Abstract
A marine diesel engine diagnostic system appears to be necessary because of the following considerations;
1) an uneconomical maintenance schedule
2) a less frequent and shorter overhaul period
3) a difficulty to recruit the skill personnel.
The object of this work is to develop some new diagnostic system for marine diesel engines by some acoustic sounds analyses, in particular with regard to the reliability, simplicity and cost of analyzing system and the treatment of information. But it is very difficult because of the engine's complex movements and the variation in its heat cycle of internal combustion.
This paper discusses the some kinds of analyses (the following three methods A, B and C) of the acoustic sounds due to changing load on an uniflow type marine diesel engine.
The method A (time series analysis) having a band-pass filtering can give the history of time series analysis and some basic informations of pattern recognition for the detection of engines abnormal conditions.
The method B (power spectrum analysis) is widely used to analyze complex time series data. The abnormal signals will be able to be detected as the deviation of the power spectrum levels of some frequency components comparison with normal ones.
The method C (three dimensional display) using the color pattern corresponding the above levels can show the informations with the time history for the power spectrum of many kinds of the running conditions.
The above three methods can offer some effective informations and basic data to a diagnostic system for marine diesel engines. Especially the method C should be noticed to give both the early recognition of the irregular time history and the detection of abnormal signals by color pattern.
