The common theme of Shipping industry is to detect abnormal condition of the diesel engine in an early stage and to prevent it from vital accidents for increasing the Safety and the Operation rate of the ship. Although various abnomal condition diagnosis devices have been developed up to the present, in all cases, temperatures and pressures of various parts of the engine are detected and engine performance is estimated and the abnormal condition diagnosis is performed by the Artificial intelligence (A1), Therefore, the accuracy of abnomal condition diagnosis is not always satisfactory. For the analysis of egine operating performance, the combustion analysis system, etc., have also been developed Howerver, at present, these systems are not fit for actual circumstance from the point of view of sensor's durability, etc.In case of the diesel engine as Gas pressure in cylinder will act on the crankshaft via the piston and connecting rod and is converted into Torque. In this paper, the author would like to introduce way of the abnormal condition diagnosis which by the observation of Torque fluctuations (Torsion angles) of the crankshaft. We put a detecting device for the crankshaft torsion angles to the test, and validated that approach.