检索结果-内蒙古大学图书馆

An actor-critic algorithm with policy gradients to solve the job shop scheduling problem using deep double recurrent agents

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Monaci, Marta Agasucci, Valerio Grani, Giorgio Sapienza University of Rome Dep. of Computer Science Control and Management Engineering Rome Italy Sapienza University of Rome Dep. of Statistics Rome Italy

In this work, we present a method that applies Deep Reinforcement Learning, an approximate dynamic programming procedure using deep neural networks, to the job shop scheduling problem (JSSP). The aim is to show that a greedy-like heuristic trained on a subset of problems, can effectively generalize to some extent to unseen instances, and be competitive compared to other methods. We model the JSSP as a Markov Decision Process and we exploit the efficacy of reinforcement learning to solve the problem. We adopt an actor-critic scheme based on policy gradients, specifically the Proximal Policy Gradient method, where the action taken by the agent is influenced by policy considerations on the state-value function. The procedures take into account the challenging nature of JSSP, where the state and the action space change for every instance and after each decision. To tackle this variability, we introduced a novel model based on two incident Long-Short Term Memory networks, followed by an encoding model, different in structure for both the actor and the critic. Experiments show the algorithm reaches good solutions in a short time, proving that is possible to generate new greedy heuristics just from learning-based methodologies. We compared our algorithms against several established heuristics, an adaptive method, a commercial solver based on branch and cut, and another approach based on Deep Reinforcement Learning, proving the validity of the proposed method in terms of time and makespan. The model can generalize, to some extent, to larger problems originating from a different *** Codes 90-08 © 2021, CC BY.

关键词： Reinforcement learning

Assessment and recognition in technical massive open on-line courses with and without on-line laboratories

学校读者我要写书评

暂无评论

Assessment and recognition in technical massive open on-line...

IEEE Education engineering (EDUCON)

作者： Sergio Martin Manuel Castro Elio Sancristobal Gabriel Diaz Felix Garcia-Loro Clara Perez Blanca Quintana Rosario Gil Pedro Plaza Africa Lopez German Carro Antonio Robles Llanos Tobarra Miguel Rodriguez-Artacho Jose A. Ruiperez-Valiente Electrical and Computer Engineering Department Universidad Nacional de Educación a Distancia Madrid Spain Dep. Sistemas de Comunicación y Control Universidad Nacional de Educación a Distancia Madrid Spain Dep. Lenguajes y Sistemas Informáticos Universidad Nacional de Educación a Distancia Madrid Spain Teaching Systems Lab Masschusetts Institute of Technology USA

This paper describes the organization and results of several MOOCs delivered about technical topics (i.e., digital electronics, VHDL design on FPGAs, open education and OERs repositories and the use of STEAM technologies to encourage diversity and inclusion), where online laboratories have been used in some of them. An analysis of the enrollment, students participating on tasks or quizzes, drop-out rate and certifications requests have been done.

关键词： VHDL Laboratories Organizations Task analysis engineering education Field programmable gate arrays Certification

Computational modelling of TRIANGULAR sub-boundary-layer vortex generators

学校读者我要写书评

暂无评论

Computational modelling of TRIANGULAR sub-boundary-layer vor...

2018 International Conference on Materials and Energy, ICOME 2018

作者： Fernandez-Gamiz, Unai Ruiz De Loizaga, Jon Errasti, Iñigo Boyano, Ana Zulueta, Ekaitz Lopez-Guede, Jose Manuel Nuclear Engineering and Fluid Mechanics Dep. University of the Basque Country Nieves Cano 12 Vitoria-Gasteiz Araba01006 Spain Nieves Cano 12 Vitoria-Gasteiz01006 Spain Automatic control and System Engineering Dep. University of the Basque Country Nieves Cano 12 Vitoria-Gasteiz Araba01006 Spain

Vortex generators (VGs) are used increasingly more by the wind turbine manufacture industry as flow control devices to improve rotor blade aerodynamic performance. The VG height is usually designed with equal thickness of the local boundary layer at the VG position. Nevertheless, these conventional VGs may produce excess residual drag in some applications. The so-called sub boundary layer VGs can provide enough momentum transfer over a region several times their own height for effective flow-separation control with much lower drag. The main objective is to investigate how well the simulations can reproduce the physics of the flow of the primary vortex generated by a triangular VG mounted on a flat plate with negligible pressure gradient with the angle of attack of the vane to the oncoming flow β= 18. Three different device heights H= 5mm, H1= 6,25mm, H2= 4,16mm have been studied and compared both qualitatively and quantitatively. To that end, computational simulations have been carried out using RANS method and at Reynolds number Re = 2600 based on the boundary layer momentum thickness θ= 2.4 mm at the VG position. The computational results show good agreement with the experimental data available in AVATAR project. © The Authors, published by EDP Sciences, 2020.

