|
Modified Recurrent Fuzzy Neural Network Sliding Mode Control for Nonlinear Systems |
|
Ming Yang, Qianwen Xie, Jienan Han(Hohai University, China) |
| In this study, a modified fuzzy neural network sliding mode control for a class of nonlinear systems is designed. Firstly, the considered nonlinear system is given and a global sliding mode control is proposed. Then, a modified fuzzy neural network (FNN) is constructed and utilized for estimating the uncertain function. Compared with the conventional FNN, the designed FNN can obtain a better generalization capability with two feedback loops.The simulation results demonstrate that the designed control method can achieve superior control capability. |
|
|
Position Estimation of Stepper Motor Using Adaptive Gain Super Twisting Algorithm Sliding Mode Observer |
|
Hyun Uk Son, Yong Woo Jeong, Chung Choo Chung(Hanyang University, Korea) |
| This paper presents an Adaptive Gain Super Twisting Sliding Mode Observer (AGSTA-SMO) for a permanent magnet stepping motor as position. Since the proposed algorithm has a different structure with the Super Twisting Algorithm Sliding Mode Observer (STA-SMO), the AGSTA SMO ensures a global, finite-time convergence even with the unknown, bounded perturbations/uncertainties. With the experimental validation, we show that the position estimation performance of AGSTA-SMO outperforms comparing to the position estimation result of STA-SMO. |
|
|
Driver Torque Control for Electric Power Steering Systems with Parameter Estimation Algorithm |
|
Gwanyeon Kim, Wonhee KIM(Chung-Ang University, Korea) |
| This paper propose driver torque control method for electric power steering (EPS) system. The torque-based EPS model is established to improve the torque control performance by modifying the general EPS model. The driver torque is estimated by extended state observer (ESO) because the torsion-bar torque which measured by TAS sensor has delay problem and different from driver torque in transient response. The parameter estimation algorithm is designed to
estimate the accurate driver torque. To achieve the robustness against disturbance, another ESO is designed to estimate the lumped disturbance. Additionally, the simple state feedback control method is designed to track the desired torque. |
|
|
Gain Optimization for SMCSPO with Gradient Descent |
|
Jin Hyeok Lee, Hyeon Jae Ryu, Min Cheol Lee(Pusan National University, Korea) |
| There is a previous method that Sliding Perturbation Observer (SPO) estimates states and perturbation of the system and makes the system with fewer gains by compensating it to the SMC. In SMCSPO, proper parameter adjustment is necessary to determine tracking performance. In this study, the algorithm is presented to optimize this control parameter by using the Gradient Descent Method with an iterative experiment. This optimizing method lets cost function with tracking errors of acceleration minimized as the iteration of experiments increases. A simple motion test of SMCSPO for a two-link robot applied with dynamics is simulated by MATLAB programming to prove this optimizing algorithm. |
|
|
Improved Adaptive Sliding Mode Control with Disturbance Observer for Velocity Control of PMSM System |
|
JaeYun Yim(Chung-Ang University, Korea), Sesun You(Chung-Ang Unversity, Korea), Youngwoo Lee(Chonnam National University, Korea), Wonhee Kim(Chung-Ang University, Korea) |
| In this paper, we propose the improved adaptive SMC for control velocity of PMSM. The proposed ASMC uses DOB algorithm to compensate for disturbances. In general, high frequency disturbances degrade DOB estimation performance. However, the proposed ASMC is designed using disturbance error calculated by DOB algorithm. It ensures robustness regardless of disturbance estimation performance. In conclusion, proposed method can select the control gain as small as possible even though there is external disturbance. |
|
|
Design Framework for Context-Adaptive Control Methods Applied to Vehicle Power Management |
|
Sudarsan Kumar Venkatesan, Rian Beck, Sorin Bengea, Joram Meskens, Bruno Depraetere(Flandersmake, Belgium) |
| The paper presents a combined data-driven and model-based design framework for context-adaptive algorithms applied to the power management of a hybrid electric vehicle. Focused on this application, a set of predefined driving contexts are derived from multiple standard driving cycles. The first contribution consists of computing optimal control parameters associated with power management for each driving context. The second contribution consists of customized supervised classification algorithms for estimating driving contexts. The context-adaptive tool chain integrates context estimation and subsequent adaptation of control parameters. |
|