Dynamic Neural Networks for Model-Free Control and Identification
1Departamento de Control Automático, Centro de Investigación y de Estudios Avanzados, Mexico City, Mexico
2Bitechnology Department (UPIBI), National Politecnical Institute (IPN), Mexico City, Mexico
3Department of Electrical, Computer and Biomedical Engineering, University of Rhode Island, Kingston, RI 02881, USA
Dynamic Neural Networks for Model-Free Control and Identification
Description
Since their strong rebirth in the last decade, Artificial Neural Networks (ANNs) are playing an increasing role in the engineering theory and practice. ANNs have shown also good identification properties in presence of uncertainties or external disturbances. Recent developments in ANN applications permitted to consider them as providing considerable promise for application of new advanced theoretical results in nonlinear control. This promise is based on their theoretical capability to approximate well arbitrary continuous nonlinear mappings. By large, the application of neural networks to automatic control is usually for building a model of the plant, and then, based on this model, to design a control law. There are two known types of NNs: static one (SNN), using the so-called “back-propagation technique” and Dynamic Neural Networks (DNNs). The first one (SNN) deals with the class of global optimization problems trying to adjust the weights of such NNs to minimize an identification error. The second approach deals with DNNs and exploits the feedback properties of the applied DNN, which permits to avoid many problems related to global extremum search converting the learning (training) process to an adequate feedback design. If the mathematical model of a considered process is incomplete or partially known, the DNN approach provides an effective instrument to attack a wide spectrum of problems such as identification, state estimation, trajectories, and tracking. There are two known types of DNNs:
- DNN functioning in discrete time and referred to as Recurrent Neural Networks
- DNN functioning in continuous time and referred to as Differential Neural Networks
This special issue will focus on recent theoretical advances in both Recurrent and Differential Neural Networks and will incorporate also papers from broad range of disciplines related to the application of DNN in actual engineering practice. Potential topics include, but are not limited to:
- DNN identification based on input-output data
- State estimation (filtering) of model-free plants
- Neurocontrol of dynamic plants
- Dynamic games of model-free systems using DNN approach
- Application of DNN controllers and state estimators to different mechanical, biotechnological, petrol industries, financial, and other systems
Before submission authors should carefully read over the journal's Author Guidelines, which are located at http://www.hindawi.com/journals/jcse/guidelines/. Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Tracking System at http://mts.hindawi.com/ according to the following timetable: