Intelligent engine control using an adaptive critic

Kulkarni, Nilesh; Krishnakumar, Kalmanje

doi:10.1109/tcst.2003.809254

Cited by 29 publications

(9 citation statements)

References 16 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Step 5: All agents participating in the task compare DP and P. 5 E l e c t i n g a n e w A d d i n g t h e ( n + 1 ) t h a g e n t t o r u n Figure 7. Basic idea of the TWTC algorithm.…”

Section: Collaborative Strategy-twtc Algorithmmentioning

confidence: 99%

An agent-based framework for collaboration among critical mechanical devices working in a cluster

Yang

Xing

Wang

2012

Proceedings of the Institution of Mechanical Engineers, Part I:

View full text Add to dashboard Cite

Critical mechanical devices, such as high-pressure air compressors, are crucial for industrial production. Critical mechanical devices are independently distributed devices that work in a cluster, but have high power requirements and need large amounts of space. Thus, the efficient control and coordination of a cluster of critical mechanical devices to achieve task goals is beset with difficulties. Motivated by an example of high-pressure air compressors in a cluster, this paper proposes an agent-based framework for realizing collaboration among critical mechanical devices that work in a cluster. The proposed agent-based framework endows critical mechanical devices with abilities such as autonomy, interaction and collaboration. The framework comprises three core components: (i) an agent-based control approach to the design of critical mechanical device agents to create autonomy; (ii) a type of critical mechanical device agent communication language to create interaction among critical mechanical devices; and (iii) a total working time-based collaboration algorithm that fully considers the physical characteristics of critical mechanical devices to create collaboration among critical mechanical device clusters. A type of critical mechanical device agent controller is implemented using the proposed approach. The agent-based framework has been successfully applied and evaluated in a real-world project. The results obtained in that case study show that the proposed framework can achieve collaboration among critical mechanical device clusters allowing them to realize their predefined goals efficiently and safely.

show abstract

“…Step 5: All agents participating in the task compare DP and P. 5 E l e c t i n g a n e w A d d i n g t h e ( n + 1 ) t h a g e n t t o r u n Figure 7. Basic idea of the TWTC algorithm.…”

Section: Collaborative Strategy-twtc Algorithmmentioning

confidence: 99%

An agent-based framework for collaboration among critical mechanical devices working in a cluster

Yang

Xing

Wang

2012

Proceedings of the Institution of Mechanical Engineers, Part I:

View full text Add to dashboard Cite

show abstract

“…Most subsystems associated with UAV, as well as the turbojet engine, can be viewed as nonlinear systems with strong variability [22,23]. The neural network (NN) has been recognized as an efficient means for nonlinear time-varying controller design because of its learning ability.…”

Section: Introductionmentioning

confidence: 99%

Single Neural Adaptive PID Control for Small UAV Micro-Turbojet Engine

Wei

Wang

et al. 2020

Sensors

View full text Add to dashboard Cite

The micro-turbojet engine (MTE) is especially suitable for unmanned aerial vehicles (UAVs). Because the rotor speed is proportional to the thrust force, the accurate speed tracking control is indispensable for MTE. Thanks to its simplicity, the proportional–integral–derivative (PID) controller is commonly used for rotor speed regulation. However, the PID controller cannot guarantee superior performance over the entire operation range due to the time-variance and strong nonlinearity of MTE. The gain scheduling approach using a family of linear controllers is recognized as an efficient alternative, but such a solution heavily relies on the model sets and pre-knowledge. To tackle such challenges, a single neural adaptive PID (SNA-PID) controller is proposed herein for rotor speed control. The new controller featuring with a single-neuron network is able to adaptively tune the gains (weights) online. The simple structure of the controller reduces the computational load and facilitates the algorithm implementation on low-cost hardware. Finally, the proposed controller is validated by numerical simulations and experiments on the MTE in laboratory conditions, and the results show that the proposed controller achieves remarkable effectiveness for speed tracking control. In comparison with the PID controller, the proposed controller yields 54% and 66% reductions on static tracking error under two typical cases.

show abstract

“…Ideally, all AC designs use a reinforcement learning technique, where the controller optimizes certain attributes of the system based on the response of the system to a control action and the action generated by the controller itself [6]. There are other control structures that use critic-actor schemes can be found in [7]- [9].…”

Section: Introductionmentioning

confidence: 99%

Near-Optimal Controller for Nonlinear Continuous-Time Systems With Unknown Dynamics Using Policy Iteration

Dutta

Patchaikani

Behera

2016

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

This paper presents a single-network adaptive critic-based controller for continuous-time systems with unknown dynamics in a policy iteration (PI) framework. It is assumed that the unknown dynamics can be estimated using the Takagi-Sugeno-Kang fuzzy model with arbitrary precision. The successful implementation of a PI scheme depends on the effective learning of critic network parameters. Network parameters must stabilize the system in each iteration in addition to approximating the critic and the cost. It is found that the critic updates according to the Hamilton-Jacobi-Bellman formulation sometimes lead to the instability of the closed-loop systems. In the proposed work, a novel critic network parameter update scheme is adopted, which not only approximates the critic at current iteration but also provides feasible solutions that keep the policy stable in the next step of training by combining a Lyapunov-based linear matrix inequalities approach with PI. The critic modeling technique presented here is the first of its kind to address this issue. Though multiple literature exists discussing the convergence of PI, however, to the best of our knowledge, there exists no literature, which focuses on the effect of critic network parameters on the convergence. Computational complexity in the proposed algorithm is reduced to the order of (Fz)(n-1) , where n is the fuzzy state dimensionality and Fz is the number of fuzzy zones in the states space. A genetic algorithm toolbox of MATLAB is used for searching stable parameters while minimizing the training error. The proposed algorithm also provides a way to solve for the initial stable control policy in the PI scheme. The algorithm is validated through real-time experiment on a commercial robotic manipulator. Results show that the algorithm successfully finds stable critic network parameters in real time for a highly nonlinear system.

show abstract

Intelligent engine control using an adaptive critic

Cited by 29 publications

References 16 publications

An agent-based framework for collaboration among critical mechanical devices working in a cluster

An agent-based framework for collaboration among critical mechanical devices working in a cluster

Single Neural Adaptive PID Control for Small UAV Micro-Turbojet Engine

Near-Optimal Controller for Nonlinear Continuous-Time Systems With Unknown Dynamics Using Policy Iteration

Contact Info

Product

Resources

About