Research On Fault Diagnosis Of Diesel Engine By Neural Nets Based On The Vibration Signals

As the familiar power equipment, the diesel engine affects the safety of operation system because of its reliability and dynamic characteristics. Effectively discovered and removed the faults in time, the diesel engine could increase its security and reliability, reduce upkeep costs, decrease economical losing, and could avoid bigness accident. At present, it is being studying moment on fault diagnosis of diesel engine. And it has very theoretic sense and practical value to diagnose the fault of the diesel engine by artificial neural nets.After analyzed the force of the vibration, a sound test system is designed, and the vibration signals are measured on the type X4105BD2 under the fuel system in order and five fault states of three working conditions (900rpm,empty load;rating speed,empty load;rating speed,over load). The feature parameters on time domain can be the fault sign vectors to study the fault diagnosis of diesel engine when they are picked-up from the signals.The six BP neural nets are established using the head and side vibration signals with the MATLAB software. And the training speeds are all rapidness after twelve swatches were used to train each net time after time. One swatch is used to prove in each states, point lays and working conditions, and the neural nets could draw right judgements except very few oddity signals. So, it is said that the method is feasible.It has some influence to the training speed and practical application of the neural net by the signals on different test points' lays. But the convergence of the training speed is faster using the head signals than the side. So, we can draw a conclusion that it is good to diagnose the fault of the diesel using the head signals than the side. It can provide theoretic base for choosing the lays of the test points to diagnose the fault of the diesel using the neural net.The test dates are pretreated by the software of MATLAB of its strong function in operating numerical value and treating dates, and the feature parameters on time domain of the vibration signals are calculated by it. The BP neural nets are established and validated used the toolbox of neural net of MATLAB software based on the vibration signals to diagnose the fault of the diesel. The language of its programme is simple to understand and use, and its speed is so faster in operating dates and training nets.Because the signals are measured on the head and side in the test, the charge is socheaper than other methods on fault diagnosis of diesel engine. So on as it is simple to realize.