WO2019164125A1 - Système de détermination des paramètres de commande pour optimiser les performances d'un accélérateur à l'aide de techniques d'apprentissage par renforcement et d'apprentissage automatique - Google Patents

Système de détermination des paramètres de commande pour optimiser les performances d'un accélérateur à l'aide de techniques d'apprentissage par renforcement et d'apprentissage automatique Download PDF

Info

Publication number
WO2019164125A1
WO2019164125A1 PCT/KR2019/000514 KR2019000514W WO2019164125A1 WO 2019164125 A1 WO2019164125 A1 WO 2019164125A1 KR 2019000514 W KR2019000514 W KR 2019000514W WO 2019164125 A1 WO2019164125 A1 WO 2019164125A1
Authority
WO
WIPO (PCT)
Prior art keywords
accelerator
learning
control
control parameter
machine learning
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2019/000514
Other languages
English (en)
Korean (ko)
Inventor
이근호
김진원
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
MOBIIS CO Ltd
Original Assignee
MOBIIS CO Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by MOBIIS CO Ltd filed Critical MOBIIS CO Ltd
Publication of WO2019164125A1 publication Critical patent/WO2019164125A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G05B13/027—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0499—Feedforward networks
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/041—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a variable is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/09—Supervised learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/092—Reinforcement learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N99/00—Subject matter not provided for in other groups of this subclass

