基于BP神经网络的火电厂技改项目风险评价

发布时间：2018-03-10 10:19

本文选题：技术改造　切入点：风险评价　出处：《昆明理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：电力行业是关系到国计民生的基础性行业。由于中国人口众多,国民经济体量大,中国既是能源生产大国,也是能源消费大国。尽管近年来我国水电、核电等能源得到了迅速发展,但目前在我国现有的发电产业结构中,火力发电仍然占有巨大的市场份额。随着环境保护压力不断增大和人类社会可持续发展的迫切需要,我国火电厂运用新技术、采用新设备进行技术改造的问题也越来越突出。火电厂技术改造是一个复杂的系统工程,存在着许多风险。因此,如何辨识、评判和管理火电厂技改项目的风险,是一个非常值得研究的课题。本文以ZX火电厂为研究对象,运用风险评估的理论和方法,开展对其典型的技改项目——汽轮机改造项目进行风险评价研究,旨在为该项目的风险管理提供参考。首先,按照风险识别过程运用检查表法开展了基本风险分析,在此基础上,选出15个二级指标作为评价指标构建了评价指标体系;其次,基于BP神经网络的理论,运用MATLAB数学软件建立了包含动量因子的优化的BP神经网络评价模型和函数工具箱的神经网络评价模型,并对各自的隐含层的节点情况作了分析,选择出最佳的隐藏层节点。通过对上述两个模型进行训练和测试,由于批量训练的方式准确率较高,因此选择批量训练方式。最后,把汽轮机技术改造的专家打分数据输入到训练后的含有动量因子的BP神经网络批量训练模型中进行仿真,通过仿真分析,得出汽轮机技改项目的风险为中级,进而为决策者提供参考,从而得到防范风险和监控风险的方法,为ZX电厂的技术改造提供了保障。
[Abstract]:The electric power industry is a basic industry related to the national economy and the people's livelihood. Because of China's large population and large national economy, China is not only a large energy producer but also a large energy consumer. With the rapid development of nuclear power and other energy sources, thermal power still occupies a huge market share in the current industrial structure of power generation in China. With the increasing pressure of environmental protection and the urgent need of sustainable development of human society, The technical transformation of thermal power plants is a complex system engineering, which has many risks. Therefore, how to identify, Judging and managing the risk of technical transformation project of thermal power plant is a topic worth studying. This paper takes ZX thermal power plant as the research object and applies the theory and method of risk assessment. In order to provide reference for the risk management of the typical technical innovation project, steam turbine retrofit project, the risk evaluation research is carried out. Firstly, the basic risk analysis is carried out according to the risk identification process by using the check table method, and on the basis of this, Fifteen secondary indexes are selected as evaluation indexes to construct the evaluation index system. Secondly, based on the theory of BP neural network, The BP neural network evaluation model including momentum factor optimization and the neural network evaluation model of function toolbox are established by using MATLAB software, and the nodes of each hidden layer are analyzed. Select the best hidden layer node. By training and testing the above two models, because the accuracy of batch training is high, we choose batch training method. Finally, The expert score data of steam turbine technical transformation is input into the BP neural network batch training model with momentum factor after training. Through simulation and analysis, it is concluded that the risk of steam turbine technical innovation project is intermediate. The method of preventing risk and monitoring risk is provided as a reference for the decision maker, which provides the guarantee for the technical transformation of ZX power plant.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM621;TP183

【参考文献】