Managing the Balance Between Project Value and Net Present Value Using Reinforcement Learning DOI Creative Commons
Cláudio Szwarcfiter, Yale T. Herer

IEEE Access, Год журнала: 2023, Номер 12, С. 7500 - 7518

Опубликована: Дек. 25, 2023

Two important goals in project management are the maximization of net present value (NPV) and value, a more recent target. The former is well-documented objective scheduling, both evaluation tools used by decision makers. literature has focused on NPV problem as separate research tracks, but consideration tradeoff between offers makers thorough when weighing alternatives. This paper introduces novel formulation that includes robust develops algorithms to solve it, illustrates objectives. proposed mixed integer program (MIP) features multimode setting, where selection an activity mode will impact cost, duration, resource usage stochastic durations. To problem, this study innovative reinforcement learning (RL) based algorithm. solution can be plot efficient frontier value. Computational experiments revealed algorithm performs well compared tabu search MIP using commercial solver, RL actions leveraged for coping with positive negative cashflows. utility our work lies its ability respond makers' information needs, providing framework analysis select most adequate plan satisfies stakeholders' requirements.

Язык: Английский

Data-driven torque and pitch control of wind turbines via reinforcement learning DOI Creative Commons
Jingjie Xie, Hongyang Dong, Xiaowei Zhao

и другие.

Renewable Energy, Год журнала: 2023, Номер 215, С. 118893 - 118893

Опубликована: Июнь 8, 2023

This paper addresses the torque and pitch control problems of wind turbines. The main contribution this work is development an innovative reinforcement learning (RL)-based method targeting turbine applications. Our RL-based framework synergistically combines advantages deep neural networks (DNNs) model predictive (MPC) technologies. proposed strategy data-driven, adapting to real-time changes in system dynamics enhancing performance robustness. Additionally, incorporation MPC structure within our design improves efficiency reduces high computational complexity typically found RL algorithms. Specifically, a DNN designed approximate based on continuously updated dataset composed state action measurements taken at specified sampling intervals. policy generated by integrating online trained into architecture. iteratively updates optimize performance. As primary result work, demonstrates superior robustness compared commonly-employed other baseline controllers presence uncertainties unexpected actuator faults. effectiveness showcased through simulations with high-fidelity simulator.

Язык: Английский

Процитировано

35
Uncertainty analysis and parameter optimization of a water yield ecosystem service model: A case study of the Qilian Mountains, China DOI
Bei Wang, Gaofeng Zhu,

Juntao Zhong

и другие.

The Science of The Total Environment, Год журнала: 2025, Номер 966, С. 178772 - 178772

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

1
A quasi-oppositional learning of updating quantum state and Q-learning based on the dung beetle algorithm for global optimization DOI Creative Commons
Zhendong Wang, Lili Huang, Shuxin Yang

и другие.

Alexandria Engineering Journal, Год журнала: 2023, Номер 81, С. 469 - 488

Опубликована: Сен. 22, 2023

There are many tricky optimization problems in real life, and metaheuristic algorithms the most effective way to solve at a lower cost. The dung beetle algorithm (DBO) is more innovative proposed 2022, which affected by action of beetles such as ball rolling, foraging, reproduction. Therefore, A based on quasi-oppositional learning Q-learning (QOLDBO). First, quantum state update idea cleverly integrated into increase randomness generated population. And best behavior pattern selected adding rolling stage improve search effect. In addition, variable spiral local domain method make up for shortage developing only around neighborhood optimum. For optimal solution each iteration, dimensional adaptive Gaussian variation retained. Experimental performance tests show that QOLDBO performs well both benchmark test functions CEC 2017. Simultaneously, validity verified several classical practical application engineering problems.

Язык: Английский

Процитировано

21
Adaptive optimal secure wind power generation control for variable speed wind turbine systems via reinforcement learning DOI
Mahmood Mazare

Applied Energy, Год журнала: 2023, Номер 353, С. 122034 - 122034

Опубликована: Окт. 14, 2023

Язык: Английский

Процитировано

21
Deep clustering of Lagrangian trajectory for multi-task learning to energy saving in intelligent buildings using cooperative multi-agent DOI
Raad Z. Homod, Hayder I. Mohammed, Aissa Abderrahmane

и другие.

Applied Energy, Год журнала: 2023, Номер 351, С. 121843 - 121843

Опубликована: Сен. 2, 2023

Язык: Английский

Процитировано

18
Wind forecasting-based model predictive control of generator, pitch, and yaw for output stabilisation – A 15-megawatt offshore DOI Creative Commons
Tenghui Li, Jin Yang, Αναστασία Ιωάννου

и другие.

