Real-time anomaly detection method based on time-frequency domain in intelligent oscilloscope

Abstract

The application discloses a real-time abnormality detection method based on time-frequency domain in an intelligent oscilloscope, wherein an input signal is sampled through a collection system to obtain a digital sampling signal and is sent to a time-frequency domain abnormality detection module in FPGA; in the time-frequency domain abnormality detection module, abnormality detection based on spectrum distribution, channel division of data in a digital domain, adaptive down-sampling, short-time Fourier transform with variable window length and abnormality detection based on time-frequency matrix representation are used to complete abnormality detection of the signal; after the detection, the collection system stores the signal judged as abnormal, and finally displays on a host computer.

Description

Technical Field

[0001] This invention belongs to the field of anomaly detection technology in data acquisition systems, and more specifically, relates to a real-time anomaly detection method based on the time and frequency domain in an intelligent oscilloscope. Background Technology

[0002] With the increasing complexity of electrical signals and the rapid development of 5G and 6G technologies, the requirements for observing high-speed electrical signals and for acquisition systems are becoming increasingly demanding. Acquisition systems are also becoming more complex, placing higher demands on their ability to capture specific abnormal signals. While general triggering functions can capture data waveforms, they can only capture anomalous signals with simple and known shapes. When dealing with complex signals, especially those with regular frequency domain patterns but complex time domain shapes, their ability to capture anomalous signals is insufficient. Due to the diversity and complexity of anomalous signals, it is difficult for users to capture them by setting existing triggering conditions, even though anomalous signals often contain the information the user wants. Therefore, the necessity of anomaly detection functionality for high-speed acquisition systems is self-evident.

[0003] In most radar and communication signal fields, signals differ from simple sine and square waves; many are modulated signals. Anomaly detection is often difficult in the time or frequency domains alone, even though these signals exhibit clear patterns in their time-frequency domain characteristics. Traditional acquisition systems lack real-time anomaly detection capabilities in the time and frequency domains, hindering their ability to effectively capture anomalies in such signals. For example, analyzing frequency-modulated (FM) signals, whose frequency changes over time, requires only basic triggering functions to capture and display waveforms, making anomaly detection challenging. Anomaly detection in both the time and frequency domains is insufficient. However, by performing a Short-Time Fourier Transform (STFT) on the signal and analyzing its time-frequency domain characteristics, the presence of anomalies can be determined.

[0004] For a broadband acquisition system, calculating and analyzing the full bandwidth signal results in extremely high computational load, high resource consumption, low throughput, and excessively long processing time. This leads to prolonged periods where abnormal signals cannot be detected, resulting in poor real-time performance and failing to meet the anomaly detection requirements of high-speed acquisition systems. Therefore, a time-frequency domain anomaly detection method is needed that can extract the time-frequency domain features of the signal in real time. By analyzing these features, anomalies during acquisition can be captured, thereby meeting the real-time requirements of high-speed acquisition systems. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a real-time anomaly detection method based on the time and frequency domain in an intelligent oscilloscope. This method determines whether a signal is abnormal by comparing time and frequency domain features and displays the captured abnormal signals in real time.

[0006] To achieve the above-mentioned objectives, the present invention provides a real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope, characterized by comprising the following steps:

[0007] (1) Power on and initialize the digital oscilloscope, and then connect the input signal;

[0008] (2) In normal detection mode, the input signal is sampled by the acquisition system in the digital oscilloscope to obtain the digital sampled signal;

[0009] (3) Input the digital sampling signals into the FFT unit and the WOLA structure respectively;

[0010] (3.1) Perform FFT operation on the digital sampled signal in the FFT unit to obtain the spectral distribution characteristics X. norm ={X norm [0],X norm [1],…,X norm [n],…,X norm [N-1]}, where N is the number of points in the FFT operation;

[0011] Errata distribution characteristics X norm Each spectral component X in norm [n], marking the spectral component X norm Spectral components with [n] greater than the threshold δ are valid spectral components;

[0012] (3.2) In the WOLA structure, the digital sampled signal is channelized digitally filtered by the channelized digital filter in each channel, and then compared with the signal at a frequency of... The digital single-tone signal is digitally mixed, and finally digitally anti-mirror filtered by a digital anti-mirror filter, thus obtaining a low-frequency digital signal in each channel; where i represents the channel number, i = 0, 1, ..., m-1, m is the number of channels, and f s It is the real-time sampling rate of the acquisition system;

[0013] Finally, the time-domain data of each low-frequency digital signal is adaptively decimated by a factor of D to obtain the output signal of each channel.

