Search

CN-121995077-A - Charged particle speed detection method based on electrostatic field Fourier transform

CN121995077ACN 121995077 ACN121995077 ACN 121995077ACN-121995077-A

Abstract

The invention provides a charged particle speed detection method based on electrostatic field Fourier transform, which is applied to fault monitoring and electrostatic detection of a heavy-duty gas turbine, and comprises the steps of firstly capturing a time domain electrostatic signal generated by the movement of charged particles falling off due to faults through an electrostatic field sensor arranged on the inner wall of a gas circuit pipeline; the method comprises the steps of carrying out anti-aliasing filtering, amplifying and analog-to-digital conversion on signals to generate a high-fidelity time domain sequence, extracting normalized Gaussian pulse signals, zero filling to a length of 2 N , converting the signals to a frequency domain by utilizing fast Fourier transform, calculating 3dB bandwidth of the signals in the frequency domain, inverting the particle motion speed based on a linear relation obtained by calibration, and finally evaluating the fault type, the fault position and the fault severity according to speed distribution characteristics.

Inventors

  • ZHANG LAN
  • WEI XUEFEI
  • REN HONGYUN
  • CHENG HAOBIN
  • WEN XIAOLONG

Assignees

  • 中国联合重型燃气轮机技术有限公司
  • 北京科技大学

Dates

Publication Date
20260508
Application Date
20251231

Claims (8)

  1. 1. A method for detecting velocity of charged particles based on electrostatic field fourier transform, applied to a heavy duty gas turbine, the method comprising: s1, acquiring time domain electrostatic signals of charged particles in a gas circuit pipeline through an electrostatic field sensor; s2, performing anti-aliasing filtering, analog-to-digital conversion and normalization processing on the time domain electrostatic signals of the charged particles to obtain normalized Gaussian pulse signals; S3, sequentially performing zero filling and fast Fourier transform on the normalized Gaussian pulse signal to obtain frequency spectrum data, and calculating a 3dB bandwidth corresponding to the frequency domain signal based on the frequency spectrum data to obtain a frequency domain signal bandwidth; s4, based on the frequency domain signal bandwidth, obtaining an approximate linear relation between the shedding particle speed v and the frequency domain signal bandwidth, and obtaining the charged particle speed.
  2. 2. The method for detecting velocity of charged particles based on fourier transform of electrostatic field according to claim 1, wherein the anti-aliasing filtering in S2 comprises: An analog low-pass filter is connected between the electrostatic field sensor and the analog-to-digital converter; setting the cut-off frequency of the analog low-pass filter to be lower than the Nyquist frequency corresponding to the system sampling frequency; wherein the sampling frequency of the system is 22Hz.
  3. 3. The method for detecting velocity of charged particles based on fourier transform of electrostatic field according to claim 2, wherein the analog low-pass filter is a first-order filter or a multi-order filter.
  4. 4. A charged particle velocity detection method according to claim 3 wherein said anti-aliasing filtering and analog-to-digital conversion processing of the time domain electrostatic signal of the charged particles in S2 comprises: after the time domain electrostatic signals of the charged particles are subjected to anti-aliasing filtering, performing analog-to-digital conversion after being processed by a signal conditioning circuit to obtain a time domain signal sequence; The signal conditioning circuit is used for amplifying and filtering the received signals.
  5. 5. The method for detecting velocity of charged particles based on fourier transform of electrostatic field according to claim 4, wherein said normalization process in S2 comprises: Processing the time domain signal sequence based on a maximum absolute value normalization processing method so that all data points are mapped into an [ -1, 1] interval; the image pattern of the normalized gaussian pulse signal in the step S2 is a bell-shaped curve point-symmetrical with respect to the signal intensity peak.
  6. 6. The method of claim 5, wherein zero-filling the normalized gaussian pulse signal in S3 comprises: Adding a plurality of data points with the value of zero at the tail end of the normalized Gaussian pulse signal, and expanding the total length of the normalized Gaussian pulse signal to the power N of 2, wherein N is an integer greater than or equal to 10.
  7. 7. The method for detecting velocity of charged particles based on Fourier transform of electrostatic field according to claim 6, wherein the fast fourier transform in S3 comprises: applying a hanning window to the normalized Gaussian pulse signal after zero filling, and performing fast Fourier transform; Wherein, zero padding is performed in the process of performing fast Fourier transform; after the fast Fourier transform is completed, calculating a single-side frequency spectrum, carrying out peak normalization and taking logarithm of a frequency axis to obtain frequency spectrum data; Wherein the frequency range is set to 0 to fs/2 hz, fs being the sampling frequency.
  8. 8. The method for detecting velocity of charged particles based on Fourier transform of electrostatic field according to claim 6, wherein the approximate linear relation between the velocity v of the dropped particles in S4 and the bandwidth of the frequency domain signal is v=k.BW+b; wherein k and b are values obtained through linear regression analysis calibration, v is the speed of the falling particles, and BW is the frequency domain signal bandwidth.

