| Electric wheel dump truck is a kind of high-power equipment with high technical content and high complexity,which integrates machinery,electricity and hydraulics.It plays an important role in open-pit mining.With the continuous improvement of the structural complexity of electric wheel dump trucks,more and more failure modes occur.The traditional artificial intelligence maintenance method only diagnoses and repairs a single electric wheel dump truck or a single subsystem,and the amount of diagnostic sample data is large.The quality of diagnosis is poor,and a new method for intelligent diagnosis is urgently needed.Therefore,this paper takes the electric wheel dump truck as the background,and studies the cooperative joint diagnosis of multiple electric wheel dump trucks.Firstly,the failure mechanism,system requirements and design scheme of the electric wheel dump truck are studied.Starting from the various subsystems of the dump truck,such as electrical system,diesel engine system,hydraulic system and auxiliary system,the failure mode analysis and fault tree modeling of the subsystems are completed,the relationship between each failure mode and the failure symptoms is summarized,and the multi-vehicle system is clarified.Demand for combined vehicle and ground diagnostics.Finally,an overall design scheme is designed in which data is collected by the multi-vehicle on-board data acquisition device,and the ground server performs real-time status monitoring and multi-vehicle collaborative diagnosis.Secondly,the multi-vehicle coordination-vehicle-ground joint fault diagnosis method is studied.The method of combining self-organizing feature mapping network and BP neural network is selected to carry out intelligent diagnosis of the underlying minecart subsystem,and the simulation verification is carried out through MATLAB.When the method is extended to complex failure modes,the accuracy is 79.9%,the diagnosis accuracy is low,and the failure mode cannot be accurately judged.Therefore,the bidirectional long-short-term memory network model is introduced for multi-vehicle collaborative diagnosis,and it is compared with other shallow neural networks for diagnosis and analysis,and finally the overall fault diagnosis architecture suitable for multi-vehicle collaboration is determined.Then,the calculation and optimization of multi-vehicle fault diagnosis confidence are studied and verified by simulation.Four methods,namely traditional calculation method,expert scoring method,failure mode,effect and criticality analysis method(FMECA),and FMECA method of fuzzy mathematical analysis are used to analyze the initial confidence calculation of the failure cause,and compare the advantages and disadvantages of various methods.And the scope of application,and finally choose the fuzzy FMECA method to calculate the confidence of multi-vehicle.The influence of fault occurrence time and fault occurrence times on confidence adjustment is studied and analyzed,and a multi-factor confidence optimization adjustment algorithm is designed,and the algorithm is verified by MATLAB.Experiments show that the algorithm is effective,reliable and stable.Finally,the software and hardware design and experimental verification of the multi-vehicle coordination-vehicle-ground joint diagnosis system are completed.In terms of hardware,the design of the data transmission board,the main control processor backplane and the data communication board has been completed.In terms of software,based on Web Storm software,using HTML language,relying on vue architecture to complete the front-end development of server software;based on IDEA software,using JAVA language to complete the development of server software back-end,with the design of MYSQL database for data storage.Finally,a ground experiment platform is built to simulate the sending of fault data,observe the server diagnosis effect,and verify the feasibility of the multi-vehicle joint diagnosis algorithm.This article has a total of 96 figures,42 tables,and 57 reference. |
PDF Full Text Request