[0014] (4) Determine the channel whose effective spectral components are located in the WOLA structure;

[0015] Errata distribution characteristics X norm Each effective spectral component is denoted as X. norm [k], then the corresponding frequency component These are the effective components, where k is the number of the effective spectral component, and f sIt is the real-time sampling rate of the acquisition system. Then the i-th channel is the effective spectral component X. norm [k] is located in the channel;

[0016] (5) Based on the channel determined in step (4), select the output signal of the corresponding channel i in the WOLA structure, perform adaptive short-time Fourier transform, then perform downsampling processing, and finally convert the downsampled data into a time-frequency matrix P. norm [i];

[0017] (6) The spectral distribution features X obtained under normal detection mode norm and time-frequency matrix P norm [i] is used as a standard spectrum distribution template and time-frequency matrix template and deployed to the FPGA;

[0018] (7) Set the digital oscilloscope to anomaly detection mode, and then repeat steps (2) to (5) to obtain the spectral distribution characteristics X of the input signal in anomaly detection mode. in ={X in [0],X in [1],…,X in [n],…,X in [N-1]} and the time-frequency matrix P in [i];

[0019] (8) Judgment of abnormal input signals;

[0020] (8.1) Errata Spectral Distribution Characteristics X norm and X in For each spectral component in X, if X norm and X in If there exists a set of spectral components that satisfy the relation: |X in [n]-X norm [n]|>X set X set If a threshold is set, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); otherwise, proceed to step (8.2).

[0021] (8.2) Traversing the time-frequency matrix P norm [i] and P in [i], compare whether the channel numbers i are the same one-to-one. If the channel numbers are different, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); if the channel numbers are the same one-to-one, the starting point search and loop comparison method are used to compare all time-frequency matrices P with the same number. norm [i] and P inIf the comparison results of elements at a certain position are different, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); if the comparison results of the time-frequency matrices corresponding to the channel number are all the same, the input signal is determined to be a normal signal, and then proceed to step (8.3).

[0022] (8.3) Input the determined signal to the host computer for real-time display.

[0023] The objective of this invention is achieved as follows:

[0024] This invention discloses a real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope. First, the input signal is sampled by an acquisition system to obtain a digital sampled signal, which is then sent to the time-frequency domain anomaly detection module in the FPGA. In the time-frequency domain anomaly detection module, anomaly detection of the signal is completed based on anomaly detection of spectral distribution, channel partitioning of data in the digital domain, adaptive downsampling, short-time Fourier transform with variable window length, and anomaly detection based on time-frequency matrix representation. After the detection is completed, the acquisition system stores the signals that are determined to be abnormal and finally displays them on the host computer.

[0025] The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope of the present invention also has the following beneficial effects:

[0026] (1) An adaptive downsampling technique based on the WOLA structure is adopted. This structure can divide the full bandwidth signal into multiple narrow bandwidth digital channels in the digital domain, and then move the sub-signals of each channel to the low frequency band through digital domain signal processing. At the same time, by real-time adaptive multi-rate extraction of the signal time-domain waveform in the digital domain, the equivalent sampling rate of the signal is reduced, thereby greatly reducing the computational load of subsequent real-time anomaly detection, thus realizing real-time calculation and anomaly detection of the signal time-frequency characteristics;

[0027] (2) In the WOLA structure, by adaptively extracting time-domain data from the low-frequency signal after mixing, the equivalent sampling rate of the signal is adaptively reduced, the computational load of the subsequent short-time Fourier transform is reduced, the real-time performance of the system is improved, and the efficiency of the system in capturing anomalies is increased.

