IPAC2019 - List of Authors (Leng, Y.B.)

Paper	Title	Page
WEPGW045	Application of Clustering by Fast Search and Find of Density Peaks to Beam Diagnostics at SSRF	2581
	R. Jiang, Y.B. Leng SSRF, Shanghai, People’s Republic of China
	With the increased technological complexity of accelera-tors, meeting the demand of beam diagnostics and opera-tion need more powerful and faster methods. And detect-ing the accuracy and stability of beam position moni-tors(BPMs) are important for all kinds of measurement systems and feedback systems in particle accelerator field. As an effective tool for data analysis and automa-tion, the machine learning methods had been used in accelerator physics field, recently. Among machine learn-ing methods, the clustering by fast search and find of density peaks as a typical unsupervised learning algo-rithms could be performed directly without training in arbitrary accelerator systems and could discover un-known patterns in the data. This paper used clustering by fast search and find of density peaks to detect faulty beam position monitor or monitoring beam orbit stability by analysis five typical parameters, that is beta oscilla-tion of X and Y direction(BetaX and BetaY), transverse oscillation of X and Y direction(AmpX and AmpY) and energy oscillation(AmpE). The results showed that cluster-ing by fast search and find of density peaks could classi-fy beam data into different clusters on the basis of their similarity. And that, aberrant run data points could be detected by decision graph. Morever, analysis results demonstrate the characteristic parameters AmpE, AmpX and BetaX amplitude have the same effect to distinguish the faulty BPMs and the AmpY and the BetaY amplitude are also.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW045
About •	paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW049	Deep Learning Applied for Multi-Slit Imaging Based Beam Size Monitor	2587
	B. Gao, Y.B. Leng SSRF, Shanghai, People’s Republic of China X.Y. Xu SINAP, Shanghai, People’s Republic of China
	In order to satisfy the requirement of high speed measurement and improve the accuracy of BSM (beam size monitor), multi-slit imaging based BSM has been proposed by SSRF at 2017. However, it is very difficult to deconvolve the image and figure out the beam size, which requires dedicated algorithms to solve this issue. Deep learning is one of the most popular algorithms, which can learn to mimic any distribution of data. In the region of Beam instrumentation, they can be taught to deal with many difficult problem. In this paper, multi-layer neural network is used to process the images from the multi-slit imaging system. Training processes, struct of the neural networks and the result of the experiments will be presented.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW049
About •	paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW057	Design of Stripline BPM for the SHINE Project	2605
	T. Wu, B. Gao, L.W. Lai, Y.B. Leng SSRF, Shanghai, People’s Republic of China S.S. Cao, J. Chen, Y.M. Zhou SINAP, Shanghai, People’s Republic of China
	As a under-constrution forth-generation light source in China, SHINE(Shanghai HIgh repetition rate XFEL aNd Extreme light facility) is expected to play an important role in basic scientific research in the field of materials and medicine. However, the performance of FEL depends critically on the completeness and quality of their beam diagnostic systems. Since the SHINE project will operate with bunch charge at 100pC, which is only one-quarter of that in the SXFEL, the measurement accuracy requirements for SBPM will increase significantly. On the other hand, the bunch repetition frequency of SHINE reached 1MHz, which shortened the threshold for measuring dead time. Fitting the requirement, the passband and the sampling rate design of stripline BPM is upgraded for the SHINE project. The final design was simulated using the data on the SXFEL, and the some inspiring results have been made.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW057
About •	paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW061	Bunch Length Measurement Using Multi-Frequency Harmonic Analysis Method at SSRF	2616
	Y.M. Zhou, B. Gao, Y.B. Leng, N. Zhang SSRF, Shanghai, People’s Republic of China
	Harmonics method in the frequency domain is an effective and inexpensive bunch length measurement method, which was implemented at the Shanghai Synchrotron Radiation Facility (SSRF). A multi-frequency bunch-bybunch length measurement system using an integrated RF conditioning module will be established to reduce the system noise and signal reflection, and to improve the bunch length measurement accuracy as well. The module consists of power splitters, band-pass filters, mixers and so on. The main function of the integrated RF conditioning module is to extract the beam signals at 500MHz, 1.5GHz, 2GHz, and 3GHz operating frequency. Raw data are acquired by a high-precision digitizer and analyzed by MATLAB code. The absolute bunch length can be obtained with a streak camera, which was used to calibrate the response coefficients of the system. Bunch-by-bunch length can be measured by the multi-frequency harmonic analysis method from the button BPM
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW061
About •	paper received ※ 15 May 2019 paper accepted ※ 18 May 2019 issue date ※ 21 June 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW064	Machine Learning Application in Bunch Longitudinal Phase Measurement	2625
	X.Y. Xu, Y.M. Zhou SINAP, Shanghai, People’s Republic of China Y.B. Leng SSRF, Shanghai, People’s Republic of China
	High resolution bunch-by-bunch longitudinal phase measurement has been realized at Shanghai Synchrotron Radiation Facility (SSRF). In order to fully exploit the potency of the bunch phase monitor, the transient state during injection is being further studied. A longitudinal phase fitting method was used to study the synchrotron damping oscillation in injection events, where we can get the energy offsets between the injector and the storage ring, refilled bunch arrived time and the synchrotron damping time. However, manual multi-parameter fitting of nonlinear functions is awfully complex and slow. Machine learning algorithms, such as gradient descent and artificial neural network (ANN) is more suitable to do this fitting. Through these methods, we can quickly obtain more accurate fitting parameters and further realize online measurement of the refilled charge arrived time, energy offsets between the injector and storage ring, and the synchrotron damping time.
DOI •	reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW064
About •	paper received ※ 13 May 2019 paper accepted ※ 20 May 2019 issue date ※ 21 June 2019
Export •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Page

