怀安热电有限公司;
【基金】：中央高校基本科研业务费专项基金资助项目(12NQ40)
【分类号】：TM621
【相似文献】
相关期刊论文 前10条
1 王雪峰,邬建华,冯英浚,王建元;运用样本更新的实时神经网络进行短期电力负荷预测[J];系统工程理论与实践;2003年04期
2 文汉云;;硫化氢燃烧的神经网络PID控制及其仿真[J];自动化与仪器仪表;2006年01期
3 张U，
本文编号：2366302
本文链接：https://www.wllwen.com/kejilunwen/dianlidianqilunwen/2366302.html
