基于神经网络的电站锅炉入炉煤粉含碳量在线测量研究

电站锅炉燃烧的煤种来源不稳定,经常与设计煤种偏差较大,造成锅炉出力不足、燃烧效率下降以及结焦等问题出现,影响锅炉燃烧的经济性和安全性。针对锅炉燃烧煤种的不稳定性,提出了一种基于静电法在线测量入炉煤粉含碳量的方法,研究静电信号与煤粉含碳量之间的关系。对煤粉颗粒静电测量过程中的几个影响因素进行了试验分析,并采集了足够具有代表性的数据,在此基础上建立了锅炉一次风管中入炉煤粉含碳量的BP神经网络模型。结果表明,所建立的模型预测效果较好,能较好预测进入锅炉燃烧的煤种。

Study on Online Measuring the Carbon Content of Injected Pulverized Coal of Utility Boiler Based on Neural Network

The sources of the coal types for the utility boiler combustion are not stable and are often deviated greatly from the design coal, which results in the insufficient performance of the boiler, the decrease of combustion efficiency, coking and other problems, thus influencing the economy and the safety of boiler combustion. For the instability of the coal types for boiler combustion, this paper proposes a method of online measuring the carbon content of the injected pulverized coal base on the electrostatic method and studies the relationship between the electrostatic signal and the carbon content of the pulverized coal. The influence factors in the electrostatic measurement process of pulverized coal particles have been analyzed through tests, and enough representative data have been collected. On this basis, the BP neural network model of the carbon content of the injected pulverized coal in the primary airpipe of the boiler has been established. The results show that the established model prediction has a good effect and can predict the coal types injected into the boiler.