RELIABLE GASOLINE ENGINE KNOCK DETECTION SYSTEM

THE PROPOSED RESEARCH IS IN DEVELOPING A RELIABLE MEANS FOR DETECTING ENGINE KNOCK. ENGINE KNOCK RESULTS FROM THE PREMATURE DETONATION OF THE UNBURNED CHARGE IN THE COMBUSTION CHAMBER. THIS CAN CAUSE SEVERE DAMAGE TO KEY ENGINE COMPONENTS IN ADDITION TO DEGRADING DRIVABILITY, EFFICIENCY, AND CONTRIBUTING SIGNIFICANTLY TO ENGINE NOISE. TRADITIONALLY, ENGINE KNOCK HAS BEEN KEPT UNDER CONTROL BY REQUIRING HIGHER OCTANE FUELS AND MODIFYING COMBUSTION CHAMBER DESIGN. RECENTLY, HOWEVER, THE INDUSTRY HAS IDENTIFIED THE NECESSITY OF DEVELOPING (REAL-TIME) ADAPTIVE KNOCK CONTROL. THE GENERAL CONSENSUS IS THAT, CURRENTLY THE MOST VIABLE TECHNIQUE FOR SENSING KNOCK IS THE MEASUREMENT OF ENGINE VIBRATION. A CRUCIAL PROBLEM, WHICH MUST STILL BE OVERCOME, LIES IN THE EXTREMELY NOISY NATURE OF THE SENSED VIBRATION SIGNAL. IT IS HEAVILY CORRUPTED BY SPURIOUS VIBRATION, E.G. THOSE DUE TO PISTON SLAP, VALVE TRAIN, AND TIMING GEARS. CLASSICAL SIGNAL PROCESSING TECHNIQUES WHICH HAVE BEEN IMPLEMENTED THUS FAR HAVE NOT BEEN SATISFACTORY. IN THE PROPOSED RESEARCH, OPTIMA SYSTEMS WILL INVESTIGATE THE USE OF NEW RESULTS IN STOCHASTIC ESTIMATION THEORY, ADAPTIVE THRESHOLDING AND DYNAMIC HYPOTHESIS TESTING. REAL ENGINE DATA UNDER VARIOUS OPERATING CONDITIONS, INCLUDING INTENTIONALLY INDUCED KNOCK, WILL BE USED FOR CONSTRUCTING STOCHASTIC MATHEMATICAL MODELS AND A RELAIBLE KNOCK DETECTION ALGORITHM.