Research on Transient Pulse Signal Recognition Algorithm in Particle Accelerator Devices.docx

Research on transient pulse signal recognition algorithms in particle accelerator devices

Abstract: Particle accelerator devices belong to complex high-tech large scientific engineering equipment, and the electromagnetic environment is exceptionally complicated. Transient pulse signals negatively impact the operation of equipment in particle accelerator devices through conduction emission, cable crosstalk, near-field coupling, and other means. Therefore, identifying transient pulse signals in particle accelerator devices is an important topic for alleviating electromagnetic compatibility issues in particle accelerators. This paper systematically analyzes the causes and main effects of transient pulse signals generated by particle accelerator devices and proposes a transient pulse signal identification algorithm based on the threshold method. First, a testing system is constructed based on the IEC-61000-4-4 standard to obtain the pulse signal spectrum; second, a graphical pixel processing method is used to denoise the test spectrum; finally, based on the threshold identification algorithm, transient pulses are extracted, and the accuracy of different thresholds in signal identification is compared using the signal-to-noise ratio to define the optimal threshold, providing support for interference signal identification and spectrum management in particle accelerator devices.

Keywords: particle accelerator device, electromagnetic compatibility, signal recognition, transient pulse signal

Introduction

There are devices in particle accelerator systems that can generate transient pulse signals, such as high-power microwaves, high-current high-voltage pulses, high-power switch power supplies, and AC power supply systems.[1] These devices are characterized by short duration, wide frequency bandwidth, and concentrated energy, which not only affect the stability and reliability of the particle accelerator systems but may also pose potential risks to the health and safety of personnel. Therefore, how to identify and extract transient pulse signals has become a very important topic in the electromagnetic compatibility of particle accelerator systems, providing support to alleviate electromagnetic compatibility issues, which is of great significance.

Currently, many studies have been conducted on the characteristics of pulse signals. Among the existing testing methods, the IEC-61000-4-5 standard, provides testing methods for voltage surge disturbances, but the rise time and duration are difficult to adapt to overvoltage situations in actual power systems. In terms of suppressing pulse signal generation, literature [3][4][5] and literature [6] [7] have analyzed the causes of overvoltage (pulse signal) generation and some suppression schemes, most of these studies focus on analyzing the causes of interference signals or suppression, but cannot accurately eliminate the sources of interference. Therefore, using pulse transient signal identification methods to quickly locate and accurately extract interference signals is the best strategy.The current methods for identifying pulse transient signals have become quite refined, such as the transient signal extraction method based on wavelet packet selection and NLLS, which performs excellently in high noise environments (astronomical and bandwidth signals), but has high computational complexity and requires significant computational resources; the MCDPMF method applied to power equipment discharge, mechanical fault diagnosis, etc., can retain the details of the signal, but has high algorithm complexity, poor real-time performance, and requires repeated optimization for different signals; using short-time Fourier transform or wavelet transform to detect transient features in blasting vibration signals is quite challenging; using the threshold valley method to extract signals by calculating dynamic thresholds through signal-to-noise ratio to separate burst signals, but the setting of dynamic thresholds has a significant impact on the results and may require multiple adjustments.

This article focuses on the transient pulse signals in particle accelerator devices and proposes a pulse transient signal recognition algorithm based on the threshold method. By accurately identifying the transient pulse signals present in the particle accelerator, the goal of mitigating electromagnetic interference is achieved. To effectively verify the accuracy of the algorithm, based onIEC61000-4-4[12], a100kHz standard pulse signal is used for algorithm research. The experimental results show that the pulse transient signal recognition algorithm based on the threshold method can effectively and accurately extract transient pulse signals. This is of great significance for analyzing transient pulse signals in particle accelerator devices.

Analysis of the causes of pulse transient signal generation

The pulsed transient signals in particle accelerator devices are the result of various electromagnetic effects and operational characteristics of the equipment. The main sources of these signals include overvoltage in power system operations, electromagnetic radiation from high-power devices, switching processes of nonlinear devices, electromagnetic crosstalk between cables, parasitic effects in high-current pulse systems, arc discharge phenomena, and electromagnetic induction caused by changes in particle beam flow. These signals propagate through conduction, radiation, and electromagnetic coupling, causing interference and instability in the operation of the equipment within the complex electromagnetic environment of the particle accelerator. Their impact may manifest as data transmission errors, abnormal device functions, or even a reduction in overall operational reliability.

