Publication

A previously validated open-source head-neck model, VIVA Open HBM [1-3], was enhanced by the addition of muscle activation. The previous model contained 129 beam elements on both the left and right side of the neck. Each element had Hill 3 elements (*MAT_MUSCLE_156) [4,5] definitions for material model. However, these muscles were defined as passive muscles model without any active tensile forces.
The first goal of this study was to implement the LS-DYNA PID (Proportional Integral Derivative) feedback control mechanism [5,6] on Finite Element (FE) models of cervical muscles, using the enhancements introduced in LS-DYNA version 9.2 [5]. The second goal was to calibrate the PID control gains by conducting a parameter identification using LS-OPT [7] with published-volunteer data in rear impact collisions [8-9] being used as reference performance data.
To activate these muscles, the closed-loop control strategy (reflexive feedback control function) was applied. To develop this strategy, the LS-DYNA PID Control function (PIDCTL) which can be defined inside the *DEFINE_CURVE_FUNCTION keyword was utilized. The method of controlling muscles activation with reflexive feedback control was adopted from Östh, et.al [10] and Olafsdottir [11].
To mimic the human body’s vestibular system, the coordinates of two nodes (Head Center of Gravity node and T1 node) and a reference node were sampled at specific times reference and used to define the controller vector. An angle was calculated between these two vectors and the difference between this angle and a reference value was calculated. This error signal then was delayed, mimicking the human’s neural delay. To model this neural delay, the DELAY function inside *DEFINE_CURVE_FUNCTION was applied. The PID controllers were given the delayed-error signalfrom the previous calculation and used to compute a control signal with an objective of zero error.
A calibration study was conducted to identify reasonable gain values of the controller so that the head displacements of the model in X and Z direction match within ± 1 Standard Deviation (SD) head displacements of the volunteer data.
The simulation results show that the LS-DYNA PIDCTL and DELAY function were successfully utilized for controlling the model muscle’s activation and by using controller’s gain value from the calibration’s result, the model was able to capture the kinematics motion of the volunteer data within the defined-corridors