Sensor Fusion Approach for Dynamic Torque Estimation with Low Cost Sensors for Boosted 4-Cylinder Engine

As the world searches for ways to reduce humanity\u27s impact on the environment, the automotive industry looks to extend the viable use of the gasoline engine by improving efficiency. One way to improve engine efficiency is through more effective control. Torque-based control is critical in modern cars and trucks for traction control, stability control, advanced driver assistance systems, and autonomous vehicle systems. Closed loop torque-based engine control systems require feedback signal(s); indicated mean effective pressure (IMEP) is a useful signal but is costly to measure directly with in-cylinder pressure sensors. Previous work has been done in torque and IMEP estimation using crankshaft acceleration and ion sensors, but these systems lack accuracy in some operating ranges and the ability to estimate cycle-cycle variation. In this study, we show that a data driven system to estimate IMEP using frequency content of crank acceleration, exhaust pressure, and ion current with optimized data windowing can effectively estimate individual cylinder cycle-cycle variation in IMEP over some engine operating regions. Fourier Transforms are used to extract features from the angle domain sensors that are useful for IMEP estimation. A neural network is used to estimate IMEP from those features. Pattern search and grid search algorithms are used to optimize feature extraction, network structure, and network training hyper-parameters with dual objectives of minimizing error and network complexity. These derivative free optimization techniques drove the IMEP estimation error down to 16 kPa over a transient drive cycle (using production possible sensors) and allow it to estimate cycle-cycle variation in some conditions