[0028] (3) When users need to capture an abnormal signal, they often need to make complex settings to the acquisition system. The detection method of the present invention can automatically detect abnormalities in the input signal without requiring users to be familiar with the various functions of the acquisition system, thereby improving the oscilloscope's ability to automatically detect abnormal signals.

[0029] (4) Some signals are difficult to detect by time domain or frequency domain anomaly detection methods alone. The detection method of the present invention can solve this problem and greatly improve the ability of the acquisition system to accurately identify and capture abnormal signals. Attached Figure Description

[0030] Figure 1 This is a flowchart of a real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to the present invention;

[0031] Figure 2 This is a schematic diagram of the input signal;

[0032] Figure 3 This is a partial screenshot of the sampled signal after FFT transformation under normal detection mode;

[0033] Figure 4 This is the time-frequency diagram of the signal obtained from channel 4 under normal detection mode;

[0034] Figure 5 This is the time-frequency diagram of the signal obtained from channel 5 under normal detection mode;

[0035] Figure 6 This is a partial screenshot of the sampled signal after FFT transformation in anomaly detection mode;

[0036] Figure 7 This is the time-frequency diagram of the signal obtained from channel 4 under anomaly detection mode;

[0037] Figure 8 This is the time-frequency diagram of the signal obtained from channel 5 under anomaly detection mode;

[0038] Figure 9 This is a diagram illustrating the starting point search and the loop comparison. Detailed Implementation

[0039] The specific embodiments of the present invention will now be described with reference to the accompanying drawings to enable those skilled in the art to better understand the invention. It should be particularly noted that in the following description, detailed descriptions of known functions and designs that might obscure the main content of the invention will be omitted here.

[0040] Example

[0041] In this embodiment, as Figure 1 As shown, the present invention discloses a real-time anomaly detection method based on the time-frequency domain in a smart oscilloscope, comprising the following steps:

[0042] S1. Power on and initialize the digital oscilloscope, then connect the input signal. The input signal can be as follows: Figure 2 The signal shown; in this embodiment, the digital oscilloscope operates in normal detection mode by default during power-on initialization;

[0043] S2. In normal detection mode, the input signal is sampled by the acquisition system in the digital oscilloscope to obtain a digital sampled signal. In this embodiment, it is necessary to collect sampling data for a sufficient duration, at least including the complete cycle of the input signal.

[0044] S3. Input the digital sampling signals to the FFT unit and the WOLA structure respectively;

[0045] S3.1 Perform FFT operation on the digital sampled signal in the FFT unit to obtain the spectral distribution characteristics X. norm ={X norm [0],X norm [1],…,X norm [n],…,X norm [N-1]}, where N is the number of points in the FFT operation. Here, N is 4096, so a 4096-point FFT is performed. The FFT result is as follows: Figure 3 As shown, only a portion of the signal contained in the FFT is extracted;

[0046] Errata distribution characteristics X norm Each spectral component X in norm [n], marking the spectral component X norm [n] Spectral components greater than the threshold δ = 150 are considered valid spectral components; in this example, the valid frequency range of the signal in the FFT corresponds to the number of FFT points from 103 to 143, and the system sampling rate is 20 GHz, so its valid frequency components are...

[0047] S3.2 The WOLA structure divides the full bandwidth of the acquisition system into m = 160 narrow bandwidth channels in the digital domain. The WOLA structure is a 1-in-160-out structure. The input is the full bandwidth digital sampling signal, and the corresponding output is 160 narrow bandwidth sampling data.