Application of Clustering by Fast Search and Find of Density Peaks to Beam Diagnostics at SSRF

2581

R. Jiang, Y.B. Leng
SSRF, Shanghai, People’s Republic of China

With the increased technological complexity of accelera-tors, meeting the demand of beam diagnostics and opera-tion need more powerful and faster methods. And detect-ing the accuracy and stability of beam position moni-tors(BPMs) are important for all kinds of measurement systems and feedback systems in particle accelerator field. As an effective tool for data analysis and automa-tion, the machine learning methods had been used in accelerator physics field, recently. Among machine learn-ing methods, the clustering by fast search and find of density peaks as a typical unsupervised learning algo-rithms could be performed directly without training in arbitrary accelerator systems and could discover un-known patterns in the data. This paper used clustering by fast search and find of density peaks to detect faulty beam position monitor or monitoring beam orbit stability by analysis five typical parameters, that is beta oscilla-tion of X and Y direction(BetaX and BetaY), transverse oscillation of X and Y direction(AmpX and AmpY) and energy oscillation(AmpE). The results showed that cluster-ing by fast search and find of density peaks could classi-fy beam data into different clusters on the basis of their similarity. And that, aberrant run data points could be detected by decision graph. Morever, analysis results demonstrate the characteristic parameters AmpE, AmpX and BetaX amplitude have the same effect to distinguish the faulty BPMs and the AmpY and the BetaY amplitude are also.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW045

About •

paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW049

Deep Learning Applied for Multi-Slit Imaging Based Beam Size Monitor

2587

B. Gao, Y.B. Leng
SSRF, Shanghai, People’s Republic of China
X.Y. Xu
SINAP, Shanghai, People’s Republic of China

In order to satisfy the requirement of high speed measurement and improve the accuracy of BSM (beam size monitor), multi-slit imaging based BSM has been proposed by SSRF at 2017. However, it is very difficult to deconvolve the image and figure out the beam size, which requires dedicated algorithms to solve this issue. Deep learning is one of the most popular algorithms, which can learn to mimic any distribution of data. In the region of Beam instrumentation, they can be taught to deal with many difficult problem. In this paper, multi-layer neural network is used to process the images from the multi-slit imaging system. Training processes, struct of the neural networks and the result of the experiments will be presented.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW049

