Method for extracting and processing a specific frequency vibration signal and application

Abstract

The application discloses a specific frequency vibration signal extraction and processing method and application, comprising the following steps: (1) receiving real-time vibration signal data and corresponding phase information; (2) performing bidirectional Butterworth filtering on the vibration signal based on a phase adaptive sliding window to obtain vibration signal of a specific frequency and corresponding phase information; (3) performing area calculation on the filtered vibration signal data based on the phase information by using a zero-mean two-section Simpson integral area calculation method, and calculating the area value of each period of the vibration signal after averaging. The specific frequency vibration signal extraction and processing method ensures that the vibration signal will not produce phase offset in the filtering process, and also avoids the influence of direct current offset or signal drift on the vibration signal in processing.

Description

Technical Field

[0001] This invention relates to the field of vibration monitoring technology, and in particular to a method and application for extracting and processing vibration signals at a specific frequency. Background Technology

[0002] In today's era of rapid technological advancement, the processing and analysis of vibration signals plays a crucial role in numerous fields. Whether it's mechanical equipment operation monitoring, fault diagnosis, structural health monitoring, environmental vibration control, or industrial process optimization, accurately extracting vibration information at specific frequencies has significant practical implications. In mechanical equipment operation monitoring, it can be used for equipment condition analysis and predictive maintenance; in fault diagnosis, it can help quickly locate potential problems in mechanical components; in structural health monitoring, it assists in assessing the stability of buildings or bridges; in environmental vibration control, it is used to identify external interference sources and optimize vibration isolation design; and in industrial process optimization, it can improve equipment operating efficiency and safety.

[0003] With continuous technological advancements, the demands for accuracy and reliability in vibration signal processing are increasing. Traditional signal processing methods often struggle to effectively extract specific frequency signal characteristics when faced with complex vibration environments. In scenarios with mechanical vibration interference or external environmental noise, this interference may mask specific frequency components of the target signal, leading to inaccurate data analysis. For example, in factory environments, machine vibration is superimposed on ambient noise, making it difficult for traditional methods to extract specific frequency signals related to equipment malfunctions.

[0004] The Butterworth filter is a type of electronic filter, also known as a maximum flatness filter. A key characteristic of the Butterworth filter is its maximally flat frequency response curve within the passband, with no ripple, while it gradually decreases to zero in the stopband. As a classic filter type, the Butterworth filter exhibits a flat frequency response characteristic, with slow and monotonic stopband attenuation. Its amplitude-squared function form ensures good signal reproduction capability. Furthermore, higher filter orders generally result in better performance, but also increased computational complexity. Due to its good stability and simple design, the Butterworth filter is widely used in vibration signal processing, mechanical diagnostics, structural monitoring, and other fields, and can be used to remove noise from vibration signals, extract specific frequency components, and achieve signal separation.

[0005] Although Butterworth filters have a flat amplitude-frequency response, they are prone to introducing phase shift during signal filtering. Phase shift can distort the timing information of the signal, affecting the analysis of dynamic signal characteristics.

[0006] Existing methods have high computational complexity when processing specific frequency signals in complex vibration environments, and are difficult to adapt flexibly to diverse application scenarios. Furthermore, they are susceptible to DC offset or signal drift when processing vibration signals. Summary of the Invention

[0007] This invention provides a method for extracting and processing vibration signals at a specific frequency, ensuring that the vibration signal does not undergo phase shift during the filtering process, and also avoiding the influence of DC offset or signal drift when processing the vibration signal.

[0008] It is applicable to a variety of different working conditions and has strong adaptability.

[0009] The technical solution of the present invention is as follows:

[0010] A method for extracting and processing vibration signals at a specific frequency, comprising:

[0011] (1) Receive real-time vibration signal data and corresponding phase information;

[0012] (2) The vibration signal is subjected to bidirectional Butterworth filtering based on a phase adaptive sliding window to obtain a vibration signal of a specific frequency and the corresponding phase information.

