A Study on Modeling of Driver's Braking Action to Avoid Rear-End Collision with Time Delay Neural Network

Modeling of occupant’s head movement behavior in motion sickness study via time delay neural network

“…To sum up, the comparison between TDNN and ANN models with the same number of hidden neurons were carried out to analyze the proposed model's accuracy. 28 Figure 4 illustrates an overview of ANN and TDNN models structures which are implemented in this study. Basically, both feed-forward network models are organized in three interconnected layers, namely an input layer, a hidden layer, and an output layer.…”

“…30 In the LM algorithm, the training process optimizes the weights through iterations based on the input-output time series. 28 The back propagation (BP) algorithm consisted of the forward and backward paths. 31 In the forward path, the feedforward network was created, the weights were initialized, and the network was trained.…”

“…36 As TDNN utilized the present and past time-series data in the learning process, it has the number of delays that showed to what extent the present and past data had been input into the model. 28 The influence of time delay on the model's accuracy was investigated. In this study, the parameter of time delay is set to be 0 until 10 with an interval of 1 and 10 until 100 with an interval of 10.…”

Artificial neural network for modelling of the correlation between lateral acceleration and head movement in a motion sickness study

“…ANN has gained widespread popularity as a non-linear function approximation tool among researchers [12]. It is one of the optimal solutions derived through learning input and output data and has been commonly used in pattern matching, image recognition, signal processing and optimisation techniques [13]. A scenario which provokes MS is set up under a real experimental environment and the naturalistic data of the vehicle, driver and passenger are measured.…”