Adaptive time-frequency joint filtering method for solid rocket motor plume electrostatic data

Abstract

This invention relates to an adaptive time-frequency joint filtering method for electrostatic data from solid rocket motor exhaust. The method includes the following steps: constructing a main channel filtering algorithm to suppress the fundamental combustion oscillation, employing an adaptive IIR notch filter, with the core frequency detected by a real-time FFT algorithm and the bandwidth dynamically adjusted according to the combustion state; constructing an auxiliary channel to suppress harmonics, consisting of multiple parallel notch filters, each corresponding to a harmonic frequency, with the harmonic order adaptively determined based on the fundamental attenuation; addressing the phase distortion problem of traditional IIR filters by using forward-backward filtering to offset phase shift; particle pulse detection is the core of solid rocket motor exhaust electrostatic signal processing, aiming to accurately identify transient pulses (1-5 μs) generated by Al2O3 particle collisions; this invention uses a wavelet domain adaptive algorithm based on multi-scale noise separation, a time-varying threshold mechanism, and the SURE optimization criterion to suppress combustion oscillations and noise in the exhaust electrostatic signal while preserving particle pulse information.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and specifically to an adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plumes. Background Technology

[0002] As the core power unit of aerospace propulsion systems, real-time monitoring of the exhaust plume is crucial for spacecraft safety control and fault diagnosis. The exhaust plume contains rich state information, with the electrostatic signal generated by condensed particle collisions, typically in the μV range and with a frequency band of 0.1–10 kHz, being a typical characteristic. In recent years, with the development of solid rocket motors towards higher energy and lighter weight, the amount of aluminum powder added to the propellant has gradually increased, leading to a significant increase in the concentration of Al2O3 particles in the exhaust plume, a significantly increased probability of particle collisions, and consequently, a stronger electrostatic signal intensity. However, the exhaust plume environment has extremely complex characteristics, such as high temperatures exceeding 2000K, strong oscillating noise, and multiphase flow coupling, posing severe challenges to exhaust plume testing based on electrostatic sensing.

[0003] Active noise cancellation technology is only effective against low-frequency noise and is insufficient for suppressing broadband electromagnetic interference in electrostatic signals. Electrostatic sensors, due to their advantages of non-contact measurement, simple structure, and microsecond-level response, have become the ideal choice for testing exhaust plume conditions.

[0004] Electrostatic signals generated by electrostatic sensors face three core types of interference during acquisition, making it difficult for existing denoising methods to simultaneously suppress noise and preserve features. First, there is the time-varying nature of combustion oscillation noise: combustion instability generates narrowband noise in the 40–500 Hz range, whose fundamental frequency and harmonic components dynamically change with propellant formulation and operating conditions. Second, there is the vulnerability of particle pulse signals: microsecond-level transient pulses generated by Al2O3 particle collisions are easily annihilated by noise. Third, there is spectral aliasing between noise and signal: in the 1–10 kHz frequency band, particle pulses overlap with high-frequency combustion harmonics and electromagnetic jet noise, especially during the combustion of high-aluminum propellants, where the increased particle collision rate exacerbates the aliasing phenomenon.

[0005] This invention proposes an adaptive time-frequency joint denoising algorithm. The main channel uses a zero-phase-delay IIR filter to track the fundamental frequency in real time, while the auxiliary channel dynamically activates harmonic suppression units and adaptively adjusts the quality factor. Combined with Stein unbiased risk estimation and pulse position detection, a low threshold is used to protect edge features in the pulse region, and a high threshold is used to suppress broadband noise in the non-pulse region. The dynamic suppression of combustion oscillation noise and the precise protection of particle pulses resolve the contradiction of "denoising equals distortion" in traditional methods.

[0006] This invention patent addresses the field of electrostatic signal processing for solid rocket motor exhaust plumes, proposing an adaptive time-frequency joint filtering algorithm for electrostatic data from solid rocket motor exhaust plumes, belonging to the field of signal processing.

[0007] The electrostatic signal of a solid rocket motor's exhaust plume is an important basis for studying engine performance, combustion processes, and exhaust plume characteristics. By measuring the electrostatic parameters of the exhaust plume, we can understand the engine's combustion efficiency and the impact of the exhaust plume on the surrounding environment, which is of great significance for engine design and research.