[0013] (3) Based on phase information, the zero-mean double-segment Simpson integral area calculation method is used to segment the filtered vibration signal data to calculate the area, and the mean value is obtained to obtain the area value of each cycle of the vibration signal.

[0014] Preferably, step (1) includes:

[0015] (1-1) The real-time received vibration signal data is added to the signal data queue for buffering, with n seconds as the sliding window; the phase information corresponding to the vibration signal data is added to the phase position queue.

[0016] (1-2) New vibration signal data is continuously added to the signal data queue through the enqueue operation, the oldest vibration signal data is removed from the sliding window, and new vibration signal data is added to the sliding window.

[0017] Adjust the position of the sliding window of the phase position queue according to the sliding window of the signal data queue, and align the sliding windows of the signal data queue and the phase position queue.

[0018] Step (2) includes:

[0019] The vibration signal data in the sliding window is filtered forward from the beginning of the signal to the end using a Butterworth filter, and then the filtered vibration signal data is filtered backward from the end of the signal to the beginning.

[0020] While forward filtering can effectively suppress noise in a signal, it can cause a phase shift in the target signal due to the phase response of the filter. Reverse filtering can introduce a phase shift of equal magnitude but opposite direction to that of forward filtering, thereby canceling the phase shift caused by the initial filtering.

[0021] The extraction method of this invention utilizes the excellent characteristics of the Butterworth filter to accurately extract vibration signals within a specific frequency range, reducing the influence of mechanical vibration interference and external noise, and improving the quality and reliability of the vibration signal. By performing bidirectional filtering on the input vibration signal, the signal components at the target frequency can be effectively separated. While ensuring the filtering effect, it is also ensured that the phase of the filtered signal is consistent with the original signal, providing a high-precision data foundation for subsequent analysis and applications.

[0022] Preferably, n is an integer between 3 and 6.

[0023] Step (3) includes:

[0024] (3-1) Based on phase information, the vibration signal is periodically divided and the signal data in each period is extracted;

[0025] (3-2) Perform zero-mean processing on the signal data in each period, including: calculating the mean of the maximum and minimum values ​​in a period, and subtracting the mean from all data in the period to obtain the zero-mean processed signal data;

[0026] (3-3) Perform piecewise numerical integration on the zero-mean-averaged signal data in each period, including: dividing the data of the period into the first half and the second half according to the mean in a period, and then using Simpson's integral method to perform numerical integration on the first half and the second half of the data respectively to obtain the corresponding areas S1 and S2, and calculate S1-S2 to obtain the area value of the period.

[0027] (3-4) Calculate the average of the area values ​​of all periods to obtain the area value of each period.

[0028] Zero-mean normalization of the signal data within each cycle eliminates the DC offset in the vibration signal, ensuring that the area calculation reflects only the fluctuation characteristics of the vibration signal. Piecewise numerical integration of the zero-mean normalized signal data within each cycle can more accurately reflect the fluctuation characteristics of the signal, especially for asymmetric signals.

[0029] After bidirectional filtering, edge effects exist at both ends of the sliding window, but the signal in the middle region is relatively smooth and stable.

[0030] Preferably, in step (3-1), after removing the data from the front and back ends of the sliding window for 0.5-1 seconds, the period is divided.

[0031] The present invention also provides a method for real-time monitoring of gas concentration, comprising:

[0032] (I) Several groups of gases with known components having a gradient concentration distribution are introduced into the gas chamber of a capacitive gas sensor based on infrared thermal effect, and vibration signal data and corresponding phase information data of the gas are collected respectively.

[0033] (II) After the vibration signal data and corresponding phase information data of the gas are extracted and processed by the vibration signal of the specific frequency mentioned above, the periodic area value of the gas is obtained.

[0034] (III) Fit the gas concentration and periodic area value to obtain the concentration prediction model;

[0035] (IV) The gas to be tested with known composition is introduced into the gas chamber of a capacitive gas sensor based on infrared thermal effect, and the vibration signal data and corresponding phase information data of the gas are collected in real time. After the specific frequency vibration signal is extracted and processed, the periodic area value of the gas to be tested is obtained. The periodic area value of the gas to be tested is input into the concentration prediction model to obtain the concentration prediction value of the gas to be tested.

