ADAPTIVE OPTIMIZATION OF EMISSIONS AND FUEL CONSUMPTION OF AN INTERNAL COMBUSTION ENGINE

In recent years air pollution has become a major societal problem and automotive emission has become a major international concern. With the advent of fuel shortages and rapidly rising fuel prices we are also faced with the problem of maximizing fuel economy. Reducing automobile emission and improving fuel economy are conflicting goals. Engines must be designed and controlled to obtain an appropriate compromise between these conflicts without unduly sacrificing performance and driveability.; The objective of the research for this dissertation was to develop adaptive on-line optimization techniques for a solution of the engine control problem. In this research a typical engine and vehicle were simulated on a computer. A recursive least square identification technique was used to estimate the parameters of a linearized discrete model of this simulated system. A discrete multivariable optimizing controller was designed using this linearized model. The adaptive on-line technique optimized the control input air-fuel, spark advance, exhaust gas recirculation rate, and throttle position to minimize a performance index. The performance index in this study was a weighted sum of the squares of the fuel consumption rate, exhaust emission rates, and the deviations of the engine and the vehicle speeds from specified values.; The solution for this optimal control problem lead to a two-point boundary value problem. The technique is applicable to general driving conditions and the control is found which minimizes the performance index over a succession of short-time segments.; The effect of changing the weighting factors in the performance index on the outputs was evaluated. These results showed fuel consumption was reduced by 9% when its weighting factor was raised from 1 to 5. In the case of exhaust emissions, the NO(,x) emission produced per distance travelled by the vehicle was reduced by 9% as its weighting factor was changed from 1 to 9. The CO and the HC emitted by the vehicle per distance travelled were also reduced by 23% and 15% as their weighting factors changed from 1 to 5, respectively. The effect of the forgetting factor in the identification algorithm was also investigated.