摘要
针对制粉系统存在的大惯性和大迟延等特点,提出了一种基于时序-神经网络的一次风量软测量模型。在建模过程中,考虑了生产过程输入变量和输出变量的时序,给出了辅助变量选取和数据预处理方法。某电厂实际运行结果表明,该模型的准确性较目前广泛应用的静态神经网络软测量模型有显著提高。该研究为磨煤机一次风量的测量提供了一定的理论基础。
In accordance with the characteristics of large inertia and large time lag of the coal purverizing system, the model of soft measurement of primary air flow based on time sequence - neural network is proposed. In the modeling process, the time sequences of input variables and output variables of productive process are considered, the methods of secondary variables selection and data pretreatment are given. The simulation verification of model is conducted using the practical historical operation data of certain power plant,the results indicate that comparing with the commonly used static neural network soft measurement models; the precision of this model is obviously improved. The research provides certain theoretical basis for the measurement of primary air flow of pulverizers.
出处
《自动化仪表》
CAS
2017年第1期46-49,共4页
Process Automation Instrumentation
关键词
火电厂
制粉系统
磨煤机
风量
软测量
时间序列
BP神经网络
人工智能
归一化处理
数据挖掘
Thermal power plant
Coal pulverizing system
Pulverizer
Air flow
Soft measurement
Time series
BP neural network
Artificial intellegence
Normalized processing
Data mining