[0036] The working process of a capacitive gas sensor based on infrared thermal effect includes: infrared light source emits infrared radiation, which is modulated by a beam cutter to form infrared radiation that changes in AC signal; the infrared radiation enters the gas chamber, and the gas in the gas chamber absorbs the infrared radiation to generate pressure changes, which in turn drive the moving capacitor plate of the thin-film capacitive monitor to move relative to the fixed plate, generating vibration signal data that changes in AC signal.

[0037] Capacitive gas sensors based on the infrared thermal effect utilize the thermal effect of the gas being measured absorbing infrared radiation, resulting in a pressure change. This pressure change is proportional to the amount of infrared radiation absorbed by the gas, and the absorption amount is related to the gas concentration. This pressure change is converted into a capacitance change in a parallel-plate capacitor. The working process of a capacitive gas sensor based on the infrared thermal effect is as follows: An infrared light source emits infrared radiation, which is modulated by a beam cutter driven by a synchronous motor (the infrared light is periodically blocked and allowed to pass through, thus generating an AC signal), converting it into an AC signal. When the modulated infrared light enters the gas chamber containing the gas being measured, the gas in the chamber absorbs infrared radiation of a specific wavelength, causing a pressure change. This pressure change is proportional to the amount of infrared radiation absorbed by the gas, and the absorption amount is related to the gas concentration. The pressure change drives the moving capacitor plate of the thin-film capacitive detector to move relative to the fixed plate, converting the change in the concentration of the measured component into a capacitance change, and outputting a signal.

[0038] Based on the output signal and phase, the corresponding area can be calculated, thereby determining the concentration of the gas being measured. However, noise from mechanical vibrations and other sources can be superimposed on the alternating current signal representing the change in capacitance, affecting the final measured gas concentration. By employing the specific frequency vibration signal extraction and processing method of this invention to process the gas vibration signal, noise interference can be effectively eliminated.

[0039] Preferably, in step (III), the least squares method is used to fit the gas concentration and periodic area value.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] (1) Sliding window algorithms typically process continuous signal data. The phase-adaptive sliding window of this invention further incorporates a phase position queue to ensure that the data processed by each sliding window is aligned with the corresponding phase position. This mechanism enables the method of this invention to accurately calculate the area of ​​the signal within different phase periods, which is of great significance, especially when processing periodic signals.

[0042] (2) The bidirectional filtering process of the present invention ensures that the signal does not produce phase shift during the filtering process. Usually, filters cause the signal phase to lag or advance, while the method of the present invention cancels out the phase shift by filtering before and after the other two times, keeping the phase of the signal consistent with the original signal.

[0043] (3) In the area calculation, the method of the present invention extracts the maximum and minimum values ​​of each cycle and adjusts the data to zero mean. This effectively eliminates DC offset or overall drift in the signal, ensuring that the calculated area is related to the relative fluctuation amplitude of the signal, rather than being affected by the absolute value of the signal, thereby improving the accuracy of the calculation results. Attached Figure Description

[0044] Figure 1 This is a schematic flowchart of the method of the present invention;

[0045] Figure 2 The image shows the vibration signal after bidirectional filtering, where the vertical axis represents the filtered signal intensity and the horizontal axis represents the time (s) corresponding to the data within the sliding window.

[0046] Figure 3 The image shows the vibration signal within a single cycle after bidirectional filtering (DC component elimination has been performed). The yellow line represents the waveform of the original signal data within this cycle, and the blue line represents the waveform of the signal data after bidirectional filtering.

[0047] Figure 4This is a graph showing the relationship between the average area of ​​each cycle after processing by this invention and the gas concentration. The horizontal axis represents the gas concentration (ppm), and the vertical axis represents the area value corresponding to different concentrations of the same gas. The yellow line and the blue line correspond to the data with vibration interference and the data without vibration, respectively. Detailed Implementation

[0048] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the embodiments described below are intended to facilitate the understanding of the present invention and do not limit it in any way.