Energy Conversion and Management, Год журнала: 2024, Номер 302, С. 118155 - 118155

Опубликована: Фев. 1, 2024

As wind energy continuously expands its share in power generation, the grid has a higher requirement for stable production. This study aims forecasting-based turbine control to mitigate fluctuation caused by uncertainties. Firstly, compass-vector transformation supports model on direction forecasting besides velocity. Wind modelling adopts general network structure of learning-shaping learn transformed vector series. speed and averaged from prediction determine three-degree-of-freedom (3-DOF) reference as objective update system configuration. Subsequently, predictive (MPC) solves real-time regulation sparse quadratic programming (QP). Besides, loop integrates generator control, compensation, output buffer coordinate generator, pitch servo, yaw servo. According simulation, long short-term memory (LSTM) ensures mean accuracy over 0.997 30-s window. Its performance is more than dense (DNN), convolutional (CNN), CNN-LSTM. Compared baseline proposed MPC can reduce 7% oscillation 12% peak-to-peak. improves rotation stability 44% at high wind. The proven contribute better quality.

Язык: Английский

Процитировано

4
Data-driven modeling and MPPT control of offshore wind turbines based on machine learning approach DOI
Junrong Li, Guolian Hou

Ocean Engineering, Год журнала: 2025, Номер 329, С. 121121 - 121121

Опубликована: Апрель 6, 2025

Язык: Английский

Процитировано

0
Online assessment of frequency support capability of the DFIG-based wind farm using a knowledge and data-driven fusion Koopman method DOI

Yimin Ruan,

Wei Yao, Qihang Zong

и другие.

Applied Energy, Год журнала: 2024, Номер 377, С. 124518 - 124518

Опубликована: Сен. 23, 2024

Язык: Английский

Процитировано

1
Micro-Grid Wind Energy Conversion Systems: Conventional and Modern Embedded Technologies - A Review DOI Creative Commons

Mutiu K. Agboola,

Kabiru Alani Hassan,

Titus O. Ajewole

и другие.

Journal of Digital Food Energy & Water Systems, Год журнала: 2024, Номер 5(2)

Опубликована: Дек. 27, 2024

This study assesses the effectiveness of an electric microgrid wind energy conversion system using both traditional techniques and contemporary embedded systems, such as artificial neural network-based control mechanisms fuzzy logic control. The text compares lists advantages disadvantages various types turbines (WTs). Moreover, this falls into one two groups: conventional power or non-traditional On other side, describes methods manually controlling turbine rotor's rotation speed computational analysis. current work, in contrast, investigates evaluates used systems (WECS), including maximum point tracking, Artificially intelligent relation to mechanism, provide complete over pitch angle, coefficient, tip ratio for best possible extraction. makes a direct comparison possible. Nonetheless, there are few drawbacks difficulties with widely utilized quality extractions: artificially networks their systems. However, combining technology integrated intelligence controllers may be workable strategy lessen even eliminate these difficulties, well advantageous upcoming studies.

Язык: Английский

Процитировано

1
A Statistical Method to Evaluate the Impact of Electrical Consumption Uncertainty in a Renewable Energy Community DOI
Mattia Calabrese, Andrea Ademollo, Carlo Carcasci

и другие.

Опубликована: Янв. 1, 2024

Renewable Energy Communities (RECs) have been introduced in Italy following the European Directive RED II. As they spread, aim is to further enhance energy generated from renewable sources by promoting shared and establishing a new model of electricity grid management.This study explores integration within contemporary landscape, emphasizing pivotal importance understanding individual participants' consumption patterns, associated with significant uncertainty.A statistical method perform techno-economic analysis proposed. The addresses uncertainty households' while maintaining constant production photovoltaic (PV) field. generates distribution subsequential economic revenues REC's users. A sensitivity then performed varying both number consumers involved PV field's power.The carried out after fitting process simulated data using theoretical curves. In this way, database curves simulate domestic electrical created. Consequently, annual sampled Monte Carlo introduction behavior parameter.Results indicate that leads lower Collective Self Consumption (CSC) values compared deterministic analysis, considering uncertainties input parameters. Economic evaluations demonstrate increased Net Present Value (NPV) REC, influencing even optimal sizing system.These findings contribute valuable insights for designing economically viable sustainable Communities.

Язык: Английский

Процитировано

0