Design And Implementation Of Drilling Pump Fault Diagnosis Expert System Based On Dynamic Fault Tree

This thesis is supported by the major science and technology special project "development and application of intelligent drilling rig"(Project No.2019zdzx0030)of Sichuan Province.As the "heart" of drilling rig,drilling pump is prone to failure due to complex working environment and harsh conditions,which affects the normal operation of drilling rig.In order to ensure the efficiency of production,it is required to locate and solve the fault quickly and accurately in the early stage of the fault.At present,fault tree analysis is a common method to diagnose complex mechanical faults,which can not represent the situation that only certain components of drilling pump fail in a fixed order can cause system faults,resulting in inaccurate diagnosis results.In order to solve this problem,dynamic fault tree analysis method is selected to represent this kind of dynamic fault and diagnose it.At the same time,the establishment of expert system can make the staff quickly determine the fault and obtain professional maintenance knowledge in the absence of experts.The main research contents are as follows:Firstly,the drilling pump is selected as the research object of this thesis because of its bad working environment,complex fault conditions and difficult to diagnose.Through the guidance of experts and the relevant knowledge provided,the failure mechanism is analyzed.Taking the power end of the drilling pump as an example,the fault tree is established,and the effectiveness of the fault tree analysis method in fault diagnosis is verified.Secondly,in view of the fact that the fault tree can not represent the system fault caused by the failure of each component in the drilling pump in sequence,a dynamic logic gate is introduced into the fault tree to form a dynamic fault tree.Combined with the fault characteristics of drilling pump,the dynamic fault tree diagnosis model based on discrete Bayesian network with low computational complexity and high accuracy is selected,and its diagnosis model is improved.By adding the corresponding relationship of "rate parameter ?--partition n ",the reliability and accuracy of diagnosis are improved.Taking the hydraulic end of drilling pump as an example,the effectiveness of dynamic fault tree in fault diagnosis is verified.Thirdly,the traditional fault tree diagnosis model is combined with the dynamic fault tree diagnosis model of the improved discrete-time Bayesian network to form a hybrid fault tree diagnosis model,which is used as the fault inference engine of the fault diagnosis expert system;According to the characteristics of basic fault information and diagnosis model knowledge of drilling pump,the appropriate representation is designed.Finally,according to the design idea,the web application of MVC framework is established,and the fault diagnosis expert system with friendly operation interface,good operation experience and perfect function guidance is developed.Through an example,the guiding significance and practical value of the system in the fault diagnosis of drilling pump are proved.