Sensor Fusion Concept for Improved Rotational Speed Measurement in Small Engines

Future developments for small engines, e.g. engines for handheld working tools, like chain saws require the integration of ECU-systems for engine control. For small engines often only a rotational speed senor is available. The application of additional engine sensors is in many cases unwanted, e.g. due to cost aspects and additional wiring. The lack of sensor data requires tailored control strategies and signal processing techniques to infer information about the engine from the sensor data. E.g. for rotational speed sensors the Δω method has been proposed, where the load is estimated from the temporal variation of the rotational speed. This approach requires a rotational speed sensor with sufficient angular resolution. In this paper we present a simulation study for a sensor fusion concept to improve the temporal resolution of engine speed measurements for low cost engines by means of an additional vibration sensor. The rotational sensor of the engine is assumed to have insufficient resolution to determine variations of the rotational speed over an engine revolution. However, variations of the rotational speed of the engine also cause vibrations of the engine chassis. A vibration sensor can be used to pick up the vibration signal with high temporal resolution. As the transfer function between the variation of the rotational speed and the sensor readings is only approximately known, a sensor fusion concept for the rotational speed sensor and the acceleration sensor has to be applied, which combines the different measurements, while simultaneously estimating the unknown transfer function. We will use an extended Kalman filter for the data fusion and an autoregressive-model for the unknown transfer function.