Reinforcement Learning for Controlling a Coupled Tank System Based on the Scheduling of Different Controllers

Diniz, A. A. R.; Pires, Paulo F.; Melo, Jorge Dantas de; Neto, Adrião Duarte Dória; Filho, Armando J. J. L.; Kanazava, Sergio M.

doi:10.1109/sbrn.2010.44

Cited by 7 publications

(5 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the RL domain, we will only provide our algorithms with a (reward function), which refers to a learning factor when it is performing well, and when it is performing poorly, the learning algorithm's task is to know how to choose actions over time to obtain big rewards. The goal of "RL" is to guide the agent to determine what action to take that maximizes (or minimizes) the sum of all RL signals (the numerical reward) or punishment, it receives over time, called the total expected reward [29]. According to Fig.…”

Section: Reinforcement Learning (Rl)mentioning

confidence: 99%

Energy Saving by Reinforcement Learning for Multi-Chillers of HVAC Systems

Hussein¹,

Ateeq²,

Homod³

2022

Proceedings of 2nd International Multi-Disciplinary Conference Theme: Integrated Sciences and Technologies, IMDC-IST 2021, 7-9

View full text Add to dashboard Cite

This paper presents a method for controlling and operating a multi-chillers system: (1) Model-based control approach was used by MATLAB/SIMULINK to model a building containing two non-identical chillers depending on thermal loads. (2) ON/OFF all chillers alternately using the model reinforcement learning controller (RL-control) to select the appropriate chiller for the building conditioning process. The results were in terms of energy efficiency and performance of the enhanced learning control for the chiller, and a control unit signal (PID) was applied to make a comparison with the signals of energy, power, and temperatures. After comparison, it was found that the energy saving through the proposed controller is 45% of the traditional (PID) strategy, where can the proposed strategy control for the chiller appropriate for the building's conditioning process.

show abstract

Section: Reinforcement Learning (Rl)mentioning

confidence: 99%

Energy Saving by Reinforcement Learning for Multi-Chillers of HVAC Systems

Hussein¹,

Ateeq²,

Homod³

2022

Proceedings of 2nd International Multi-Disciplinary Conference Theme: Integrated Sciences and Technologies, IMDC-IST 2021, 7-9

View full text Add to dashboard Cite

show abstract

“…Following this idea, [8] define reinforcement learning (RL) as learning what to do -as mapping situations to actions -in a way that maximizes a numerical reward.…”

Section: Reinforcement Learningmentioning

confidence: 99%

“…Hence, a policy , is the mapping from states to actions , taken from that state, and represents the probability of selecting each possible action, in such a way that the best actions correspond to the highest probability of choice [11]. [8] explains that to evaluate the quality of the actions taken by the agent can be applied the concept of the "actionvalue function for policy ", that represents an estimation of the total return expected, i. e., the quality of the action taken by the agent when it is following some policy . This function represents the value of the expected total return to the state (current state) when the action is chosen and it follows, from that state, the policy , as shown in (7).…”

Section: Reinforcement Learningmentioning

confidence: 99%

See 1 more Smart Citation

Modeling a system for monitoring an object using artificial neural networks and reinforcement learning

Peixoto

Diniz

Almeida

et al. 2011

The 2011 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

This paper presents a modeling of a system designed to monitor a moving object from images captured by a camera. The research was focused on defining the steps necessary to the functioning of systems, they are: capture and image processing, pattern recognition with artificial neural networks and seek the best path for moving the camera, using reinforcement learning. The results show the viability of the proposed system, being a relevant alternative to monitoring and security environments.

show abstract

“…Non-linear model gives a more accurate prediction for a wider operating range of control [4] . The couple tank system considered in this study is a typical example of the plant with a high degree of non-linearity [26,27] . The non-linearity in the CTS is mainly due to the basic dynamic equations of the CTS, the characteristics of the valves and as a result of the nonlinear flow characteristics in the tank system [4] .…”

Section: Introductionmentioning

confidence: 99%

A wavelet neural network based non-linear model predictive controller for a multi-variable coupled tank system

Owa

Sharma

Sutton

2014

Int. J. Autom. Comput.

View full text Add to dashboard Cite

Abstract:In this paper, a novel real time non-linear model predictive controller (NMPC) for a multi-variable coupled tank system (CTS) is designed. CTSs are highly non-linear and can be found in many industrial process applications. The involvement of multi-input multi-output (MIMO) system makes the design of an effective controller a challenging task. MIMO systems have inherent couplings, interactions in-between the process input-output variables and generally have an complex internal structure. The aim of this paper is to design, simulate, and implement a novel real time constrained NMPC for a multi-variable CTS with the aid of intelligent system techniques. There are two major formidable challenges hindering the success of the implementation of a NMPC strategy in the MIMO case. The first is the difficulty of obtaining a good non-linear model by training a non-convex complex network to avoid being trapped in a local minimum solution. The second is the online real time optimisation (RTO) of the manipulated variable at every sampling time. A novel wavelet neural network (WNN) with high predicting precision and time-frequency localisation characteristic was selected for an MIMO model and a fast stochastic wavelet gradient algorithm was used for initial training of the network. Furthermore, a genetic algorithm was used to obtain the optimised parameters of the WNN as well as the RTO during the NMPC strategy. The proposed strategy performed well in both simulation and real time on an MIMO CTS. The results indicated that WNN provided better trajectory regulation with less mean-squared-error and average control energy compared to an artificial neural network. It is also shown that the WNN is more robust during abnormal operating conditions.

show abstract

Reinforcement Learning for Controlling a Coupled Tank System Based on the Scheduling of Different Controllers

Cited by 7 publications

References 3 publications

Energy Saving by Reinforcement Learning for Multi-Chillers of HVAC Systems

Energy Saving by Reinforcement Learning for Multi-Chillers of HVAC Systems

Modeling a system for monitoring an object using artificial neural networks and reinforcement learning

A wavelet neural network based non-linear model predictive controller for a multi-variable coupled tank system

Contact Info

Product

Resources

About