About •

paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW057

Design of Stripline BPM for the SHINE Project

2605

T. Wu, B. Gao, L.W. Lai, Y.B. Leng
SSRF, Shanghai, People’s Republic of China
S.S. Cao, J. Chen, Y.M. Zhou
SINAP, Shanghai, People’s Republic of China

As a under-constrution forth-generation light source in China, SHINE(Shanghai HIgh repetition rate XFEL aNd Extreme light facility) is expected to play an important role in basic scientific research in the field of materials and medicine. However, the performance of FEL depends critically on the completeness and quality of their beam diagnostic systems. Since the SHINE project will operate with bunch charge at 100pC, which is only one-quarter of that in the SXFEL, the measurement accuracy requirements for SBPM will increase significantly. On the other hand, the bunch repetition frequency of SHINE reached 1MHz, which shortened the threshold for measuring dead time. Fitting the requirement, the passband and the sampling rate design of stripline BPM is upgraded for the SHINE project. The final design was simulated using the data on the SXFEL, and the some inspiring results have been made.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW057

About •

paper received ※ 15 May 2019 paper accepted ※ 21 May 2019 issue date ※ 21 June 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW061

Bunch Length Measurement Using Multi-Frequency Harmonic Analysis Method at SSRF

2616

Y.M. Zhou, B. Gao, Y.B. Leng, N. Zhang
SSRF, Shanghai, People’s Republic of China

Harmonics method in the frequency domain is an effective and inexpensive bunch length measurement method, which was implemented at the Shanghai Synchrotron Radiation Facility (SSRF). A multi-frequency bunch-bybunch length measurement system using an integrated RF conditioning module will be established to reduce the system noise and signal reflection, and to improve the bunch length measurement accuracy as well. The module consists of power splitters, band-pass filters, mixers and so on. The main function of the integrated RF conditioning module is to extract the beam signals at 500MHz, 1.5GHz, 2GHz, and 3GHz operating frequency. Raw data are acquired by a high-precision digitizer and analyzed by MATLAB code. The absolute bunch length can be obtained with a streak camera, which was used to calibrate the response coefficients of the system. Bunch-by-bunch length can be measured by the multi-frequency harmonic analysis method from the button BPM

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW061

About •

paper received ※ 15 May 2019 paper accepted ※ 18 May 2019 issue date ※ 21 June 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

WEPGW064

Machine Learning Application in Bunch Longitudinal Phase Measurement

2625

X.Y. Xu, Y.M. Zhou
SINAP, Shanghai, People’s Republic of China
Y.B. Leng
SSRF, Shanghai, People’s Republic of China

High resolution bunch-by-bunch longitudinal phase measurement has been realized at Shanghai Synchrotron Radiation Facility (SSRF). In order to fully exploit the potency of the bunch phase monitor, the transient state during injection is being further studied. A longitudinal phase fitting method was used to study the synchrotron damping oscillation in injection events, where we can get the energy offsets between the injector and the storage ring, refilled bunch arrived time and the synchrotron damping time. However, manual multi-parameter fitting of nonlinear functions is awfully complex and slow. Machine learning algorithms, such as gradient descent and artificial neural network (ANN) is more suitable to do this fitting. Through these methods, we can quickly obtain more accurate fitting parameters and further realize online measurement of the refilled charge arrived time, energy offsets between the injector and storage ring, and the synchrotron damping time.

DOI •

reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2019-WEPGW064

About •

paper received ※ 13 May 2019 paper accepted ※ 20 May 2019 issue date ※ 21 June 2019

Export •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)