Definitions

  • the present invention relates to a parameter determination system for accelerator control devices, and more particularly, to a system for determining parameter values for each accelerator control device through artificial neural network based learning and simulation.
  • a particle accelerator is a device that accelerates an atomic nucleus or elementary particle, but ultimately a device that tries to observe and determine the microstructure of matter through particle collision.
  • Such particle accelerators include cation accelerators, anion accelerators, heavy ion accelerators, electron accelerators (radiation accelerators), and the like, and linear accelerators and circular accelerators, depending on the type of acceleration.
  • an electron gun that radiates a powerful ultraviolet laser to copper to eject electrons
  • a linear accelerator that greatly compresses the length of the electron beam by the electron gun
  • a compressed accelerated electron beam passes between the permanent magnets.
  • An undulator that produces X-ray radiation that is considerably brighter than light, and an X-ray experimental apparatus (beamline) that outputs X-ray radiation to reveal the structure and phenomena of matter to the molecular structure form one acceleration system.
  • various control devices related to electron gun control, particle acceleration control, emission control and the like are gathered to form one accelerator control system.
  • the physical model simulation using the current Matlab and the optimization by the equipment according to the actual operation experience can be partially optimized.However, the final output quality of the accelerator (for example, Q-BPM total) can be achieved according to the mutual influence of various control parameters. There is a limit to searching / determining the optimal control parameter that maximizes the value).
  • the present invention has been made to solve the above-mentioned conventional problems, and an object thereof is to provide a system for calculating / searching an optimum value for maximizing accelerator final output quality for various control parameters included in an accelerator control system. will be.
  • a parameter determination system for accelerator control apparatuses includes a plurality of apparatus simulators corresponding to each of a plurality of accelerator control apparatuses and performing learning and simulation based on an artificial neural network; Designate a control parameter collection set corresponding to the collection of at least one control parameter included in the plurality of device simulators, and change the value of the control parameters included in the control parameter collection set and thereby adjust the accelerator final output quality accordingly.
  • a machine learning processing unit for calculating the value of the optimal control parameters to the highest final output quality of the accelerator.
  • the machine learning processing unit may calculate optimal control parameter values one by one in the order included in the corresponding control parameter collection set among the control parameters included in the control parameter collection set.
  • each of the plurality of device simulators is matched to a specific arrangement order, and the control parameter set may include control parameters included in the plurality of device simulators in the same order as that of the corresponding device simulator. have.
  • the machine learning processor may calculate optimal control parameter values through a Monte Carlo Tree Search (MCTS) algorithm using a reinforcement learning model.
  • MCTS Monte Carlo Tree Search
  • the machine learning processor includes a plurality of learning data in which an output of an accelerator injector, a control parameter value of each accelerator control device, and an accelerator final output quality value are matched, and a predetermined number of the plurality of learning data is set. After learning as much as the learning data through supervised learning (SL), reinforcement learning (RL) may be performed.
  • SL supervised learning
  • RL reinforcement learning
  • FIG. 1 is a functional block diagram of a parameter determination system according to an embodiment of the present invention.
  • FIG. 2 is an example of a structure of an artificial neural network of the device simulator of FIG. 1,
  • FIG. 3 is an example of data used by the machine learning processor of FIG. 1 for supervised learning.
  • FIG. 5 illustrates a combination between a parameter determination system and a conventional EPIC system according to an embodiment of the present invention compared to FIG. 4.
  • FIG. 1 is a functional block diagram of a parameter determination system (hereinafter referred to as a 'parameter determination system') for accelerator control apparatuses according to an embodiment of the present invention.
  • a parameter determination system hereinafter referred to as a 'parameter determination system'
  • the parameter determination system 100 includes a plurality of device simulators 110 and a machine learning processor 120.
  • each device simulator 110 corresponds to each accelerator control device, and as mentioned in the background art, in order to operate one accelerator, dozens or hundreds of accelerator control devices are required, if necessary.
  • Each device simulator 110 of FIG. 1 implements such accelerator control devices, respectively.
  • the device simulator 110 may be configured to perform learning or simulation based on artificial neural networks.
  • the device simulator 110 may configure an artificial neural network as shown in FIG. 2, and may determine an optimal internal parameter value for displaying the same characteristics as the actual accelerator control device through machine learning.
  • FIG. 2 is only a form well known in the art of constructing an artificial neural network for machine learning, and thus, a description of well-known techniques for optimally setting a weight in each layer by machine learning will be omitted.
  • each device simulator 110 receives a result of the device simulator 110 located at the front end as a sensor parameter, and also receives a control parameter, and then selects an internal hidden layer (hidden layer).
  • hidden layer The shape, weight, etc. may vary depending on the function of each accelerator control device corresponding to the device simulator 110), and finally, the sensor parameter may be transmitted to the device simulator 110 located at the next stage in the output layer.
  • each device simulator 110 is configured to output a result according to the internal control parameter value after receiving the result value of the device simulator 110 located at the front end.
  • each device simulator 110 in the 'front stage' or 'next stage' means that the accelerator control devices for actual accelerator operation are arranged in such an order.
  • the plurality of device simulators 110 are each matched to a specific arrangement order, and the arrangement order matched to these device simulators 110 corresponds to the arrangement order of accelerator control devices for actual accelerator operation.
  • the internal variable weight value of each layer (artificial neural network layer) of the device simulator 110 is determined.
  • the devices, that is, the control parameters of the device simulators 110 are calculated or determined by the machine learning processor 120, which will be described in detail below.
  • the machine learning processor 120 designates a control parameter set set corresponding to a collection of at least one control parameter included in the plurality of device simulators 110, and assigns a value to the values of the parameters included in the control parameter collection set. After learning about the change and the final output quality of the accelerator through the artificial neural network-based learning process, and performs the function of calculating the value of the optimal control parameters to the highest accelerator final output quality.
  • the machine learning processor 120 may refer to them as a set of parameter sets ' ⁇ C0_1, C1_1, C2_1, C3_1'. ⁇ 'And then change the values of each of the parameters (ie, C0_1, C1_1, C2_1, C3_1) included in the set of parameters through machine learning to ensure that the final output quality is the highest. To calculate a value.
