Online determination of temperature model parameters of a Selective Catalytic Reduction System

Abstract

Emission reduction, this expression is one of the most frequently used in the last years. Its meaning is known very well for the ordinary people, ordinary in the sense of being not associated with the automotive industry. One of the strategies for NOx emissions reduction in personal vehicles with combustion engine is the Selective Catalytic Reduction System. The critical issue of such systems can be too high urea temperature at the injector outlet. Engine compartment around the injector influences very much the injector temperature. This temperature can be estimated with a model based on interpolation maps and correlation factors. Parameters of the model can be evaluated using genetic algorithms. Main issue of this work was researching for best identification method of the parameters of the engine compartment. First, easy genetic algorithms has been designed and coupled to the temperature model, than two experiment have been defined to check various method, where changing of settings of genetic algorithms and then several runs have been performed. To evaluate which setting are most suitable, spread of the found parameters have been compared. In first two experiment absolute values of the parameters have not been exposed, third experiment has been introduced to validate short the chosen method. Output of this experiment are time course of the temperature throughout the whole simulation duration, population fitness and found parameters. The identification was running parallel to the process, e.g. during the system calibration phase in a car. Matlab Simulink environment was applied for design of the function. As a development platform for running of the algorithm and doing measurement a dedicated Rapid Prototyping System based on Micro Auto Box from dSpace was applied.