Abstract: A fault diagnosis strategy was proposed based on ensemble empirical mode decomposition of crankshaft instantaneous speed signals to handle the faults of single-cylinder fire and double-cylinder fire for a ten-cylinder V-type diesel engine. This strategy was arranged to analyze the impact of multiple speed conditions on fire fault diagnosis. Firstly, the real-time data collected from the engine at various speeds were categorized into three operating conditions, normal, single-cylinder fire, and double-cylinder fire, based on the diesel engine's fuel injection advance angle. Preprocessing the data of three operating conditions with the multi-cycle averaging method and decomposing the data with the ensemble empirical mode decomposition method, the fault diagnosis system was designed to decompose adaptively the crankshaft speed signal into several intrinsic mode functions. And then, according to the abnormal fluctuations in the amplitude of each intrinsic mode functions obtained with ensemble empirical mode decomposition, the intrinsic mode functions with fault information was identified. In order to further extract features, the intrinsic mode functions was processed with fast Fourier transform to obtain the fault characteristics based on the amplitude of the dominant frequency component. Finally, the diagnostic process was conducted under multiple speed conditions to derive the diagnostic accuracy for each speed condition, enabling a sensitivity analysis of the diagnostic algorithm to different speed conditions. The experiment results show that this method can effectively extract fault features and achieve fire fault diagnosis in a ten-cylinder diesel engine based on multiple instantaneous speed measurements.