Reinforcement Learning-Based Backstepping Control for Container Cranes

Sun, Xiao; Xie, Zhihang

doi:10.1155/2020/2548319

Cited by 4 publications

(6 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To construct controller and analyse stability condition for TS systems (9), the LMIs approach is employed.…”

Section: Control Designmentioning

confidence: 99%

See 1 more Smart Citation

Takagi-Sugeno Fuzzy Approach with Compressed Representation for Overhead Crane System

2022

JST: SSAD

View full text Add to dashboard Cite

This paper proposes a Takagi-Sugeno (TS) fuzzy approach for 2-dimensional of freedom overhead crane. Since this system is underactuated with nonlinear mathematical dynamic model, the requirement of achieving accurate positioning while eliminating oscillation is a challenging issue. The nonlinear sector decomposition is taken into implementation which uses the nonlinear terms in the dynamic model as scheduling variables to transform the initial system to TS representation. However, the number of TS fuzzy rules is exponential in the number of nonlinear elements in the system, leading to the increase of computational expense. Therefore, the reduced complexity method is introduced to minimize the scheduling variables in TS system. In addition, the uncertain system’s components are also taken into consideration to enhance the robust property of the system when working in practical environment. The controller is constructed based on parallel distributed compensation (PDC) approach while the linear matrix inequalities (LMIs) technique is employed to analyse the system’s stability. The effectiveness of the proposed method is demonstrated through numeral simulations.

show abstract

“…To construct controller and analyse stability condition for TS systems (9), the LMIs approach is employed.…”

Section: Control Designmentioning

confidence: 99%

“…Neural network-based approaches [7] are proposed to approximate system parameters and external disturbances. Data-driven based control method [8], reinforcement learning [9] are used data from closed-loop experiments and optimal methods to construct dynamics model and adaptive parameters.…”

Section: Introductionmentioning

confidence: 99%

Takagi-Sugeno Fuzzy Approach with Compressed Representation for Overhead Crane System

2022

JST: SSAD

View full text Add to dashboard Cite

show abstract

“…The advantages of RL and fuzzy logic has led to design proposed model for monitoring the Irrigation tanks (Tank bund), bridges and barrages continuously to avoid disasters. RL having capability of interacting environment applies policy for gaining reward in accordance with the decision obtained from Fuzzy logic, as fuzzy logic helps in minimizing the states [15]. The decision for scheduling the requests by adapting the network resource to the changing demand of tank bund and user in view of ensuring better QoS is carried out.…”

Section: Related Workmentioning

confidence: 99%

“…URMA updates the availability of user resources. 15. If adequate resources are not available at user, then processing is forwarded to nearest fog device.…”

Section: A Device Agencymentioning

confidence: 99%

Multi-Agent System for Resource Adaptation in IoT to Monitor Leakage in Irrigation Tanks

Hatti

Sutagundar²

2021

Preprint

View full text Add to dashboard Cite

The sporadic changes in the requirements by the end users has led to the problems in managing the resources of IoT devices. The problem of managing the heterogeneous requests with the available resources in view of ensuring Quality of service (QoS) to end users is challenging issue. The proposed model inculcates the adaptation policy for adapting the resources to fulfill the requirements of the user. The dynamic changes in the environment are handled by Reinforcement learning model with Fuzzy Interference system to apply the policy. Periodically monitoring of irrigation tank to alert the flow of water above Full Tank Level (FTL) by Reinforcement learning agent, prioritization of tasks (requests) by Fuzzy logic is performed. It is done by interacting through agency, providing video conferencing or video calling facility to the user based on availability of the user resources. It also adapts according to communication and computational resources. The proposed model is simulated to monitor and control the leakage in tank. It helps to remotely control the leakages in irrigation tanks/ bridges through Multi-Agent Fuzzy Q learning model. It focuses on adapting the resources of the devices and the action considering the user resources. The algorithm is simulated in Ifogsim and python and performance are evaluated in terms of resource cost, latency, execution time, energy consumption and network usage.

show abstract

“…In recent years, the backstepping method has been gradually applied to the antisway control of cranes. In [26], a control method is proposed that combines fuzzy reinforcement learning with backstepping. The suppression of the payload swing is achieved by determining the parameters of the backstepping controller through reinforcement learning.…”

Section: Introductionmentioning

confidence: 99%

Flatness-Based Backstepping Antisway Control of Underactuated Crane Systems under Wind Disturbance

Niu

2023

Electronics

View full text Add to dashboard Cite

A control method that combines trajectory planning and backstepping is proposed for the antisway problem of underactuated overhead cranes under wind disturbance. First, a set of flat outputs is proposed so that the crane system dynamics can be represented by each order of flat outputs. Sufficient relevant constraints are given to ensure that the trolley can arrive at the desired position in a limited time under variable rope lengths, and that the swing angle can be suppressed when the payload is lifted or lowered during operation. The planned trajectory is obtained by solving for the optimal parameters of the flat output. Next, to reduce the deviation caused by wind disturbance on the actual control of the trajectory, a tracking controller is designed. Because the system output space and flat output space are differentiable homeomorphisms, the backstepping controller constructed based on the flat output can indirectly control the system output, which makes the backstepping method applicable to underactuated cranes. The simulation results show that the proposed method is effective and has strong robustness.

show abstract

Reinforcement Learning-Based Backstepping Control for Container Cranes

Cited by 4 publications

References 17 publications

Takagi-Sugeno Fuzzy Approach with Compressed Representation for Overhead Crane System

Takagi-Sugeno Fuzzy Approach with Compressed Representation for Overhead Crane System

Multi-Agent System for Resource Adaptation in IoT to Monitor Leakage in Irrigation Tanks

Flatness-Based Backstepping Antisway Control of Underactuated Crane Systems under Wind Disturbance

Contact Info

Product

Resources

About