Neural Network Transmission Control

This Application is using Multilayer Perceptron Algorithm to predict the shifting gear timing based on throttle percentage, vehicle velocity, time history and engine speed, in order to enhance the fuel efficiency, shifting time, power loss and driver’s comfort during shifting. The model makes no assumptions about how transmission performance aspects like gears slipping time or the existence of distinct environments like high temperature affects the power train system performance. Instead, the model can learn to model any type of distortion or additive noise in the sensor data; induced in the CAN protocol, through sufficient training data characterization. The network will be implemented on an AMESim model to characterize the operating temperatures and optimize the gearbox design to work with an optimal temperature and advance to this temperature during operation. Multilayer Perceptron Algorithm possesses important properties such as many degrees of freedom, non-linearity, and dissipations. The identification of stability attractors of the lyapunov function with associative memories and input - output mappers is one of the foundations of neural network paradigms through controlling the locations of the attractors in the state space of the system. The Algorithm role to take the form of the nonlinear dynamic equation that manipulates the locations of the attractors for the purpose of encoding information in a desired form. In this way, it is possible to establish an intimate relationship between the physics of the transmission and the algorithms computations.