  • the machine learning processor 120 may calculate optimal parameter values one by one in the order included in the corresponding control parameter set among the parameters included in the control parameter set.
  • the machine learning processor 120 designates a value for the first parameter C0_1 and sets the value as a fixed value. In this state, machine learning is performed to calculate C1_1 satisfying the optimum final output quality.
  • the machine learning processing unit 120 calculates C2_1 satisfying the optimum final output quality through machine learning while setting the values of C0_1 and C1_1 previously specified or calculated as fixed values, and similarly, C0_1, C1_1, and C2_1 With a fixed value, machine learning can yield C3_1 that satisfies the optimal final output quality.
  • control parameter set may include control parameters included in the plurality of device simulators 110 in the same order as the arrangement order of the corresponding device simulator 110.
  • each of the plurality of device simulators 110 may have a specific arrangement order according to the corresponding accelerator control device, and the control parameter set may include each control according to the arrangement order of each device simulator 110.
  • the parameters are included, and the machine learning processor 120 determines / calculates values of each control parameter in the order included in the control parameter set.
  • the first accelerator control device, the second accelerator control device, the third accelerator control device, and the fourth accelerator control device should be arranged in that order, and the first accelerator control device may include the first control parameter C0_1.
  • the second control parameter C1_1 may be set in the second accelerator control apparatus
  • the third control parameter C2_1 may be set in the third accelerator control apparatus
  • the fourth control parameter C3_1 may be set in the fourth accelerator control apparatus.
  • the set of control parameter collections may be configured as ' ⁇ C0_1, C1_1, C2_1, C3_1 ⁇ '
  • the machine learning processor 120 performs a process of determining each parameter value in that order.
  • each accelerator control apparatus has one parameter, but each accelerator control apparatus may have a plurality of parameters, and of course, a priority among the plurality of parameters may exist.
  • the machine learning processor 120 may use a Monte Carlo Tree Search (MCTS) algorithm based on the reinforcement learning model. .
  • MCTS Monte Carlo Tree Search
  • MCTS algorithm is a term for an algorithm that randomly calculates the value of a function using random numbers. If the value to be calculated is not represented as a closed value or is complicated, it is used to approximate it.
  • Monte-Carlo Tree Search is a method of finding the best decision. It is a heuristic search algorithm for decision making and is often used when it is not easy to find a solution. It is used as a way to find the number of.
  • each number in the game is a node, and the whole process of the game is represented by a tree of each number.
  • the odds are recorded in each node, and the process of finding the best number in the game can be approximated by finding the node with the highest win rate, and MCTS calculates the odds for each node and finds the node with the highest win rate. have.
  • MCTS does not explore all of the possibilities, but instead obtains game results through multiple random simulations and applies them to the odds of nodes. do.
  • the MCTS algorithm consists of four stages: Selection, Expansion, Simulation, and Back propagation.
  • the machine learning processor 120 includes a plurality of learning data in which the output of the accelerator injector, the control parameter value of each accelerator control device, and the accelerator final output quality value are matched, and among the plurality of learning data, After learning through supervised learning (SL) using the set number of learning data, optimal control parameter values may be calculated by reinforcement learning (RL).
  • SL supervised learning
  • RL reinforcement learning
  • 'supervised learning' means that supervised learning is learning in a state in which a label (explicitly correct answer) for data is given, and is manually selected by previously calculated big data (ie, a conventional researcher or the like). Learning based on artificial neural networks using the specified control parameters and the resulting accelerator output quality value, and reinforcement learning means that the agent takes some action for a given state and rewards it from it. It means learning while getting).
  • Such supervised learning or reinforcement learning also has a theoretical content corresponding to a known technique, and the feature of the present invention is that the learning method is used to calculate the optimal parameters of the accelerator control devices.
  • 3 is an example of data that is stored in advance for 'supervised learning' of the machine learning processor 120.
  • each row of FIG. 3 corresponds to the above-described learning data
  • the machine learning processor 120 extracts the learning data corresponding to several cases in a predetermined order or randomly from among the plurality of learning data collected in this manner. After that, 'supervised learning' can be performed in the artificial neural network using the extracted learning data.
  • 'I' is an output value of the injector
  • Q-BPM is an accelerator final output quality value
  • the rest corresponds to control parameters of each accelerator control device.
  • the parameter determination system 100 described in the above embodiments may be operated in conjunction with, for example, an EPICS based accelerator control system.
  • FIG. 4 shows a conventional EPICS based accelerator control system
  • FIG. 5 shows that the control devices in this EPICS based accelerator control system are replaced with the device simulator 110 according to the present invention and the machine learning processor 120 is added and operated. The form is shown.
  • accelerator control devices such as LLRF and BPM are arranged for several decades in the entire accelerator, and perform optimization in each section.
  • the accelerator control devices are configured for Pv data for each time period given to each device in the EPICS IOC.
  • the EPICS IOC Input Output Controller
  • the EPICS IOC Input Output Controller
  • the EPICS IOC Input Output Controller
  • AA Archiver Appliance
  • FIG. 5 illustrates a state in which such an accelerator control device is replaced with a device simulator 110 and a machine learning processor 120 communicating with the device simulator 110 is added.
  • the device simulator 110 By replacing the accelerator control device which is actually operated with the device simulator 110 as shown in FIG. 5, it is possible not only to prevent the operation of the accelerator for deriving optimal control parameters, but also the device simulator 110 can learn from the artificial neural network. By having the shape made, it is possible to be quite close to the characteristics of the actual accelerator control device.
  • the process of performing each of the above-described embodiments can be performed by a program or an application stored in a predetermined recording medium (for example, computer readable).
  • the recording medium includes both an electronic recording medium such as a random access memory (RAM), a magnetic recording medium such as a hard disk, an optical recording medium such as a compact disk (CD), and the like.
  • the program stored in the recording medium may be executed on hardware such as a computer or a smartphone to perform the above-described embodiments.
  • at least one of the above-described functional blocks of the present invention may be implemented by such a program or application.
  • the present invention by performing artificial neural network-based machine learning on the control parameters of each accelerator control device, it is possible to quickly determine the value of the optimum control parameters to increase the accelerator final output quality.
  • the device simulator has a form in which the learning is performed based on artificial neural networks. It can be very close to the characteristics of, which increases the reliability of the simulation results.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)