[0008] The effective electrostatic signal from the engine exhaust is weak, with strong signal-noise interference, making it difficult to separate the effective signal. Based on the electrostatic signal from the exhaust obtained by the electrostatic sensor, precise filtering and signal extraction are required to ensure the accuracy and reliability of the measurement results. The filtering process can eliminate the influence of the sensor's own systematic errors, environmental factors, and changes in exhaust characteristics, thereby improving measurement accuracy.

[0009] To detect electrostatic signals from solid rocket motors, the data acquisition card used must have a sampling frequency of at least 100 kHz, which means it needs to process 10⁵ data points per second. This places high demands on the speed and accuracy of the algorithm.

[0010] Current methods for processing electrostatic signals from exhaust plumes are characterized by "hardware first, algorithm later." At the hardware level, an FPGA+ARM architecture is often used to process the original circuit signals according to certain rules.

[0011] Progress in electrostatic signal processing algorithms for plume exhaust has been relatively slow, lacking a dedicated algorithm system for handling the complex noise environment of exhaust exhaust. More novel methods include Empirical Mode Decomposition (EMD), which uses pre-defined modes to determine the mode in which the electrostatic signal belongs and then uses that mode for signal processing; Variational Mode Decomposition (VMD), which uses a pre-defined number of modes to switch between modes for filtering; and Deep Learning methods, which identify characteristics of the electrostatic signal in the exhaust exhaust by labeling the electrostatic signal data and then perform filtering.

[0012] Traditional hardware filtering, such as high-pass and low-pass filtering, is not very effective at preserving signal characteristics. Once the circuit is fixed, the electrostatic signal is also fixed.

[0013] Empirical mode decomposition (EMD), while capable of handling non-stationary signals, is sensitive to high-frequency oscillating mode aliasing and has insufficient classification accuracy under oscillating combustion conditions.

[0014] Variational mode decomposition requires a preset number of modes, making it difficult to adapt to sudden changes in exhaust flame conditions.

[0015] Deep learning methods rely on a large amount of labeled data, but the exhaust flame test samples are scarce, resulting in poor model generalization ability. Summary of the Invention

[0016] Based on the above-mentioned technical problems, this invention proposes an adaptive time-frequency joint denoising method to solve the contradiction of "denoising equals distortion" in traditional methods, and lay the foundation for tail flame velocity testing based on electrostatic signals.

[0017] To address the aforementioned technical problems, one objective of this invention is to provide an adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plumes, the specific steps of which are as follows: S1: To address the unique combustion oscillation noise of solid rocket motor exhaust, a dual-channel variable bandwidth notch filter is constructed. S2: Particle pulse detection accurately identifies transient pulses (1-5μs) generated by Al2O3 particle collisions. S3: Wavelet domain adaptive denoising to reconstruct electrostatic signals of the tail flame; S4: Using the algorithms from steps S1 to S3, construct an adaptive time-frequency joint denoising algorithm.

[0018] Furthermore, the specific steps of S1 are as follows: S11: Main channel filtering algorithm. A main channel filtering algorithm is constructed to suppress the fundamental frequency of combustion oscillation. An adaptive IIR notch filter is used. The core frequency is detected by a real-time FFT algorithm, and the bandwidth is dynamically adjusted according to the combustion state. S12: Auxiliary channel filtering algorithm. The auxiliary channel is used to suppress harmonics and consists of multiple parallel notch filters. Each notch filter corresponds to a harmonic frequency, and the harmonic order is adaptively determined according to the fundamental frequency attenuation. S13: Zero-phase filtering algorithm, which solves the phase distortion problem of traditional IIR filters by canceling phase shift through forward-backward filtering.

[0019] Further, S2 includes the following steps: S21: Calculation of particle collision physical properties, gradient mutation characteristics; S22: Detection algorithm structure, gradient calculation.

[0020] Further, S3 includes the following steps: S31: The filtering algorithm is optimized using the SURE optimization criterion; S32: 5-layer Db4 wavelet packet decomposition.

[0021] Further, S4 includes the following steps: S41: Signal preprocessing, input signal normalization, initialization of notch filter state vector; S42: Dual-channel adaptive notch filter; S43: Pulse detection to eliminate transient pulse noise; S44: SURE transient protected wavelet threshold denoising; S45: Reconstruct the signal using wavelet transform.

[0022] Further, S42 includes the following steps: S421: Fundamental frequency detection, performing real-time spectrum analysis on the input signal; S422: Determines combustion status based on base frequency; S423: Sets the notch filter bandwidth and quality factor; S424: Construct the main channel IIR notch filter; S425: Calculate the harmonic order, where the harmonic amplitude is less than 10% of the fundamental amplitude; S426: Design an auxiliary channel notch filter for each harmonic frequency.

