Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Enhanced learning method for calibrating beam deviation of accelerator

A technology of reinforcement learning and accelerators, applied in neural learning methods, biological neural network models, electrical components, etc., can solve problems such as high-energy safety hazards, track deviation, and inefficiency

Active Publication Date: 2019-09-24
LANZHOU UNIVERSITY
View PDF3 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The medium-energy beam transmission section of the proton linear accelerator is installed by multiple quadrupole magnets along the center of the axis. Due to the interaction between the installation accuracy and the surrounding complex magnetic field, it is inevitable that the accelerated high-energy proton beam will be damaged during the movement. Orbital offset, too much offset will affect the quality of the protons entering the superconducting cavity, and even have high-energy safety hazards
The current proton beam orbital offset correction mainly relies on complex physical methods and a large number of mathematical operations to calculate the orbital offset, and then continuously input the voltage value of the magnet coil for calibration. Since the proton linear accelerator system is a complex with many variables system, it is very inefficient to use repeated debugging coil by coil, and there is basically no method to directly and automatically correct the voltage value of the magnet coil according to the position information of the beam current movement.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Enhanced learning method for calibrating beam deviation of accelerator
  • Enhanced learning method for calibrating beam deviation of accelerator
  • Enhanced learning method for calibrating beam deviation of accelerator

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0018] In order to make the content, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail in conjunction with the accompanying drawings. Modeling the environment for reinforcement learning in Figure-1:

[0019] Step 1): Establish a reinforcement learning method, mainly including environment, agent, calibration beam reward mechanism, deterministic strategy, etc.

[0020] Step 1.1: The environment is the calibration coil voltage value and position detector (BPM) value of the accelerator system, which can be directly read in the system (or directly read from the system database).

[0021] Step 1.2: The agent is the core component of deep reinforcement learning, and its specific neural network is described in step 4.2.

[0022] Step 1.3: The calibration beam reward mechanism is a reward and punishment rule for judging whether the position is good or bad according to the position after the beam position is c...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an enhanced learning method for calibrating the beam deviation of an accelerator. In an intermediate energy beam transmission section of the accelerator, the position of the beam deviates due to the influence of the installation accuracy of equipment and the surrounding complex environment, which seriously affects the energy level to which the beam can reach. According to the traditional method, the calibration voltage value is acquired through complex physical calculation, a script program is used to automatically input for continuously trying, and the process is complex and tedious. According to the invention, calibration coils integrated in three groups of horizontal and vertical quadrupole magnets in the intermediate energy beam transmission section are analyzed, and the accelerator environment is modeled by using the characteristic of interactive learning between the environment and an intelligent agent by relying on enhanced learning, thereby being a beam deviation calibration method which uses a deterministic strategy to explore the continuous large state space and motion space and approaches to the optimal calibration voltage value by using a neural network.

Description

technical field [0001] The invention relates to a reinforcement learning method for calibrating accelerator beam offset. Background technique [0002] The proton linear accelerator is a scientific device with high beam intensity and easy particle injection and extraction, which is composed of high-frequency power ion source, accelerating electrode, target chamber, and direct-space system. The medium-energy beam transmission section of the proton linear accelerator is installed by multiple quadrupole magnets along the center of the axis. Due to the interaction between the installation accuracy and the surrounding complex magnetic field, it is inevitable that the accelerated high-energy proton beam will be damaged during the movement. Orbital offset, too much offset will affect the quality of the protons entering the superconducting cavity, and even pose high-energy safety hazards. The current proton beam orbital offset correction mainly relies on complex physical methods and...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): H05H7/00H05H7/22G06N3/08
CPCG06N3/08H05H7/001H05H7/22H05H2007/002Y04S10/50
Inventor 周庆国王金强杨旭辉雍宾宾申泽邦谢启荣武强
Owner LANZHOU UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products