Force control of a magnetorheological damper using an elementary hysteresis model-based feedforward neural network

Ekkachai, Kittipong; Tungpimolrut, Kanokvate; Nilkhamhang, Itthisek

doi:10.1088/0964-1726/22/11/115030

Cited by 14 publications

(11 citation statements)

References 21 publications

(37 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The MR damper model is shown in Figure 1A. The model was trained by using data from the experimental system of an actual MR damper, Lord RD-8040-1, described in Ekkachai et al (2013).…”

Section: System Descriptionmentioning

confidence: 99%

Reinforcement Q-Learning Control With Reward Shaping Function for Swing Phase Control in a Semi-active Prosthetic Knee

et al. 2020

Self Cite

View full text Add to dashboard Cite

In this study, we investigated a control algorithm for a semi-active prosthetic knee based on reinforcement learning (RL). Model-free reinforcement Q-learning control with a reward shaping function was proposed as the voltage controller of a magnetorheological damper based on the prosthetic knee. The reward function was designed as a function of the performance index that accounts for the trajectory of the subject-specific knee angle. We compared our proposed reward function to a conventional single reward function under the same random initialization of a Q-matrix. We trained this control algorithm to adapt to several walking speed datasets under one control policy and subsequently compared its performance with that of other control algorithms. The results showed that our proposed reward function performed better than the conventional single reward function in terms of the normalized root mean squared error and also showed a faster convergence trend. Furthermore, our control strategy converged within our desired performance index and could adapt to several walking speeds. Our proposed control structure has also an overall better performance compared to user-adaptive control, while some of its walking speeds performed better than the neural network predictive control from existing studies.

show abstract

“…The MR damper model is shown in Figure 1A. The model was trained by using data from the experimental system of an actual MR damper, Lord RD-8040-1, described in Ekkachai et al (2013).…”

Section: System Descriptionmentioning

confidence: 99%

Reinforcement Q-Learning Control With Reward Shaping Function for Swing Phase Control in a Semi-active Prosthetic Knee

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…If there are N sets of training data, the dimension of y, A, and u are N 3 1, N 3 6 and 6 3 1, respectively. Here, 6 is the number of the consequent parameters.û, also called estimated u, is computed by the LSE method as shown in equation (17). Then, the output yˆfor the training data estimated by ANFIS is calculated according to equation ( 18) ŷ = Aû ð18Þ u is calculated to minimize the root-mean-square error (RMSE) between the overall output y and the estimated output yˆfor the training data.…”

Section: Sfla-based Anfis Strategy For Inverse Modelingmentioning

confidence: 99%

“…On the other hand, the non-parametric inverse modeling techniques for the MR damper have received considerable attention recently. These techniques mainly include fuzzy logic system, 14 artificial neural networks (ANN), 3,[15][16][17] and adaptive neuro-fuzzy inference system (ANFIS). 4,18 As for the fuzzy modeling technique, it has attractive advantages in handling uncertainties, high nonlinearities, and heuristic knowledge.…”

Section: Introductionmentioning

confidence: 99%

Optimal inverse magnetorheological damper modeling using shuffled frog-leaping algorithm–based adaptive neuro-fuzzy inference system approach

Lin

Chen

2016

Advances in Mechanical Engineering

View full text Add to dashboard Cite

Magnetorheological dampers have become prominent semi-active control devices for vibration mitigation of structures which are subjected to severe loads. However, the damping force cannot be controlled directly due to the inherent nonlinear characteristics of the magnetorheological dampers. Therefore, for fully exploiting the capabilities of the magnetorheological dampers, one of the challenging aspects is to develop an accurate inverse model which can appropriately predict the input voltage to control the damping force. In this article, a hybrid modeling strategy combining shuffled frogleaping algorithm and adaptive-network-based fuzzy inference system is proposed to model the inverse dynamic characteristics of the magnetorheological dampers for improving the modeling accuracy. The shuffled frog-leaping algorithm is employed to optimize the premise parameters of the adaptive-network-based fuzzy inference system while the consequent parameters are tuned by a least square estimation method, here known as shuffled frog-leaping algorithm-based adaptive-network-based fuzzy inference system approach. To evaluate the effectiveness of the proposed approach, the inverse modeling results based on the shuffled frog-leaping algorithm-based adaptive-network-based fuzzy inference system approach are compared with those based on the adaptive-network-based fuzzy inference system and genetic algorithm-based adaptive-network-based fuzzy inference system approaches. Analysis of variance test is carried out to statistically compare the performance of the proposed methods and the results demonstrate that the shuffled frogleaping algorithm-based adaptive-network-based fuzzy inference system strategy outperforms the other two methods in terms of modeling (training) accuracy and checking accuracy.

show abstract

“…The non-parametric models employ analytical expressions to describe the characteristics of MR dampers based on both testing data analysis and the working principles of the device. Various types of nonparametric models have been developed, including the neural networks model (Du et al, 2006;Ekkachai et al, 2013;Witters and Swevers, 2010;Zheng et al, 2011) and the fuzzy model (Ahn et al, 2009;Liem et al, 2015;Ahn, 2010, 2011). Chang and Roschke (1998) proposed a nonparametric model based on the multi-layer perception neural network integrated optimization method.…”

Section: Introductionmentioning

confidence: 99%

A hybrid model of magnetorheological dampers based on generalized hysteretic biviscous operators

Zhao

Pan

2018

Journal of Intelligent Material Systems and Structures

View full text Add to dashboard Cite

Magnetorheological dampers are promising vibration control devices that are widely used for vibration mitigation applications. However, due to the inherent hysteresis of magnetorheological dampers, achieving high performance is a challenging issue because it requires the development of models that can accurately describe the unique characteristics of this hysteresis. In this article, first, a generalized hysteretic biviscous operator is proposed to roughly describe the hysteretic property. Then, the superposition of the weighted generalized hysteretic biviscous operators is performed to model the magnetorheological damper, and modified particle swarm optimization is utilized to regulate the weights. This methodology provides the hybrid model with the advantages of both parametric and non-parametric models. Moreover, the model has a rather simple architecture and can be easily determined. The experimental results demonstrate the feasibility of the proposed method.

show abstract

Force control of a magnetorheological damper using an elementary hysteresis model-based feedforward neural network

Cited by 14 publications

References 21 publications

Reinforcement Q-Learning Control With Reward Shaping Function for Swing Phase Control in a Semi-active Prosthetic Knee

Reinforcement Q-Learning Control With Reward Shaping Function for Swing Phase Control in a Semi-active Prosthetic Knee

Optimal inverse magnetorheological damper modeling using shuffled frog-leaping algorithm–based adaptive neuro-fuzzy inference system approach

A hybrid model of magnetorheological dampers based on generalized hysteretic biviscous operators

Contact Info

Product

Resources

About