[0023] Further, S43 includes the following steps: S431: Calculate the signal gradient; S432: Perform robust standard deviation estimation; S432: Set a threshold to merge pulse edge regions.

[0024] Further, S44 includes the following steps: S441: Select wavelet basis and number of decomposition levels S442: Perform wavelet decomposition on the notch-filtered signal to obtain approximation coefficients (low frequency) and detail coefficients (high frequency). S443: For each layer detail coefficient, estimate the noise standard deviation of that layer; S444: For each level of detail coefficients, calculate the global threshold according to the SURE criterion or using an improved threshold function; S445: For each level of detail coefficient, based on the pulse detection results, set a lower threshold in the pulse region and a higher threshold in the non-pulse region; S446: Apply a soft threshold or a modified threshold function to handle detail coefficients.

[0025] The above-described one or more technical solutions of the present invention have at least one or more of the following technical effects: This paper proposes an adaptive time-frequency joint filtering algorithm for electrostatic data of solid rocket motor exhaust plumes. This algorithm combines dual-channel adaptive notch filtering and wavelet adaptive denoising, effectively suppressing combustion oscillations and noise while preserving particle pulse information. Notch filtering effectively removes combustion oscillations (fundamental frequency and harmonics) and reduces periodic interference. Wavelet denoising further reduces background noise and improves the signal-to-noise ratio. The pulse detection algorithm accurately identifies particle pulses with high recall and precision. The entire processing flow improves signal quality while maintaining low waveform distortion, laying the foundation for subsequent particle parameter inversion and combustion state diagnosis. Compared with existing technologies, the advantages of this invention are: (1) The dynamic notch filtering technology for combustion oscillation noise is adopted, and the dual-channel parallel notch filtering structure is used to achieve frequency drift tracking through FFT peak detection and harmonic attenuation criteria; (2) Design a pulse-guided time-varying threshold function, a threshold segmentation mechanism inspired by gradient detection, and use an exponential decay threshold in the pulse region to overcome the edge distortion problem caused by traditional wavelet hard thresholding. (3) An online noise parameter estimation module is adopted, which is based on the threshold self-optimization of the SURE criterion. No prior noise statistics are required, and it can adapt to sudden changes in the tail flame environment. Attached Figure Description

[0026] Figure 1 . Block diagram of adaptive time-frequency joint denoising algorithm. Detailed Implementation

[0027] This invention discloses an adaptive time-frequency joint filtering algorithm for electrostatic data from solid rocket motor exhaust. The method includes the following steps: constructing a main channel filtering algorithm to suppress the fundamental frequency of combustion oscillations, employing an adaptive IIR notch filter, with the core frequency detected by a real-time FFT algorithm and the bandwidth dynamically adjusted according to the combustion state; constructing an auxiliary channel to suppress harmonics, consisting of multiple parallel notch filters, each notch filter corresponding to a harmonic frequency, the harmonic order being adaptively determined based on the fundamental frequency attenuation; using forward-backward filtering to offset phase shifts and solve the phase distortion problem of traditional IIR filters; particle pulse detection is the core step in solid rocket motor exhaust electrostatic signal processing, its goal being to accurately identify transient pulses (1-5 μs) generated by Al2O3 particle collisions; using a wavelet domain adaptive algorithm based on multi-scale noise separation, time-varying threshold mechanism, and SURE optimization criterion to suppress combustion oscillations and noise in the exhaust electrostatic signal while retaining particle pulse information.

[0028] This invention focuses on the electrostatic signal of solid rocket motor exhaust plumes, primarily investigating an adaptive time-frequency joint filtering method, including the following steps: Step S1: To address the unique combustion oscillation noise of solid rocket motor exhaust, a dual-channel variable bandwidth notch filter is constructed to solve the three major limitations of traditional notch filters: 1) Spectral drift problem: changes in combustion state cause the noise fundamental frequency to drift in the range of 40-500Hz; 2) Harmonic residue problem: oscillating combustion generates harmonics greater than the 5th order; 3) Phase distortion problem: traditional IIR filters cause particle pulse timing shift.

[0029] Step S2: Particle pulse detection is the core step in the electrostatic signal processing of solid rocket motor exhaust. Its goal is to accurately identify the transient pulses (1-5μs) generated by Al2O3 particle collisions.