关键词： Boundary layers

Learning Based Model Predictive control for Quadcopters with Dual Gaussian Process

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Liu, Yuhan Tóth, Roland The Harbin Institute of Technology Department of Control Science and Engineering Harbin China The Eindhoven University of Tech. Dep. of Electrical Eng. Control Systems Group Eindhoven Netherlands The Institute for Computer Science and Control Budapest Hungary

An important issue in quadcopter control is that an accurate dynamic model of the system is nonlinear, complex, and costly to obtain. This limits achievable control performance in practice. Gaussian process (GP) based estimation is an effective tool to learn unknown dynamics from input/output data. However, conventional GP-based control methods often ignore the computational cost associated with accumulating data during the operation of the system and how to handle forgetting in continuous adaption. In this paper, we present a novel Dual Gaussian Process (DGP) based model predictive control strategy that improves the performance of a quadcopter during trajectory tracking. The bio-inspired DGP structure is a combination of a long-term GP and a short-term GP, where the long-term GP is used to keep the learned knowledge in memory and the short-term GP is employed to rapidly compensate unknown dynamics during online operation. Furthermore, a novel recursive online update strategy for the short-term GP is proposed to successively improve the learnt model during online operation in an efficient way. Numerical simulations are used to demonstrate the effectiveness of the proposed strategy. © 2021, CC BY-NC-SA.

关键词： Model predictive control

Accelerated nonlinear model predictive control by exploiting saturation

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Dyrska, Raphael Mitze, Ruth Mönnigmann, Martin Automatic Control and Systems Theory Dep. of Mechanical Engineering Ruhr-Universität Bochum Bochum44801 Germany

We present an approach for accelerating nonlinear model predictive control. If the current optimal input signal is saturated, also the optimal signals in subsequent time steps often are. We propose to use the open-loop optimal input signals whenever the first and some subsequent input signals are saturated. We only solve the next optimal control problem, when a non-saturated signal is encountered, or the end of the horizon is reached. In this way, we can save a significant number of NLPs to be solved while on the other hand keep the performance loss small. Furthermore, the NMPC is reactivated in time when it comes to controlling the system safely to its reference. Copyright © 2020, The Authors. All rights reserved.

关键词： Optimal control systems

State space sets with common optimal feedback laws for nonlinear MPC

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Mitze, Ruth Dyrska, Raphael König, Kai Mönnigmann, Martin Automatic Control and Systems Theory Dep. of Mechanical Engineering Ruhr-Universität Bochum Bochum44801 Germany

In model predictive control (MPC), an optimal control problem (OCP) is solved for the current state and the first input of the solution, the optimal feedback law, is applied to the system. This procedure requires to solve the OCP in every time step. Recently, a new approach was suggested for linear MPC. The parametric solution of a linear quadratic OCP is a piecewise-affine feedback law. The solution at a point in state space provides an optimal feedback law and a domain on which this law is the optimal solution. As long as the system remains in the domain, the law can be reused and the calculation of an OCP is avoided. In some domains the optimal feedback laws are identical. By uniting the corresponding domains, bigger domains are achieved and the optimal feedback law can be reused more often. In the present paper, we investigate in how far this approach can be extended from linear to nonlinear MPC, we propose an algorithm and we illustrate the achieved savings with an example. Copyright © 2020, The Authors. All rights reserved.

