The selection and configuration of sensors can strongly influence the closed-loop dynamics of a system. Therefore a methodology for finding the best sensor placement is a valuable tool. This paper deals with this problem by formulating an optimization problem and applies the new method on an SI engine.
The best sensor configuration is one that minimizes the overall system costs, yet still meets the system constraints. Before solving the optimization problem, the system is modeled, different sensor configurations are defined, the appropriate controller and the feedback term are developed, and the locations and size of the various errors present in the model are determined. Then, the objective function and the system constraints are defined and the optimization problem is solved considering the worst-case combination of modeling errors, which is computed using genetic algorithms. The objective function is defined as the sum of the sensor costs and of a penalty term. This term describes additional costs that depend on the control quality achieved, which is a function of the worst-case modeling error combination.
The new method is then applied to the case of the AF ratio control in an SI engine. In this application various sensor configurations are possible. The feedforward term is important for a fast compensation of changes in the engine's operating point, but due to the complexity of the plant various modeling errors are present that can strongly influence its effect. The objective function includes the costs of the sensors and of the catalytic converter. The latter represents the penalty term that depends on the performance of the control structure and therefore on the worst-case modeling error combination. The larger the transients are, the larger the catalytic converter has to be for fulfilling the emissions legislation and the higher the resulting costs are. The proposed method is then validated with test bench measurements.