[0030] Step S3: Wavelet domain adaptive denoising to reconstruct the electrostatic signal of the tail flame.

[0031] Step S4: Using the algorithms from steps S1 to S3, construct an adaptive time-frequency joint denoising algorithm.

[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments obtained. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0033] This invention discloses an adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plumes, comprising the following steps: Step S1: Construct a dual-channel variable bandwidth notch filter to address the combustion oscillation noise unique to solid rocket motor exhaust flames.

[0034] Step S2: Particle pulse detection accurately identifies transient pulses (1-5μs) generated by Al2O3 particle collisions.

[0035] Step S3: Wavelet domain adaptive denoising to reconstruct the electrostatic signal of the tail flame.

[0036] Step S4: Using the algorithms from steps S1 to S3, construct an adaptive time-frequency joint denoising algorithm.

[0037] Step S1 includes the following steps: Step S11: Main channel filtering algorithm. A main channel filtering algorithm is constructed to suppress the fundamental wave of combustion oscillation. An adaptive IIR notch filter is used. The core frequency is detected by a real-time FFT algorithm, and the bandwidth is dynamically adjusted according to the combustion state.

[0038] Step S12: Auxiliary channel filtering algorithm. The auxiliary channel is used to suppress harmonics and consists of multiple parallel notch filters. Each notch filter corresponds to a harmonic frequency, and the harmonic order is adaptively determined according to the fundamental frequency attenuation.

[0039] Step S13: Zero-phase filtering algorithm, which solves the phase distortion problem of traditional IIR filters by canceling phase shift through forward-backward filtering.

[0040] Step S2 includes the following steps: Step S21: Calculation of particle collision physical properties and gradient mutation characteristics.

[0041] Step S22: Detect the algorithm structure and calculate the gradient.

[0042] Step S3 includes the following steps: Step S31: Optimize the filtering algorithm using the SURE optimization criterion.

[0043] Step S32: 5-layer Db4 wavelet packet decomposition; Step S4 includes the following steps: Step S41: Signal preprocessing, input signal normalization, and initialization of the notch filter state vector.

[0044] Step S42: Dual-channel adaptive notch filtering.

[0045] Step S43: Pulse detection to eliminate transient pulse noise.

[0046] Step S44: SURE transient protection wavelet threshold denoising.

[0047] Step S45: Reconstruct the signal using wavelet transform.

[0048] Step S42 includes the following steps: Step S421: Fundamental frequency detection, real-time spectrum analysis of the input signal.

[0049] Step S422: Determine the combustion state based on the base frequency.

[0050] Step S423: Set the notch filter bandwidth and quality factor.

[0051] Step S424: Construct the main channel IIR notch filter.

[0052] Step S425: Calculate the harmonic order, where the harmonic amplitude is less than 10% of the fundamental amplitude.

[0053] Step S426: Design an auxiliary channel notch filter for each harmonic frequency.

[0054] Step S43 includes the following steps: Step S431: Calculate the signal gradient. Step S432: Perform robust standard deviation estimation.

[0055] Step S432: Set a threshold to merge pulse edge regions.

[0056] Step S44 includes the following steps: Step S441: Select wavelet basis and number of decomposition layers Step S442: Perform wavelet decomposition on the notch-filtered signal to obtain approximation coefficients (low frequency) and detail coefficients (high frequency).

[0057] Step S443: For each layer detail coefficient, estimate the noise standard deviation of that layer.

[0058] Step S444: For each level of detail coefficients, calculate the global threshold according to the SURE criterion or using an improved threshold function.

[0059] Step S445: For each layer of detail coefficients, based on the pulse detection results, set a lower threshold in the pulse region and a higher threshold in the non-pulse region.

[0060] Step S446: Apply a soft threshold or an improved threshold function to process the detail coefficients.

[0061] The electrostatic data of the exhaust plume is collected by a spatially displaced electrostatic sensor and is time-series data with time delay.

[0062] The dynamic notch filtering technology for combustion oscillation noise based on a dual-channel parallel notch filter structure achieves frequency drift tracking through FFT peak detection and harmonic attenuation criteria.

[0063] Among them, the design of the pulse-guided time-varying threshold function, the threshold segmentation mechanism inspired by gradient detection, and the use of an exponentially decaying threshold in the pulse region are all employed.

[0064] Among them, the online noise parameter estimation module is based on the threshold self-optimization of the SURE criterion, which does not require prior noise statistics and can adapt to sudden changes in the exhaust flame environment.