[0048] In this embodiment, based on the real-time sampling rate f of the acquisition system s =20G and digital signal processing frequency f in the digital domain fpga =250MHz to determine the division of the full bandwidth into m channels, the number of channels m satisfies:

[0049]

[0050] The bandwidth of each channel is:

[0051] The i-th channel M i The frequency range is:

[0052] M i={125MHz*i, 125MHz*(i+1)},0≤i≤m-1

[0053] In the WOLA architecture, the digital sampled signal is channelized digitally filtered by a channelized digital filter in each channel, ensuring that the full bandwidth of the input sampled data is evenly distributed across m = 160 channels. In this embodiment, each channel has a corresponding channelized digital filter with a bandwidth of 125MHz. Each channel-specific channelized digital filter is a bandpass digital filter, and its passband range is the frequency range M of that channel. i ={125MHz*i, 125MHz*(i+1)},0≤i≤m-1;

[0054] In each channel, the channelized digital filtering result is digitally mixed with a digital single-tone signal with a frequency of 125MHz*i. Then, a digital anti-mirror filter is used to filter out the high-frequency components after mixing, thereby shifting the signal frequency to a low-frequency band, achieving spectrum shifting. Finally, a low-frequency digital signal is obtained in each channel. Here, i represents the channel number, i = 0, 1, ..., m-1, and m is the number of channels. In this embodiment, the anti-mirror filter is a low-pass filter with a cutoff frequency of 125MHz. The digital mixing and digital anti-mirror filtering operations are completed in the FPGA. After digital anti-mirror filtering, the signal of the channel is shifted to the frequency band 0-125MHz. Therefore, after spectrum shifting of the signals of m channels, m low-frequency signals are obtained.

[0055] Finally, D-rate adaptive decimation of time-domain data is performed on each low-frequency digital signal, with the decimation rate and channel bandwidth as the factors. and the real-time sampling rate f of the acquisition system s Regarding the extraction multiplier D, it satisfies the following relationship: In this embodiment, the D value is set to 80, so that the data after spectrum shifting is adaptively extracted to obtain the final output of m=160 channels in the WOLA structure;

[0056] S4. Determine the channel whose effective spectral components are located in the WOLA structure;

[0057] Errata distribution characteristics X norm Each effective spectral component is denoted as X. norm [k], then the corresponding frequency component These are the effective components, where k is the number of the effective spectral component, and f s It is the real-time sampling rate of the acquisition system. Then the i-th channel is the effective spectral component X. norm [k] is located in the channel; in this example, as shown in S3.1, the signal is located in the two channels i=4 and i=5;

[0058] S5. Based on the channels determined in step S4, select the output signals of the two channels corresponding to i=4 and i=5 in the WOLA structure, perform adaptive short-time Fourier transform, then perform downsampling processing, and finally convert the downsampled data into a time-frequency matrix P. norm [i]; In this embodiment, the time-frequency matrix P of the normal signal is obtained from the 4th and 5th channels. norm [4] and P norm [5] respectively as Figure 4 and Figure 5 As shown;

[0059] S6. The spectral distribution characteristics X obtained under normal detection mode norm and time-frequency matrix P norm [i] is used as a standard spectrum distribution template and time-frequency matrix template and deployed to the FPGA;

[0060] S7. Set the digital oscilloscope to anomaly detection mode, and then repeat steps (2) to (5) to obtain the spectral distribution characteristics X of the input signal in anomaly detection mode. in ={X in [0],X in [1],…,X in [n],…,X in [N-1]} and the time-frequency matrix P in [i];

[0061] In this embodiment, the spectral distribution characteristics X of the input signal in anomaly detection mode in like Figure 6 As shown, the time-frequency matrix P of the input signal is obtained from channels 4 and 5. in [4] and P in [5] respectively as Figure 7 and Figure 8 As shown;

[0062] S8. Abnormal judgment of input signal;

[0063] S8.1, Errata Spectral Distribution Characteristics X norm and X in For each spectral component in X, if X norm and X in If there exists a set of spectral components that satisfy the relation: |X in [n]-X norm [n]|>X set X set=20 is the threshold value. If the input signal is set to be abnormal, then proceed to step S8.3; otherwise, proceed to step S8.2. In the example above, this step determines that the signal is not abnormal and proceeds to step S8.2 for judgment.

[0064] S8.2, Traversing the time-frequency matrix P norm [i] and P in [i], compare whether the channel numbers i correspond one-to-one. If the channel numbers are different, determine that the input signal is an abnormal signal, and then proceed to step S8.3; In this embodiment, the two channel numbers in the WOLA structure in both the normal detection mode and the abnormal detection mode are 4 and 5, so the channel numbers are the same.