Abstract

L'invention concerne un système de détermination de paramètres pour dispositifs de commande d'accélérateur. Le système de détermination de paramètres pour dispositifs de commande d'accélérateur de la présente invention comprend : une pluralité de simulateurs de dispositif correspondant à une pluralité de dispositifs de commande d'accélérateur respectifs, destinés à l'apprentissage et la simulation sur la base d'un réseau neuronal artificiel ; et une unité de traitement d'apprentissage automatique destinée à spécifier un ensemble de collecte de paramètres de commande correspondant à au moins une collecte de paramètres de commande dont dispose la pluralité de simulateurs de dispositif, et destinée à apprendre, par l'intermédiaire d'un processus d'apprentissage sur la base d'un réseau neuronal artificiel, des changements des valeurs des paramètres de commande inclus dans l'ensemble de collecte de paramètres de commande et la qualité de sortie finale d'un accélérateur en fonction des changements, puis calculer les valeurs optimales des paramètres de commande qui peuvent maximiser la qualité de sortie finale de l'accélérateur.
PCT/KR2019/000514 2018-02-23 2019-01-14 Système de détermination des paramètres de commande pour optimiser les performances d'un accélérateur à l'aide de techniques d'apprentissage par renforcement et d'apprentissage automatique Ceased WO2019164125A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2018-0021958 2018-02-23
KR1020180021958A KR20190101677A (ko) 2018-02-23 2018-02-23 강화학습과 머신러닝 기법을 이용한 가속기 성능 최적화를 위한 제어 패러미터 결정 시스템

Publications (1)

Publication Number Publication Date
WO2019164125A1 true WO2019164125A1 (fr) 2019-08-29

Family

ID=67687779

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2019/000514 Ceased WO2019164125A1 (fr) 2018-02-23 2019-01-14 Système de détermination des paramètres de commande pour optimiser les performances d'un accélérateur à l'aide de techniques d'apprentissage par renforcement et d'apprentissage automatique

Country Status (2)

Country Link
KR (1) KR20190101677A (fr)
WO (1) WO2019164125A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3835894A1 (fr) * 2019-12-09 2021-06-16 Siemens Aktiengesellschaft Procédé et appareil de configuration d'une commande de machine

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102418794B1 (ko) * 2020-06-02 2022-07-08 오픈엣지테크놀로지 주식회사 하드웨어 가속기를 위한 파라미터를 메모리로부터 액세스하는 방법 및 이를 이용한 장치
KR102610429B1 (ko) 2021-09-13 2023-12-06 연세대학교 산학협력단 인공신경망과 연산 가속기 구조 통합 탐색 장치 및 방법
US11995577B2 (en) * 2022-03-03 2024-05-28 Caterpillar Inc. System and method for estimating a machine's potential usage, profitability, and cost of ownership based on machine's value and mechanical state
KR20240144573A (ko) 2023-03-24 2024-10-02 연세대학교 산학협력단 강성 제약을 고려한 인공신경망과 연산 가속기 구조 통합 탐색 장치 및 방법