[0065] To achieve accurate filtering and feature extraction of electrostatic signals from solid rocket motor exhaust plumes, an adaptive time-frequency joint filtering algorithm for solid rocket motor exhaust electrostatic data is proposed. The algorithm flowchart is shown below. Figure 1 As shown. Includes the following steps: Step S1: Dual-channel adaptive notch filter. To address the unique combustion oscillation noise of solid rocket motor exhaust, a dual-channel variable bandwidth notch filter is constructed to solve the three major limitations of traditional notch filters: 1) Spectral drift problem: changes in combustion state cause the noise fundamental frequency to drift in the range of 40-500Hz; 2) Harmonic residue problem: oscillating combustion generates harmonics greater than the 5th order; 3) Phase distortion problem: traditional IIR filters cause particle pulse timing shift.

[0066] The filter equation is

[0067] Furthermore, step S1 includes the following steps: Step S11: Main channel filtering algorithm. A main channel filtering algorithm is constructed to suppress the fundamental wave of combustion oscillation. An adaptive IIR notch filter is used. The core frequency is detected by a real-time FFT algorithm, and the bandwidth is dynamically adjusted according to the combustion state.

[0068] Step S12: Auxiliary channel filtering algorithm. The auxiliary channel is used to suppress harmonics and consists of multiple parallel notch filters. Each notch filter corresponds to a harmonic frequency, and the harmonic order is adaptively determined according to the fundamental frequency attenuation.

[0069]

[0070] Step S13: Zero-phase filtering algorithm, which solves the phase distortion problem of traditional IIR filters by canceling phase shift through forward-backward filtering.

[0071]

[0072] Step S2: Particle pulse detection algorithm. Particle pulse detection is the core component of electrostatic signal processing in the exhaust plume of a solid rocket motor. Its goal is to accurately identify transient pulses (1-5 μs) generated by Al2O3 particle collisions. Further, step S2 includes the following steps: Step S21: Calculate the physical properties of particle collisions, and satisfy the gradient mutation property;

[0073] Step S22: Detect algorithm structure and calculate gradients:

[0074] Pulse marker:

[0075] Step S3: Wavelet adaptive denoising, the signal is reconstructed using a wavelet algorithm. Further, step S3 includes the following steps: Step S31: Optimize the filtering algorithm using the SURE optimization criterion.

[0076]

[0077] Step S32: 5-layer Db4 wavelet packet decomposition;

[0078] Step S4: Construct an adaptive time-frequency joint denoising algorithm using the algorithms from steps S1 to S3. Further, step S4 includes the following steps: Step S41: Signal preprocessing, input signal normalization, and initialization of the notch filter state vector.

[0079] Step S42: Dual-channel adaptive notch filtering.

[0080] Step S421: Fundamental frequency detection, real-time spectrum analysis of the input signal.

[0081] Step S422: Determine the combustion state based on the base frequency.

[0082] Step S423: Set the notch filter bandwidth and quality factor.

[0083] Step S424: Construct the main channel IIR notch filter.

[0084] Step S425: Calculate the harmonic order, where the harmonic amplitude is less than 10% of the fundamental amplitude.

[0085] Step S426: Design an auxiliary channel notch filter for each harmonic frequency.

[0086] Step S43: Pulse detection to eliminate transient pulse noise.

[0087] Step S431: Calculate the signal gradient. Step S432: Perform robust standard deviation estimation.

[0088] Step S432: Set a threshold to merge pulse edge regions.

[0089] Step S44: SURE transient protection wavelet threshold denoising.

[0090] Step S441: Select wavelet basis and number of decomposition layers Step S442: Perform wavelet decomposition on the notch-filtered signal to obtain approximation coefficients (low frequency) and detail coefficients (high frequency).

[0091] Step S443: For each layer detail coefficient, estimate the noise standard deviation of that layer.

[0092] Step S444: For each level of detail coefficients, calculate the global threshold according to the SURE criterion or using an improved threshold function.

[0093] Step S445: For each layer of detail coefficients, based on the pulse detection results, set a lower threshold in the pulse region and a higher threshold in the non-pulse region.

[0094] Step S446: Apply a soft threshold or an improved threshold function to process the detail coefficients.

[0095] Step S45: Reconstruct the signal using wavelet transform.