[0065] If the channel numbers are identical in a one-to-one correspondence, then a starting point search and cyclic comparison method are used to compare all time-frequency matrices P with the same number. norm [i] and P in If the comparison results of elements at a certain position are different for each element in [i], the input signal is determined to be an abnormal signal, and then proceed to step S8.3; if the comparison results of the time-frequency matrices corresponding to the channel numbers are all the same, the input signal is determined to be a normal signal, and then proceed to step S8.3.

[0066] Wherein, the time-frequency matrix P norm [i] and P in [i] The comparison method uses starting point search and loop comparison as follows: Figure 9 As shown, assume P norm [i] and P in The size of [i] is J×T. We first compare P. norm [4] and P in [4], then compare P norm [5] and P in [5] If the comparison results of the two sets of time-frequency matrices are the same, the input signal is determined to be a normal signal; otherwise, the signal is determined to be abnormal.

[0067] S8.3 Input the determined signal to the host computer for real-time display.

[0068] We take P norm [4] and P in [4] Taking this as an example, the specific comparison process is as follows:

[0069] 1) Time-frequency matrix P norm [4] and P in [4] starting point search;

[0070] The time-frequency matrix P in [i] is the first column P in [i]1 and the time-frequency matrix Pnorm [4] is compared with each column to find the time-frequency matrix P. in All elements in the first column of [4] satisfy the condition. Starting from the column t0, proceed to step 2); otherwise, determine that the input signal is an abnormal signal and end the comparison; where P set To set the threshold, the value is 20, P in [4] 1,v Represents the time-frequency matrix P in [4] The vth element in the first column, in this embodiment, can be found by comparison to be P in The first column in [4] is the starting point;

[0071] 2) Perform cyclical comparisons based on the starting point;

[0072] P in The first column of [i] and P norm Compare with column t0 of [4], P in The second column in [i] and P norm [4] The comparison is performed on column t0+1, and so on, until the comparison reaches column P. norm After the last column of [i], return to P. norm [4] Continue comparing the first column until the time-frequency matrix P norm [4] and P in [4] All columns have been compared. During the comparison process, if any element in any column satisfies |P in [4] 1,v -P norm [4] t,v |>P set At that time, it was believed that P in [4] and P norm [4] Different, in this example P in [4] and P norm The second half of [4] is different, thus it is determined that the signal is an abnormal signal.

[0073] Although the illustrative specific embodiments of the present invention have been described above to enable those skilled in the art to understand the invention, it should be understood that the invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the invention as defined and determined by the appended claims, and all inventions utilizing the concept of the present invention are protected.

