Estimation of fuel flow for telematics-enabled adaptive fuel and time efficient vehicle routing

This paper reports the development of vehicle fuel flow estimation algorithms based entirely on signals available through the standard OBD-II interface. The paper also illustrates the use of the resulting fuel flow estimates for adaptation and optimization. The fuel flow estimation algorithm functionality differs depending on the powertrain type (gasoline versus diesel, naturally aspirated versus boosted, conventional versus hybrid electric, etc.). To facilitate fuel and time efficient vehicle routing, an adaptation algorithm based on the recursive least squares (Kalman filtering) is defined. This adaptation algorithm learns the expected values and the variances of fuel consumption and travel time from multiple drives of a given vehicle over a given route segment. The use of adaptation from data reduces the need for accurate predictive modeling of vehicle fuel consumption and travel time which depend on difficult to predict and incorporate into the model traffic conditions, topographical road information, weather conditions, and inherently present vehicle-to-vehicle, driver-to-driver and fuel variability. The use of the adaptive models for optimization of vehicle travel is showcased with a simple example of optimizing time of day of departure decisions for a service vehicle. Finally, the use of a large interconnected network of adaptive models for vehicle fleet operation optimization is discussed.