[0096] This invention proposes an adaptive time-frequency joint denoising algorithm. The main channel uses a zero-phase-delay IIR filter to track the fundamental frequency in real time, while the auxiliary channel dynamically activates harmonic suppression units and adaptively adjusts the quality factor. Combining Stein unbiased risk estimation and pulse position detection, a low threshold is used to protect edge features in the pulse region, and a high threshold is used to suppress broadband noise in the non-pulse region. The dynamic suppression of combustion oscillation noise and the precise protection of particle pulses resolve the contradiction of "denoising equals distortion" in traditional methods, laying the foundation for tail flame velocity testing based on electrostatic signals.

[0097] Obviously, those skilled in the art can make various modifications and variations to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations to the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention also intends to include these modifications and variations.

[0098] The filtering algorithm of this invention has been verified through simulation and experimentation, which has demonstrated the effectiveness and reliability of the method.

Claims

1. An adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume, characterized in that, The specific steps are as follows: S1: To address the unique combustion oscillation noise of solid rocket motor exhaust, a dual-channel variable bandwidth notch filter is constructed. S2: Particle pulse detection accurately identifies transient pulses (1-5μs) generated by Al2O3 particle collisions. S3: Wavelet domain adaptive denoising to reconstruct electrostatic signals of the tail flame; S4: Using the algorithms from steps S1 to S3, construct an adaptive time-frequency joint denoising algorithm.

2. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plumes according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Main channel filtering algorithm. A main channel filtering algorithm is constructed to suppress the fundamental wave of combustion oscillation. An adaptive IIR notch filter is used. The core frequency is detected by a real-time FFT algorithm, and the bandwidth is dynamically adjusted according to the combustion state. S12: Auxiliary channel filtering algorithm. The auxiliary channel is used to suppress harmonics and consists of multiple parallel notch filters. Each notch filter corresponds to a harmonic frequency, and the harmonic order is adaptively determined according to the fundamental frequency attenuation. S13: Zero-phase filtering algorithm, which solves the phase distortion problem of traditional IIR filters by canceling phase shift through forward-backward filtering.

3. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plumes according to claim 1, characterized in that, S2 includes the following steps: S21: Calculation of particle collision physical properties, gradient mutation characteristics; S22: Detection algorithm structure, gradient calculation.

4. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 1, characterized in that, S3 includes the following steps: S31: The filtering algorithm is optimized using the SURE optimization criterion; S32: 5-layer Db4 wavelet packet decomposition.

5. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 1, characterized in that, S4 includes the following steps: S41: Signal preprocessing, input signal normalization, initialization of notch filter state vector; S42: Dual-channel adaptive notch filter; S43: Pulse detection to eliminate transient pulse noise; S44: SURE transient protected wavelet threshold denoising; S45: Reconstruct the signal using wavelet transform.

6. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 5, characterized in that, S42 includes the following steps: S421: Fundamental frequency detection, performing real-time spectrum analysis on the input signal; S422: Determines combustion status based on base frequency; S423: Sets the notch filter bandwidth and quality factor; S424: Construct the main channel IIR notch filter; S425: Calculate the harmonic order, where the harmonic amplitude is less than 10% of the fundamental amplitude; S426: Design an auxiliary channel notch filter for each harmonic frequency.

7. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 5, characterized in that, S43 includes the following steps: S431: Calculate the signal gradient; S432: Perform robust standard deviation estimation; S432: Set a threshold to merge pulse edge regions.

8. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 5, characterized in that, S44 includes the following steps: S441: Select wavelet basis and number of decomposition levels S442: Perform wavelet decomposition on the notch-filtered signal to obtain approximation coefficients (low frequency) and detail coefficients (high frequency). S443: For each layer detail coefficient, estimate the noise standard deviation of that layer; S444: For each level of detail coefficients, calculate the global threshold according to the SURE criterion or using an improved threshold function; S445: For each level of detail coefficient, based on the pulse detection results, set a lower threshold in the pulse region and a higher threshold in the non-pulse region; S446: Apply a soft threshold or a modified threshold function to handle detail coefficients.

9. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 1, characterized in that: The electrostatic data of the exhaust plume was collected by a spatially displaced electrostatic sensor, and the electrostatic data of the exhaust plume is time-series data with time delay.

10. The adaptive time-frequency joint filtering method for electrostatic data of solid rocket motor exhaust plume according to claim 1, characterized in that: The dynamic notch filtering technology for combustion oscillation noise based on a dual-channel parallel notch filter structure achieves frequency drift tracking through FFT peak detection and harmonic attenuation criteria.

Images