An Investigation On The Diagnosis For Abnormal Sound Fault In Motorcycle Engine Valve Train

Motorcycle engine is a typical complicated system with multiple layered hierarchical structure Limited to fault diagnosis application and technology, fault diagnosis in present motorcycle production industry still remains in man-measured stage. Its low efficiency can't satisfy large-scale modern production's requirement. Developing intelligent fault diagnosis system for motorcycle engine aims to realize automation and intelligence of fault diagnosis with great value.Based on current fault diagnosis state and valve train entity of motorcycle engine, the paper gives a detailed analysis of valve train vibration and noise of Motorcycle 125 as well as an elemental discussion of its abnormal noise. The fault simulating empiric analysis is done, using fuzzy-logic theory to set up fault diagnosing rules and presenting the expert-system, the nucleus technology based on virtual apparatus. Introducing the valve train constitution and its single-independent degree dynamic model, the paper gives a detailed analysis of its vibration theoretically, completing its noise experiment and analyzing its characteristic. According to the above analysis, the feasibility to use the surface vibration signals of cylinder head to diagnose the valve train faults is discussed, completing the valve train fault simulating experiment and building the method and principles for abstracting the fault characteristic. Based on the designing task, the paper gives a comparative analysis of current fault diagnosing methods such as reasoning-based, model-based, artificial neuron-based and case-based, The paper presents an intelligent diagnosing system composed of expert system and virtual apparatus where virtual apparatus is used as man-machine conversation interface to collect data and call expert-system and expert-system is the core for diagnosis analysis, elementally projecting the diagnosis system. Presenting a machine self-learning method with membership function coefficient, the offline and online learning ability for the fault diagnosing system is strengthened, thus promoting the accuracy and learning ability of the diagnosing system's determination.