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20150016538A (ko) * 2012-05-07 2015-02-12 독립행정법인 방사선 의학 종합연구소 조사 계획 장치, 및 하전입자 조사 시스템
KR20160030839A (ko) * 2014-09-11 2016-03-21 스미도모쥬기가이 이온 테크놀로지 가부시키가이샤 이온주입장치 및 이온빔의 조정방법
KR101729694B1 (ko) * 2017-01-02 2017-04-25 한국과학기술정보연구원 시뮬레이션 결과 예측 방법 및 장치
JP2017182129A (ja) * 2016-03-28 2017-10-05 ソニー株式会社 情報処理装置。
KR20170110190A (ko) * 2016-03-22 2017-10-11 한국원자력연구원 입자 가속기의 빔 전하량 극대화 장치 및 이를 이용한 입자 가속기의 빔 전하량 극대화 방법

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20070054457A (ko) 2005-11-23 2007-05-29 삼성전자주식회사 입자 가속기

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20150016538A (ko) * 2012-05-07 2015-02-12 독립행정법인 방사선 의학 종합연구소 조사 계획 장치, 및 하전입자 조사 시스템
KR20160030839A (ko) * 2014-09-11 2016-03-21 스미도모쥬기가이 이온 테크놀로지 가부시키가이샤 이온주입장치 및 이온빔의 조정방법
KR20170110190A (ko) * 2016-03-22 2017-10-11 한국원자력연구원 입자 가속기의 빔 전하량 극대화 장치 및 이를 이용한 입자 가속기의 빔 전하량 극대화 방법
JP2017182129A (ja) * 2016-03-28 2017-10-05 ソニー株式会社 情報処理装置。
KR101729694B1 (ko) * 2017-01-02 2017-04-25 한국과학기술정보연구원 시뮬레이션 결과 예측 방법 및 장치

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3835894A1 (fr) * 2019-12-09 2021-06-16 Siemens Aktiengesellschaft Procédé et appareil de configuration d'une commande de machine

Also Published As

Publication number Publication date
KR20190101677A (ko) 2019-09-02

Similar Documents

Publication Publication Date Title
WO2019164125A1 (fr) Système de détermination des paramètres de commande pour optimiser les performances d'un accélérateur à l'aide de techniques d'apprentissage par renforcement et d'apprentissage automatique
Wu et al. A novel sparrow search algorithm for the traveling salesman problem
US20160110640A1 (en) Time-division multiplexed neurosynaptic module with implicit memory addressing for implementing a neural network
US10204118B2 (en) Mapping neural dynamics of a neural model on to a coarsely grained look-up table
WO2020159052A1 (fr) Système de commande d'accélérateur en temps réel utilisant un simulateur de réseau neuronal artificiel et un contrôleur d'apprentissage par renforcement
CN117728064A (zh) 一种退役动力电池拆解工艺的优化方法
WO2022124449A1 (fr) Procédé d'optimisation d'hyper paramètre d'un algorithme d'intelligence artificielle légère à l'aide d'un algorithme génétique
Huang et al. Graph theory-based approach for automatic recognition of CAD data
CN111282281B (zh) 图像处理方法及装置、电子设备和计算机可读存储介质
CN117806170B (zh) 一种微束聚焦控制方法及装置
Herajy et al. An improved simulation of hybrid biological models with many stochastic events and quasi-disjoint subnets
Sayama Generative network automata: A generalized framework for modeling complex dynamical systems with autonomously varying topologies
Tian et al. Learning Bayesian networks with hidden variables using the combination of EM and evolutionary algorithms
Cuomo et al. A performance evaluation of a parallel biological network microcircuit in neuron
Azumagakito et al. Cyclic behavior in gene-culture coevolution mediated by phenotypic plasticity in language
Sehitoglu et al. A building block favoring reordering method for gene positions in genetic algorithms
Colvett et al. Simulating cyberattacks with a Petri net discrete event simulator
Alidmat et al. Cellular automata model for pedestrian evacuation in fire spreading conditions
Du et al. Analyzing behavior and intention in multi-agent systems using graph neural networks
CN113077055A (zh) 一种基于FBPF-EIKF-FastSLAM的分布式多源融合定位方法
CN113935461A (zh) 一种带有注意力机制批量训练的树循环神经网络算法
WO2021167257A1 (fr) Procédé et dispositif de recommandation ppt
CN114363095B (zh) 基于petri网的系统脆弱性分析方法、系统和介质
Mishra et al. Modified multi-verse optimiser used for global optimisation
Tan et al. The analysis of deterministic and stochastic SIM model with general information intervention

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19757626

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19757626

Country of ref document: EP

Kind code of ref document: A1