[0049] The infrared gas analyzer's light source emits infrared radiation, which is modulated by a beam cutter driven by a synchronous motor and converted into an AC signal. When the infrared light enters the receiving chamber, the gas pressure inside changes due to the infrared radiation, causing the moving capacitor plate of the thin-film capacitor detector to move relative to the stationary plate. This converts the change in the concentration of the analyte into a change in capacitance, which is then output as a signal. Based on the output signal and phase, the corresponding area is calculated, and the gas concentration is predicted.

[0050] like Figure 1 As shown, the method for extracting a specific frequency vibration signal according to the present invention includes the following steps:

[0051] (1) Data Acquisition and Input

[0052] The light source emits infrared radiation, which is modulated by a beam cutter driven by a synchronous motor and converted into an AC signal. When the infrared light enters the receiving gas chamber, the gas pressure inside the chamber changes due to the infrared radiation, which in turn drives the moving capacitor plate of the thin-film capacitor detector to move relative to the stationary plate, converting the change in the concentration of the measured component into a change in capacitance, which is then output in real time.

[0053] To facilitate use, the frequency signal extraction algorithm model is encapsulated as an API. Its input parameters include data collected within one second and its corresponding phase information. Therefore, the data output from the acquisition unit needs to be buffered. After the data has accumulated for one second, the API is called to input the buffered data into the algorithm model for processing.

[0054] (2) Data preprocessing

[0055] Because the amount of data per second is relatively small, the input data within that second needs to be cached in a queue to ensure the accuracy of subsequent processing. The queue has a maximum capacity of 4 seconds of data. When the queue is full, the oldest data from the previous second is removed before new data is added, ensuring that the queue always retains the latest 4 seconds of data. To reduce redundant computation and optimize performance, this process employs a phase-adaptive sliding window queue algorithm, thereby achieving more efficient data processing.

[0056] The sliding window algorithm is a commonly used algorithm for processing continuous data streams, capable of efficiently calculating statistical information or features within a fixed-size window. A sliding window algorithm implemented using a queue can improve performance by reducing unnecessary redundant calculations while maintaining the accuracy of the results.

[0057] In the algorithm of this project, queues are used to manage data buffers and phase positions to efficiently process data within the sliding window for filtering and area calculation. The initialization and updating of the queues play a crucial role in the algorithm's execution efficiency and correctness.

[0058] ① Queue initialization

[0059] At the start of the algorithm, two queues are initialized:

[0060] (a) Data queue: Used to store input signal data buff_data to implement the sliding window function.

[0061] (b) Phase position queue (phase_position): Used to store the phase position of the signal (buff_phase_position), providing key position information when calculating the area.

[0062] ② Queue Update

[0063] Queue updates mainly involve the following two parts:

[0064] (a) Data queue updates: New signal data is continuously added to the queue through enqueue operations. The queue update mechanism ensures that the window always maintains a fixed size of data. Each time the queue is full, the oldest data is removed (i.e., a portion of the sliding window), and the new data is added to the end of the queue.

[0065] (b) Phase position queue update: The phase position queue update operation is similar to that of the data queue, but it involves adjusting the phase positions. To ensure that the data corresponding to the phase positions are correctly aligned within the sliding window, the algorithm adjusts the storage position of the phases as the window slides.

[0066] (3) The filter processes the signal by removing interference frequencies.

[0067] During the data acquisition phase, the beam cutter driven by the synchronous motor modulates the infrared radiation, resulting in an AC signal output from the data acquisition unit with the same frequency as the synchronous motor. Therefore, when filtering the acquired data, it is necessary to extract the signal component with the same frequency as the synchronous motor. Simultaneously, since the corresponding area needs to be calculated based on the phase information, a bidirectional phase-shift-free Butterworth filtering algorithm is used to avoid introducing phase shift during the filtering process. This algorithm preserves the original phase characteristics of the signal during filtering, thus ensuring the accuracy of the area calculation and preventing errors caused by phase shift.

