Using a fuzzy logic decision system to detect engine leaks

Fuzzy logic is a valuable tool to use in making decisions where a binary output is not desirable. In engine testing, a simple pass/fail decision does not account for situations where an engine could only marginally fail a test that is based on a single control limit. In this application, we design a system to determine an appropriate leak rate limit for an engine oil system test. Although the rate is affected by a variety of components, which can be present in several different combinations, the leak rate is currently compared against one limit for all engine configurations. Fuzzy logic allows us to predict leak rate based on what components are on the engine, and also to determine if the rate is acceptable, unacceptable, or requires further investigation. Results from testing show that the fuzzy system is more accurate than the old system in detecting leaks in engines with a normally low rate of air flow. In addition, the fuzzy system predicts a higher leak rate for engines with naturally higher air flow, allowing them to pass the test.