Description

Charged particle speed detection method based on electrostatic field Fourier transform Technical Field The invention relates to the field of heavy gas turbine fault monitoring and static electricity detection, in particular to a charged particle speed detection method based on electrostatic field Fourier transform. Background Heavy duty gas turbines are large thermal machines that integrate precision aerodynamics, high temperature materials science and advanced control technology, and the core operating principle is based on the brayton cycle, which converts chemical energy of fuel (e.g., natural gas) into mechanical energy by continuously operating three components, the compressor, the combustor and the turbine. The device has the outstanding advantages of high single machine power, high heat efficiency, quick start and stop, flexible adjustment and the like, is widely applied to the core fields of power generation, petrochemical industry, aerospace, steel and the like, and is key equipment in a modern energy system. Heavy gas turbine systems generally operate in high temperature and high pressure environments, and various components are affected by extremely high thermal stress, mechanical stress, chemical corrosion and the like, so that gas circuit faults are very easy to occur. The failure of the gas path of the heavy-duty gas turbine pipeline can lead to the decrease of the output of the unit and the increase of the heat rate, and even causes unplanned shutdown when serious, thereby causing huge economic loss. The detection of the gas path pipeline faults of the heavy-duty gas turbine has important significance for enterprises to maximize equipment availability and reduce maintenance cost. At present, although the traditional detection methods such as performance monitoring, vibration analysis and the like can realize the detection of the gas path pipeline faults of the heavy gas turbine, obvious bottlenecks still exist: 1. Limitations of performance monitoring techniques Performance monitoring methods infer efficiency decay (e.g., fouling, corrosion) or flow loss (e.g., seal wear) by monitoring changes in compressor discharge pressure, discharge temperature, turbine discharge temperature, fuel flow, etc. parameters. However, these methods are slow to respond, typically alarm after significant degradation in performance, and are difficult to accurately locate and identify faults, subject to sensor drift and operating condition variations. 2. Deficiencies of vibration analysis techniques Vibration analysis technology is effective for mechanical faults such as rotor unbalance, misalignment, bearing wear and the like, but is insensitive to early gas circuit faults such as blade corrosion, coating spalling, hot channel cracks and the like, and can be detected only when the faults develop to influence dynamic balance, and the early warning capability is lacking. 3. Current state of the art of electrostatic detection The electrostatic detection is used as a necessary technical basis for realizing safe production, quality control and process optimization in modern industry, and can convert time domain signals related to an industrial electrostatic field into a frequency domain through electrostatic field Fourier transform, thereby realizing accurate quantification of the performance of a static dissipation material, tracing and positioning of periodic defects in the production process, and effective separation and filtering of noise in monitoring signals. Disclosure of Invention In order to solve the problems, the invention provides a charged particle speed detection method based on electrostatic field Fourier transform, which is used for monitoring abnormal particles in a gas circuit pipeline of a heavy gas turbine and realizing early warning of fault monitoring, and specifically comprises the following steps: a charged particle velocity detection method based on electrostatic field fourier transform, applied to a heavy duty gas turbine, the method comprising: s1, acquiring time domain electrostatic signals of charged particles in a gas circuit pipeline through an electrostatic field sensor; s2, performing anti-aliasing filtering, analog-to-digital conversion and normalization processing on the time domain electrostatic signals of the charged particles to obtain normalized Gaussian pulse signals; S3, sequentially performing zero filling and fast Fourier transform on the normalized Gaussian pulse signal to obtain frequency spectrum data, and calculating a 3dB bandwidth corresponding to the frequency domain signal based on the frequency spectrum data to obtain a frequency domain signal bandwidth; s4, based on the frequency domain signal bandwidth, obtaining an approximate linear relation between the shedding particle speed v and the frequency domain signal bandwidth, and obtaining the charged particle speed. Optionally, the anti-aliasing filtering in S2 includes: An analog low-pass filter is connected between