Modeling and control of McKibben artificial muscle enhanced with echo state networks

Utilizing online learning based on echo-state networks for the control of a hydraulic excavator

“…Furthermore, the training process corresponds to solving a linear regression problem and thus, does not suffer from problems such as slow convergence or sub-optimality that are inherent in most gradient-based methods used for training of RNNs. Due to this advantage, ESNs have found various applications such as predicting chaotic and nonlinear systems [23][24][25], motor speed control [26], online classification of visual tasks [27], learning grammatical structures [28], automatic speech recognition [29], control of shape memory alloys [30], forecasting short-term electric load [31], nonlinear adaptive filtering of complex signals [32], and modeling and control of pneumatic artificial muscles [33].…”

An Integrated Intelligent Nonlinear Control Method for a Pneumatic Artificial Muscle

“…Remark 1: The spool dynamics of the studied valve are ignored. Based on previous studies [7], [8], [22], [27], [30], it can be seen that the system dynamics (Eqs. (1) - (2)) are not only nonlinear functions of the inputs ( , , )…”

“…By applying Newton's second law and previous results ( [7], [8], [18], [30]), the force dynamics of the system can be presented under a simple form as 2 2 ( , )…”

“…However, unmodeled and unknown terms were not still covered by the controllers. To solve these issues of the model-based approaches, intelligent methods have been developed, such as neural network-based control [28]- [30], fuzzy PD+I learning control [31], advanced intelligent nonlinear PID controllers [32]- [34], hybrid fuzzyneural control [35], tuning fuzzy controllers [36]- [38], and switching predictive approaches [39]. Here, the control objective was realized by utilizing neural networks or fuzzy schemes to represent the system behaviors as black-box models or to adjust the control parameters via various learning algorithms such as back propagation (BP), recursive least squares (RLS), or bumpless transition (BT) mechanisms.…”

Modelling of pneumatic muscle actuator and antagonistic joint using linearised parameters

“…Andrikopoulos et al (2012) attempted to develop a dynamic model based on a so called switch system approach and obtained good results. In order to develop a more precise model, a novel network training method was also applied, which was proved to be robust and have good performance (Xing et al, 2012). A model of PM expressed with an exponential function, which has six parameters, was proposed and studied (Szepe, 2011).…”