Study On The Diesel Engine Fault Diagnosis Method Based On KNN Algorithm

The diesel engine, as an important power machinery, is widely used in all fields of national economy. The diesel engine holds the extremely important status in ships as the main power source. The operating condition of diesel engine has a direct impact on the whole equipment. However, due to some characters of diesel, such as poor working conditions,numerous components and complicated motion, the diesel engine is easy to be faulty. Once a failure happening, the running state and the safe operation of the diesel engine will be affected.It would cause heavy economic damages,sometimes resulting in casualties. On this condition,it makes a lot of sense to study the technology of diesel engine fault diagnosis. In the paper,take the D114 diesel engine as the research object, the simulation model of diesel engine was established by using Boost software,use the simulation model,typical faults includes compression ratio decreasing, injection timing fault are simulated and calculated. A fault diagnosis method for diesel engine based on KNN algorithm is studied.Firstly, through the analysis of the diesel engine work process simulation, determine the research strategy of diesel engine performance simulation and fault diagnosis. D4114 diesel engine simulation model is established by using AVL BOOST software. Simulated different working conditions of the model, compared with the experimental data to verify the correctness of the model.Secondly, by using simulation model, eight typical faults which includes compression ratio decreasing, injection timing fault, cylinder fuel supply uneven, turbocharger efficiency decreasing, inflow and exhaust valve gap increase and crankcase blow-by are simulated, by means of quantitative analysis, the change rules of eight thermal parameters under different faults and different severity are acquired, a fault database of the thermal parameters under different faults is established.Finally, the feasibility to apply KNN algorithm to fault diagnosis of diesel engine is discussed. According to the insufficient of the KNN algorithm, such as strong dependence on sample set and classification speed is slow, improved KNN algorithm. Design fault diagnosis program based on improved KNN algorithm by Matlab. Acquire the distance between fault samples and the fault database and the probability between samples and each type. Based on this method detected the fault type. It is proved that KNN algorithm is effective. Then,established test bed of diesel fault diagnosis, program of operation parameters acquisition is designed through Lab VIEW. The experimental verification of the fault diagnosis of valve clearance is completed. It is proved that the fault diagnosis method of diesel engine based on KNN algorithm is feasible.