摘要
在解决抽油机电力故障诊断问题的过程中,为提高知识表达的准确率和推理精度,首先,定义生产异常规则结构(PAC结构),并改进动量法,采用BP神经网络实现对特征变化趋势和规律的描述;其次,采用规则-知识衍生方式,提出多级框架结构,设计抽油机电力故障知识库结构;最后,采用正向推理方式,提出逆向定位算法,应用论据累积的贝叶斯方法实现冲突消解,完成基于混合智能技术的抽油机电力故障诊断方法(Him F方法)的研究,以此达到精确描述抽油机电力故障原因,扩展知识表达方式,提高故障诊断速度和准确率的目的。最后结合油井生产开发过程,利用Him F方法实现抽油机生产电力故障诊断系统的设计。
In order to improve the accuracy and accuracy of the knowledge expression in solving the problem of pumping unit power fault diagnosis,firstly,the production rule of abnormal structure（PAC structure）is defined,and the momentum method is improved. The BP neural network is used to describe the trend and law of feature change. Secondly,a multi-stage frame structure is proposed by using rule knowledge derivation,and the structure of power fault knowledge base of pumping unit is designed. Then,using the forward reasoning method,the reverse localization algorithm is proposesd. Bayesian method argument cumulative can realize conflict resolution,complete fault diagnosis method of pumping oil machine power based on Hybrid Intelligent Technology（Him F）research,to accurately describe the reasons for the power failure of the pumping unit,expand the knowledge expression mode,and improve the speed and accuracy of fault diagnosis. Finally,combined with the development process of oil well,the Him F method is used to design the power fault diagnosis system of pumping unit.
作者
高伟华
GAO Weihua(Customer Service Center,Daqing Power Supply Company of State Grid,Daqing 163300)
出处
《计算机与数字工程》
2018年第9期1905-1910,1870,共7页
Computer & Digital Engineering
基金
黑龙江省自然科学基金面上项目(编号:F2015020)
大庆市指导性科技计划项目(编号:zd-2016-010)资助
