The invention provides a method for diagnosing faults of a
diesel engine based on instant rotary speed clustering analysis. Instant rotary speed signals in the process of
diesel engine operation are collected, the collected signals are filtered,
noise interference is eliminated, main shaft instant rotary speed data are resolved according to a
top dead center signal and a
firing order of air cylinders, and the instant rotary speed information corresponding to each air cylinder is obtained;
time domain analysis and
frequency domain analysis are conducted on the processed signals,
time domain features and
frequency domain features of the instant rotary speed signals are obtained, and therefore a two-dimensional array is formed; clustering analysis is conducted through the random restarting K-means
algorithm, the performance states of the air cylinders of the multi-cylinder
diesel engine are horizontally compared, the different classification number K is set, clustering is conducted repeatedly, and the most significant clustering result is selected to serve as a final diagnosis result, so that the faulted air cylinder is diagnosed. When the random restarting K-means
algorithm is used for clustering, a large number of experiential parameters do not need to be set and influence caused by the experiment parameters on the clustering result is avoided; the faulted air cylinder can be rapidly and accurately positioned.