Using recurrent neuro-fuzzy techniques for the identification and simulation of dynamic systems

Abstract

The identi"cation and simulation of dynamic systems is still a challenging problem. In this article some basic aspects of neuro-fuzzy techniques for the identi"cation and simulation of time-dependent physical systems are presented. In particular, a neuro-fuzzy model that can be used for the identi"cation and the (real-time) simulation of viscoelastic models, is described. The presented model is motivated by a cooperative neuro-fuzzy approach based on a vectorized recurrent neural network architecture. The physical motivation of this model is illustrated and speci"c propagation procedures and a learning algorithm are presented. Moreover, the usability in practice is demonstrated by an application of the model in the area of surgical simulation.

A fuzzy-neural multi-model for nonlinear systems identification and control

“…The [4,7,10,22,23] proposed as a dynamic function in the consequent part of the T-S rules to use an RNN. Some extension of this T-S approach for identification of dynamic plants with distributed parameters is given in [26]. The difference between the approach used in [22,23] fuzzy neural model and the approach used in [7,10,4] is that the first one uses the FGS-RNN [11] model, which is sequential one, and the second one uses the RTNN model [4], which is completely parallel one.…”

Real-Time Nonlinear FEM with Neural Network for Simulating Soft Organ Model Deformation

“…As well as our method, the methods in [4,8,12] incorporate neural networks into their deformation simulators. In [4,8], the networks are used to determine the parameters of mass-spring models.…”

“…In [4,8], the networks are used to determine the parameters of mass-spring models. The soft organ deformation in [12] is regarded as the propagation process of a potential energy generated by an external force.…”

Cellular neural network based deformation simulation with haptic force feedback

“…There are several investigations that combine neural network with deformable modelling [16,17]. However, in these methods, neural networks are mainly used to determine the parameters of mass-spring models.…”