[0068] For the target data, a Butterworth filter is applied in both forward and reverse directions. While the first filter effectively suppresses noise in the signal, its phase response causes a phase shift in the target signal, which may adversely affect subsequent area calculations based on phase information. To eliminate this phase shift, the signal after the first filter is further filtered in reverse, i.e., the same filter is used to re-filter the signal from the end to the beginning. In this way, the second filter introduces a phase shift of equal magnitude but opposite direction to the first filter, thus canceling out the phase shift from the first filter. This ensures that the phase of the filtered signal is consistent with the original signal while maintaining the filtering effect.

[0069] like Figure 2 The image shows the signal after bidirectional filtering, with corresponding phase information marked. It can be observed that although edge effects exist at both ends of the window, the signal in the middle region is relatively smooth and stable. Therefore, in subsequent steps, we need to select the relatively stable 1.0s-3.0s portion in the middle for area calculation.

[0070] This invention employs bidirectional filtering to process signals. Bidirectional filtering ensures that no phase shift occurs during the filtering process, making it applicable to various working conditions and highly adaptable. For example... Figure 3 As shown, within a single period, the original signal and the filtered signal maintain the same phase without any shift. Although the signal strength is weakened during filtering, this loss ratio is constant for the target signal at the same frequency, and therefore does not affect the subsequent gas concentration prediction based on area calculation.

[0071] (4) Calculate the area.

[0072] Based on phase information, the zero-mean bi-segment Simpson integral area calculation method is used to segment the filtered data to calculate the area, and finally the average area of ​​each segment is used as the final result for subsequent gas concentration prediction.

[0073] The zero-mean, two-piece Simpson integral area calculation method is primarily designed for periodic signals. It extracts signal data within a period and performs numerical integration to calculate the area of ​​each period. This process includes signal preprocessing, piecewise processing, and numerical integration to ensure that the calculation results accurately reflect the overall characteristics of the signal within each period. The detailed processing steps of this method are as follows:

[0074] (a) Periodization and Data Extraction: Based on phase information, the signal is divided into multiple periods, and the signal data within each period is extracted. This ensures that the data of each period can be processed separately, preparing for subsequent area calculations.

[0075] (b) Zero-mean normalization: The signal for each period is zero-mean normalized by finding the midpoint between the maximum and minimum values ​​and subtracting this midpoint from all data. This step eliminates DC offset in the signal, ensuring that the area calculation reflects only the signal's fluctuation characteristics.

[0076] (c) Piecewise numerical integration: After zero-mean normalization, the midpoint of the data within each period lies on the horizontal axis of the coordinate system. This point divides the data of that period into a first half and a second half. Simpson's integral is then used to numerically integrate these two parts to obtain the corresponding areas A and B. Finally, the final area value for that period is obtained through AB. Piecewise integration can more accurately reflect the fluctuation characteristics of a signal, especially for asymmetric signals.

[0077] (d) Calculate the average area of ​​each period.

[0078] Figure 4 Taking CO as an example, the relationship between the average calculated area of ​​each cycle and the gas concentration is shown. It can be seen that, regardless of whether the data includes vibration interference or not, the calculation results show a good linear relationship with the gas concentration. This indicates that the algorithm of this invention has strong anti-noise interference capability, strong adaptability, and can be applied to different working conditions.

[0079] (5) Fitting the linear model. The area calculated in the previous step is linearly related to the gas concentration for the same gas. Therefore, during the training phase, it is necessary to use different concentrations and their corresponding areas to fit the linear model. The fitting method used here is the least squares method, so that the model can calculate the corresponding gas concentration based on the area.

[0080] (6) Output results. In the practical application stage, after calculating the area corresponding to the gas to be tested through steps (1)-(4), the area is input into the linear model fitted in step (5) to calculate the corresponding gas concentration. The output data is the concentration of the gas to be tested.

