Cargando…
Frontiers of intelligent control and information processing /
The current research and development in intelligent control and information processing have been driven increasingly by advancements made from fields outside the traditional control areas, into new frontiers of intelligent control and information processing so as to deal with ever more complex syste...
| Clasificación: | Libro Electrónico |
|---|---|
| Otros Autores: | |
| Formato: | Electrónico eBook |
| Idioma: | Inglés |
| Publicado: |
[Hackensack?] New Jersey :
World Scientific,
[2014]
|
| Temas: | |
| Acceso en línea: | Texto completo |
Tabla de Contenidos:
- Preface; Contents; 1. Dynamic Graphical Games: Online Adaptive Learning Solutions Using Approximate Dynamic Programming; 1.1 Introduction; 1.2 Graphs and Synchronization of Multi-Agent Dynamical Systems; 1.2.1 Graphs; 1.2.2 Synchronization and tracking error dynamics; 1.3 Multiple Player CooperativeGames on Graphs; 1.3.1 Graphical games; 1.3.2 Comparison of graphical games with standard dynamic games; 1.3.3 Nash equilibrium for graphical games; 1.3.4 Hamiltonian equation for dynamic graphical games; 1.3.5 Bellman equation for dynamic graphical games.
- 1.3.6 Discrete Hamilton-Jacobi theory: Equivalence of Bellman and discrete-time Hamilton Jacobi equations1.3.7 Stability and Nash solution of the graphical games; 1.4 Approximate Dynamic Programming for Graphical Games; 1.4.1 Heuristic dynamic programming for graphical games; 1.4.2 Dual heuristic programming for graphical games; 1.5 Coupled Riccati Recursions; 1.6 Graphical Game Solutions by Actor-Critic Learning; 1.6.1 Actor-critic networks and tuning; 1.6.2 Actor-critic offline tuning with exploration; 1.6.3 Actor-critic online tuning in real-time.
- 1.7 Graphical Game Example and Simulation Results1.7.1 Riccati recursion offline solution; 1.7.2 Simulation results using offline actor-critic tuning; 1.7.3 Simulation results using online actor-critic tuning; 1.8 Conclusions; Acknowledgement; References; 2. Reinforcement-Learning-Based Online Learning Control for Discrete-Time Unknown Nonaffine Nonlinear Systems; 2.1 Introduction; 2.2 Problem Statement and Preliminaries; 2.2.1 Dynamics of nonaffine nonlinear discrete-time systems; 2.2.2 A single-hidden layer neural network; 2.3 Controller Design via Reinforcement Learning.
- 2.3.1 A basic controller design approach2.3.2 Critic neural network and weight update law; 2.3.3 Action neural network and weight update law; 2.4 Stability Analysis and Performance of the Closed-Loop System; 2.5 Numerical Examples; 2.5.1 Example 1; 2.5.2 Example 2; 2.6 Conclusions; Acknowledgement; References; 3. Experimental Studies on Data-Driven Heuristic Dynamic Programming for POMDP; 3.1 Introduction; 3.2 Markov Decision Process and Partially Observable Markov Decision Process; 3.2.1 Markov decision process; 3.2.2 Partially observable Markov decision process.
- 3.3 Problem Formulation with the State Estimator3.4 Data-Driven HDP Algorithm for POMDP; 3.4.1 Learning in the state estimator network; 3.4.2 Learning in the critic and the action network; 3.5 Simulation Study; 3.5.1 Case study one; 3.5.2 Case study two; 3.5.3 Case study three; 3.6 Conclusions and Discussion; Acknowledgement; References; 4. Online Reinforcement Learning for Continuous-State Systems; 4.1 Introduction; 4.2 Background of Reinforcement Learning; 4.3 RLSPI Algorithm; 4.3.1 Policy iteration; 4.3.2 RLSPI; 4.4 Examples of RLSPI; 4.4.1 Linear discrete-time system.