Development of optimized model for filter and integrator of knock detection integrated circuit in automotive engine management system

In this work two main blocks of integrated circuit for knock signal detection has been modeled. The main goal of modeling key components of knock detection IC is to provide a software tool in order to reduce the number of experiments during calibration of knock parameters in ECU. In conventional method experimental data requires for calibration of knock related parameters for each central frequency of the filter and for several operating conditions of the engine. Since there are several parameters such as beginning and end of knock window phase, time constant of Integrator circuit, input signal gain and filter central frequency, the calibration task requires lots of measurements which is time consuming and expensive. Therefore we propose a behavioral model for two main blocks of knock signal detection integrated circuit, which enable us to use simulation instead of measurement for calibration purpose. A model of knock IC filter has been developed based of frequency response analysis, and then this model has been optimized using genetic algorithm. Similar technique has been used to model the integrator block of the IC. Comparing the model simulation results with experimental data under different engine speed and load and at several central frequencies demonstrates that the model has a reasonable accuracy with maximum error of 4.78%