[0081] The embodiments described above provide a detailed explanation of the technical solutions and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for extracting and processing vibration signals at a specific frequency, characterized in that, include: (1) Receive real-time vibration signal data and corresponding phase information; (2) The vibration signal is subjected to bidirectional Butterworth filtering based on a phase adaptive sliding window to obtain a vibration signal of a specific frequency and the corresponding phase information. (3) Based on phase information, the zero-mean double-segment Simpson integral area calculation method is used to segment the filtered vibration signal data to calculate the area, and the mean value is obtained to obtain the area value of each cycle of the vibration signal.

2. The method for extracting and processing vibration signals of a specific frequency according to claim 1, characterized in that, Step (1) includes: (1-1) The real-time received vibration signal data is added to the signal data queue for buffering, with n seconds as the sliding window; the phase information corresponding to the vibration signal data is added to the phase position queue. (1-2) New vibration signal data is continuously added to the signal data queue through the enqueue operation, the oldest vibration signal data is removed from the sliding window, and new vibration signal data is added to the sliding window. Adjust the position of the sliding window of the phase position queue according to the sliding window of the signal data queue, and align the sliding windows of the signal data queue and the phase position queue.

3. The method for extracting and processing vibration signals of a specific frequency according to claim 2, characterized in that, The n mentioned is an integer between 3 and 6.

4. The method for extracting and processing vibration signals of a specific frequency according to claim 2, characterized in that, Step (2) includes: The vibration signal data in the sliding window is filtered forward from the beginning of the signal to the end using a Butterworth filter, and then the filtered vibration signal data is filtered backward from the end of the signal to the beginning.

5. The method for extracting and processing vibration signals of a specific frequency according to claim 1, characterized in that, Step (3) includes: (3-1) Based on phase information, the vibration signal is periodically divided and the signal data in each period is extracted; (3-2) Perform zero-mean processing on the signal data in each period, including: calculating the mean of the maximum and minimum values ​​in a period, and subtracting the mean from all data in the period to obtain the zero-mean processed signal data; (3-3) Perform piecewise numerical integration on the zero-mean-averaged signal data in each period, including: dividing the data of the period into the first half and the second half according to the mean in a period, and then using Simpson's integral method to perform numerical integration on the first half and the second half of the data respectively to obtain the corresponding areas S1 and S2, and calculate S1-S2 to obtain the area value of the period. (3-4) Calculate the average of the area values ​​of all periods to obtain the area value of each period.

6. The method for extracting and processing vibration signals of a specific frequency according to claim 5, characterized in that, In step (3-1), after removing the data from the front and back ends of the sliding window for 0.5-1 seconds, the period is divided.

7. A method for real-time monitoring of gas concentration, characterized in that, include: (I) Several groups of gases with known components having a gradient concentration distribution are introduced into the gas chamber of a capacitive gas sensor based on infrared thermal effect, and vibration signal data and corresponding phase information data of the gas are collected respectively. (II) After the vibration signal data of the gas is processed by the extraction and processing method of the specific frequency vibration signal as described in any one of claims 1-6, the periodic area value of the gas is obtained. (III) Fit the gas concentration and periodic area value to obtain the concentration prediction model; (IV) The gas to be tested with known composition is introduced into the gas chamber of a capacitive gas sensor based on infrared thermal effect, and the vibration signal data and corresponding phase information data of the gas are collected in real time; after being processed by the extraction and processing method of the specific frequency vibration signal as described in any one of claims 1-6, the periodic area value corresponding to the gas to be tested is obtained; the periodic area value of the gas to be tested is input into the concentration prediction model to obtain the concentration prediction value of the gas to be tested.

8. The method for real-time monitoring of gas concentration according to claim 7, characterized in that, The working process of a capacitive gas sensor based on infrared thermal effect includes: infrared light source emits infrared radiation, which is modulated by a beam cutter to form infrared radiation that changes in AC signal; the infrared radiation enters the gas chamber, and the gas in the gas chamber absorbs the infrared radiation to generate pressure changes, which in turn drive the moving capacitor plate of the thin-film capacitive monitor to move relative to the fixed plate, generating vibration signal data that changes in AC signal.

9. The method for real-time monitoring of gas concentration according to claim 7, characterized in that, In step (III), the least squares method is used to fit the gas concentration and periodic area value.

Images