Essentially, these transient pulse signals originate from the rapid changes in current, voltage, and magnetic fields during the operation of the device. Taking induced voltage as an example, when a highly inductive load is disconnected by a circuit breaker, the self-induction effect caused by the sudden change in current can trigger high-amplitude transient voltage; meanwhile, between cables, due to distributed capacitance and magnetic coupling, high-frequency signals in the main cable can be transmitted to adjacent cables through electromagnetic coupling, resulting in crosstalk interference. In addition, the switching actions of nonlinear devices, due to the rich harmonic components, can also generate wideband transient interference signals. Furthermore, the rapid changes in particle beam currents further complicate the operational environment of the equipment through induced voltage or parasitic effects. The amplitude variation of these pulse signals can be expressed by the formula 1:

$V_{} (t) =L \frac{ⅆⅈ (t)}{ⅆt} +C \frac{ⅆv (t)}{ⅆt} +N \frac{ⅆΦ}{ⅆt}$

Among them: $L \frac{ⅆⅈ (t)}{ⅆt}$ is the induced voltage caused by sudden changes in the device's operating current, $C \frac{ⅆv (t)}{ⅆt}$ is the capacitive coupling interference caused by rapid voltage changes, $N \frac{ⅆΦ}{ⅆt}$ is the induced voltage caused by changes in the magnetic field.

Research on recognition algorithms

Experimental data acquisition

In order to obtain standard transient pulse signals for effective research on signal recognition algorithms, this article is based onIEC61000-4-4requirements to construct a test platform, as shown inFigure 1. The goal is to acquire100kHztransient pulse signals through a complete testing link. The main equipment is shown inTable 1.

Test frequency bandis9kHz~30MHz,the purpose is to collect100kHztransient pulse signals while maintaining synchronization with the conducted emission test frequency band. After testing,2*1578matrix spectrum data was obtained.

Table1Each devicemodeland parameters

Device name	Model	Parameter
Current probe	R&S EZ-17	20Hz-100MHz
Capacitive coupling clamp	ISO7637-3	Voltage:1000V, Capacitance:100pF
Anti-jamming signal simulator	Compact NX5 bs-1-300-16	1Hz-1000kHz
Spectrum analyzer	R&S FSVR7	10Hz-40GHz

Figure1 Schematic diagram of the test platform

Denoising process

Due to the high signal-to-noise ratio noise in the spectral data, it is not conducive to the identification and extraction of transient pulse signals. Therefore, further denoising of the spectral data is required. First, the spectral data is processed using image pixel values and read in grayscale mode. Secondly, adaptive thresholding is used to extract the brightness values of the pixels. Finally, the data points are inferred from the pixel positions to obtain the corresponding amplitude and frequency, thus proposing pixel points to achieve the denoising function. The comparison of data before and after denoising is shown in Figure 2.

Figure2Before denoising (left), after denoising (right)

Signal recognition and extraction

Thresholdis a common signal processing method used to extract specific signals or features from data. By setting a specific value, the data is divided into two parts: one is the signal part that meets the criteria, and the other is the noise or irrelevant signal part that is below the threshold. Specifically, given the thresholdT, the classification rule for the signalxcan be expressed usingformula2:

$x^{} = {\begin{matrix} x, & if x> T \\ 0, & if x ≤T \end{matrix}$

（2）

Among them, $x^{}$ represents the result after threshold processing, $x$ represents the signal value, $T$ represents the threshold. When the signal $x$ is less than or equal to the threshold $T$ , it is considered noise or an irrelevant signal, set to0or ignored. Figure3shows the extraction results of the pulse signal under different thresholds.

From the figure, it can be seen that the extraction effects vary with different thresholds. The lower the threshold, the more noise is included in the extracted signal. As the threshold increases, when it reaches a certain level, the extracted signal consists only of pulse signals.The signal-to-noise ratio (SNR) represents the ratio of the energy of the signal to the energy of the noise, usually expressed in decibels (dBm) as the unit. $x_{}$ is the value of the original signal, $_{}$ is the value of the extracted signal, $N$ is the number of signal sampling points.The higher the SNR, the more dominant the signal is, and the less impact noise has on the signal (as shown in formula3).

$SNR =10* (\frac{\sum_{}^{}}{\sum_{}^{} {(x_{} -_{})}^{}})$

(3)

As shown in4, the curve of SNR varies with different threshold values.When the threshold reaches26dBm, the extraction effect of the signal is optimal, with a signal amplitude of13dBm; when the threshold continues to increase, its amplitude value remains unchanged until it reaches the maximum value of the pulse signal. Comparing with the original signal graph (Figure2), it is found that after processing with the threshold method, when the threshold is above26dBm, we can accurately extract the pulse transient signal without any other noise signals (as shown in Figure5), indicating that this threshold method is suitable for identifying and extracting pulse signals.

Figure4 SNR variation curve with different thresholds

Figure5Pulse Transient Signal Extraction Result

Conclusion

This article focuses on thetransient pulse signalsin particle accelerator devices, and proposes a method for identifying and analyzing pulse transient signals based on the threshold method. This method can effectively detecttransientpulse signalsgenerated due to reasons such as the removal of inductive loads or switching power supplies, throughthe graphical pixel processing methodfor noise reduction and extraction of complex signal data in the spectrum, and then effectively extract transientpulse signalswith a threshold of26dBmor above using the threshold method. Multiple tests have proven that this method is simple and convenient, has good adaptability, and a high signal recognition rate, capable of accurately identifyingtransientpulse signals.Subsequently, this method will be applied to the measured spectral data of particle accelerator devices for the identification and extraction of transient pulse signals, combining theory with practice, which not only effectively verifies the validity of the algorithm but also provides a basis for the identification and mitigation of electromagnetic interference signals in particle accelerator devices.

