A diesel engine with a misfiring cylinder is likely to break down in a short time when the engine is used without maintenance. The authors have developed a method of detecting the occurrence of cylinder misfiring and specifying the misfiring cylinder by using wavelet analysis and statistical analysis on the root mean square (r.m.s.) values of two kinds of acceleration waveforms. This detection method is considered superior to traditional methods, which require several accelerometers, especially in terms of simplicity. Since the development of the method, research has been conducted on semi-automatic processing of measured signals and signal transmission from the worksite to a remote office. Field tests confirmed that 1_x000D_
105 points of dynamic data could be transmitted through the mobile multimedia network within 15 minutes, and the time for signal processing was within 15 minutes. Final determination on the condition of a diesel engine in heavy machinery could be achieved within 30 minutes. Shortened processing time is expected by improvement of the processing software.