关键词： Model predictive control

Toward HAZOP 4.0 Approach for Managing the Complexities of the Hazard and Operability of an Industrial Polymerization Reactor

学校读者我要写书评

暂无评论

IFAC-PapersOnLine 2020年第2期53卷 13593-13600页

作者： R. Mokhtarname A.A. Safavi L. Urbas F. Salimi M.M. Zerafat N. Harasi Advanced Control Laboratory School of Electrical and Computer Engineering Shiraz University Shiraz Iran Mechanical Science and Eng. Dep. Technische Universität Dresden Dresden Germany ADEPP Academy France Nano-chemical Eng. Dep. Shiraz University Shiraz Iran

Conventional HAZOP is a collaborative and multidisciplinary activity to identify the hazards associated to operability of the chemical processes. Then the consequences and required safeguards of each potential deviation or failure are assessed qualitatively one by one. It is performed with the assumption of "one failure at the time" for the process parameters. Furthermore, process complexities such as multiple failures, domino effects, time and amplitude dep.ndent deviations etc. are avoided to keep the HAZOP brainstorming sessions systematic and effective. Therefore, the quality and outcomes of HAZOP study are relatively subjective and dep.nd on the experience and competency of the HAZOP team. This simplified approach cannot fully cover the hazard identification and risk assessment of the complex processes such as polymerization. Furthermore, the incident investigations show that almost all major accidents have occurred due to multiple failures or domino effects. This paper aims at developing a practical methodology in the context of "Industry 4.0" and particularly illustrate how dynamic simulation liberates the HAZOP team from the simplification assumptions such as one failure at the time or neglecting the domino effects during the lifecycle of the complex processes. An industrial styrene bulk free radical polymerization process has been chosen as the case study to dep.ct the applicability of the proposed method. In continuation of this research, the dynamic simulation integrated with Artificial Intelligence (AI) algorithms and Multivariable Process Monitoring (MPM) together with virtual collaboration tools will be invoked towards a more practical and effective HAZOP 4.0 platform. Such a platform can be the foundation of the further Process Safety Management (PSM) elements such as "Operating manuals", "Training and Competency management", "Condition monitoring and predictive maintenance", "Management of Change", "Pre-Start-up Safety Review", etc.

关键词： Dynamic HAZOP Dynamic Simulation Polymerization Plant Process Complexity HAZOP 4.0

An analysis of lockergoga ransomware

学校读者我要写书评

暂无评论

An analysis of lockergoga ransomware

2019 IEEE East-West Design and Test Symposium, EWDTS 2019

作者： Adamov, Alexander Carlsson, Anders Surmacz, Tomasz NioGuard Security Lab Kharkiv National University of Radio Electronics Kharkiv Ukraine Dep. of Computer Science Blekinge Institute of Technology Karlskrona Sweden Institute of Computer Engineering Control and Robotics Wroclaw University of Science and Technology Wroclaw Poland

ISBN: (纸本)9781728110035

This paper contains an analysis of the LockerGoga ransomware that was used in the range of targeted cyberattacks in the first half of 2019 against Norsk Hydra-A world top 5 aluminum manufacturer, as well as the US chemical enterprises Hexion, and Momentive-Those companies are only the tip of the iceberg that reported the attack to the public. The ransomware was executed by attackers from inside a corporate network to encrypt the data on enterprise servers and, thus, taking down the information control systems. The intruders asked for a ransom to release a master key and decryption tool that can be used to decrypt the affected files. The purpose of the analysis is to find out tactics and techniques used by the LockerGoga ransomware during the cryptolocker attack as well as an encryption model to answer the question if the encrypted files can be decrypted with or without paying a ransom. The scientific novelty of the paper lies in an analysis methodology that is based on various reverse engineering techniques such as multi-process debugging and using open source code of a cryptographic library to find out a ransomware encryption model. © 2019 IEEE.

关键词： Reverse engineering

A dynamic programming approach to solving constrained linear-quadratic optimal control problems

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Mitze, Ruth Mönnigmann, Martin Automatic Control and Systems Theory Dep. of Mechanical Engineering Ruhr-Universität Bochum Bochum44801 Germany

The solution of a constrained linear-quadratic regulator problem is determined by the set of its optimal active sets. We propose an algorithm that constructs this set of active sets for a desired horizon N from that for horizon N − 1. While it is not obvious how to extend the optimal feedback law itself for horizon N − 1 to horizon N, a simple relation between the optimal active sets for two successive horizon lengths exists. Specifically, every optimal active set for horizon N is a superset of an optimal active set for horizon N − 1 if the constraints are ordered stage by stage. The stagewise treatment results in a favorable computational effort. In addition, it is easy to detect the solution of the current horizon is equal to the infinite-horizon solution, if such a finite horizon exists, with the proposed algorithm. Copyright © 2019, The Authors. All rights reserved.

关键词： Quadratic programming