References

[1] H. Zhao, H. Xu, G. Xiao. "Huizhou accelerator complex facility and its future development (in Chinese)," Sci Sin-Phys Mech Astron,2020,50: 112006

[2] IEC 61000-4-5, “Electromagnetic compatibility (EMC) – Part 4-5: Testing and measurement techniques – Surge immunity test”.

[3] H. Nie, X. Liu, Y. Wang, Y. Yao, Z. Gu and C. Zhang, "Breaking Overvoltage of Dry-Type Air-Core Shunt Reactors and its Cumulative Effect on the Interturn Insulation," IEEE Access, vol. 7, pp. 55707-55720, 2019.

[4] X. Lei, T. Cui, X. Wu, H. Zeng, Q. Liu, "Research on Resonance Overvoltage of EHV Transmission Lines Caused by Lightning Strike," IEEE Access, vol. 9, pp. 123777-123785, 2021.

[5] C. Li, R. Chang, N. Zhang, "Transient Characteristics of the Vacuum Circuit Breaker Cut off the No-Load Distribution Transformer in 10kV Large Current Grounding System," Chinese Journal OfVacuum Science And Technology, vol. 42, pp. 943-951, 2022.

[6] K. Zhang, Y. Ma, Z. Liu, "Experimental Analysis Of SF6 Circuit Breaker Switching 500kV Unloaded Transformer," High Voltage Apparatus, vol. 2, pp.37-39, 2002

[7] H. Liu, S. Chen, X. Sun, "Simulation Research on Overvoltage Suppression of Vacuum Circuit Breaker Switching on 10 kV Shunt Reactor," High Voltage Apparatus, vol. 55, pp. 0032-0037, 2019．

[8] M. Chendeb, G. Plantier and A. Sourice, "Parameters and DOA estimation based on wavelet packets selection for transient signals in colored noise," 2007 15th European Signal Processing Conference, Poznan, Poland, 2007, pp. 2469-2473.

[9]Yuan ChaojieResearch on Strong Transient Signal Recognition and Data Compression Storage Technology[D].Beijing University of Information Science and Technology,2019.DOI:10.26966/d.cnki.gbjjc.2019.000008.

[10]Zhang YanchengResearch on Extraction Methods of Transient Signals Based on Human Ear Auditory Characteristics[D].Northeast University,2014.

[11]Wang LiyunResearch on Interference Signal Statistics and Data Management Methods at the Radio Astronomy Observatory[D].Xinjiang University,2019.

[12]IEC 61000-4-4, “Electromagnetic compatibility (EMC) – Part 4-4: Testing and measurement techniques – Electrical fast transient/burst immunity test, IDT”．

[13]Qiu NeminResearch and Application of High-Efficiency Target Detection Technology Based on Pulse Signals[D].Beijing University of Posts and Telecommunications, 2024. DOI:10.26969/d.cnki.gbydu.2024.000965.

[14]Yuan LiLiu QiSun Zhengwenet al.Method for detecting and identifying electromagnetic interference in equipment areas under on-site conditions[J].Journal of Electromagnetic Waves and Applications,2017,32(06):650-656.DOI:10.13443/j.cjors.2017082301.

[15]Yang HuarongFan YouwenLiu Ganget al.A method for converting spectrograms into electronic data[J].Ship Science and Technology,2010,32(09):80-83.

[16]P. Li, Y. Zheng, C. Han, L. -j. Song and B. -f. Cao, "Wavelet threshold methods used in lightning transient electrical signals denoising," 2012 9th International Conference on Fuzzy Systems and Knowledge Discovery, Chongqing, China, 2012, pp. 2027-2029, doi: 10.1109/FSKD.2012.6234045.

[17]J. Bai, J. Yao, Z. Yang, Y. Guo and L. Wang, "Transient Electromagnetic Weak Signal Extraction Method Based on Multi-Scale Combined Difference Product Morphological Filtering," in IEEE Access, vol. 10, pp. 22140-22149, 2022, doi: 10.1109/ACCESS.2022.3145974.

[18]Y. Hu, "Underwater transient signal detection under two noise reduction methods," 2024 IEEE 4th International Conference on Electronic Technology, Communication and Information (ICETCI), Changchun, China, 2024, pp. 929-932, doi: 10.1109/ICETCI61221.2024.10594443.

[19]Wei Zhang and Sumin Han, "Analysis of transient interference signal on switch operating system," 2010 International Conference on Computer, Mechatronics, Control and Electronic Engineering, Changchun, 2010, pp. 143-146, doi: 10.1109/CMCE.2010.5610202.

[20]Cao BaofengResearch on Weak Signal Detection Technology of Electromagnetic Pulse from Nuclear Explosion in Broadband Noise[D].Academy of Military Sciences,2019.