Claims

1. A real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope, characterized in that, Includes the following steps: (1) Power on and initialize the digital oscilloscope, and then connect the input signal; (2) In normal detection mode, the input signal is sampled by the acquisition system in the digital oscilloscope to obtain the digital sampled signal; (3) Input the digital sampling signals into the FFT unit and the WOLA structure respectively; (3.1) Perform FFT operation on the digital sampled signal in the FFT unit to obtain the spectral distribution characteristics X. norm ={X norm [0],X norm [1],…,X norm [n],…,X norm [N-1]}, where N is the number of points in the FFT operation; Errata distribution characteristics X norm Each spectral component X in norm [n], marking the spectral component X norm Spectral components with [n] greater than the threshold δ are valid spectral components; (3.2) In the WOLA structure, the digital sampled signal is channelized digitally filtered by the channelized digital filter in each channel, and then compared with the signal at a frequency of... The digital single-tone signal is digitally mixed, and finally digitally anti-mirror filtered by a digital anti-mirror filter, thus obtaining a low-frequency digital signal in each channel; where i represents the channel number, i = 0, 1, ..., m-1, m is the number of channels, and f s It is the real-time sampling rate of the acquisition system; Finally, the time-domain data of each low-frequency digital signal is adaptively decimated by a factor of D to obtain the output signal of each channel. (4) Determine the channel whose effective spectral components are located in the WOLA structure; Errata distribution characteristics X norm Each effective spectral component is denoted as X. norm [k], then the corresponding frequency component These are the effective components, where k is the number of the effective spectral component, and f s It is the real-time sampling rate of the acquisition system. Then the i-th channel is the effective spectral component X. norm [k] is located in the channel; (5) Based on the channel determined in step (4), select the output signal of the corresponding channel i in the WOLA structure, perform adaptive short-time Fourier transform, then perform downsampling processing, and finally convert the downsampled data into a time-frequency matrix P. norm [i]; (6) The spectral distribution features X obtained under normal detection mode norm and time-frequency matrix P norm [i] is used as a standard spectrum distribution template and time-frequency matrix template and deployed to the FPGA; (7) Set the digital oscilloscope to anomaly detection mode, and then repeat steps (2) to (5) to obtain the spectral distribution characteristics X of the input signal in anomaly detection mode. in ={X in [0],X in [1],…,X in [n],…,X in [N-1]} and the time-frequency matrix P in [i]; (8) Judgment of abnormal input signals; (8.1) Errata Spectral Distribution Characteristics X norm and X in For each spectral component in X, if X norm and X in If there exists a set of spectral components that satisfy the relation: |X in [n]-X norm [n]|>X set X set If a threshold is set, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); otherwise, proceed to step (8.2). (8.2) Traversing the time-frequency matrix P norm [i] and P in [i], compare whether the channel numbers i are the same one-to-one. If the channel numbers are different, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); if the channel numbers are the same one-to-one, the starting point search and loop comparison method are used to compare all time-frequency matrices P with the same number. norm [i] and P in If the comparison results of elements at a certain position are different, the input signal is determined to be an abnormal signal, and then proceed to step (8.3); if the comparison results of the time-frequency matrices corresponding to the channel number are all the same, the input signal is determined to be a normal signal, and then proceed to step (8.3). (8.3) Input the determined signal to the host computer for real-time display.

2. The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to claim 1, characterized in that, The WOLA structure divides the full bandwidth of the acquisition system into m narrow-band channels in the digital domain. The WOLA structure is a 1-in-m-out structure, where the input is the full-bandwidth digital sampling signal and the corresponding output is m narrow-bandwidth sampling data. Its number of channels m satisfies: Among them, f s f is the real-time sampling rate of the data acquisition system. fpga The frequency of digital signal processing in the digital domain; The bandwidth of each channel is: The i-th channel M i The frequency range is:

3. The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to claim 1, characterized in that, The bandwidth corresponding to the channelization digital filter in each channel is: Each channel corresponds to a channelized digital filter, which is a bandpass digital filter whose passband range is the frequency range of that channel.

4. The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to claim 1, characterized in that, The anti-mirror filter is a low-pass filter with a cutoff frequency of .

5. The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to claim 1, characterized in that, The extraction multiplier D satisfies:

6. The real-time anomaly detection method based on the time-frequency domain in an intelligent oscilloscope according to claim 1, characterized in that, The time-frequency matrix P norm [i] and P in [i] The comparison method using starting point search and loop comparison is as follows: (6.1) Time-frequency matrix P norm [i] and P in Find the starting point of [i]; The time-frequency matrix P in [i] is the first column P in [i]1 and the time-frequency matrix P norm Compare each column of [i] to find the match with the time-frequency matrix P. in All elements in the first column of [i] satisfy the condition that... Starting from the column t0, proceed to step (6.2); otherwise, determine that the input signal is an abnormal signal and end the comparison; where P set To set the threshold, P in [i] 1,v Represents the time-frequency matrix P in [i] The v-th element in the first column; (6.2) Perform cyclical comparisons based on the starting point; P in The first column of [i] and P norm Compare with column t0 of [i], P in The second column in [i] and P norm The comparison is performed on the (t0+1)th column of [i], and so on, until the comparison reaches P. norm After the last column of [i], return to P. norm The comparison continues in the first column of [i] until the time-frequency matrix P is reached. norm [i] and P in All columns of [i] have been compared. During the comparison process, if any element in any column satisfies... At that time, it was believed that P in [i] and P norm If [i] is different, the input signal is determined to be an abnormal signal, and the comparison ends.

Images