Improvement of the reproducibility in consumption measurements by standardisation

摘要

The method involves recording input variables such as accelerator position, as fuel consumption during a test, and forming a neural network with the recorded variables, where the network specifies an output variable such as fuel consumption, after a training phase. The output variable is simulated by the network trained at a real vehicle using a standardized target curve and/or a speed curve. The simulated output variable corresponds to the real vehicle such that the vehicle accurately follows the target curve.