Multi-feature fusion crossing machine signal recognition method, device, equipment and medium

Abstract

The application provides a multi-feature fusion crossing machine signal recognition method, device, equipment and medium, comprising: acquiring time-frequency features of a collected signal, judging whether there is a crossing machine suspected signal based on the time-frequency features; performing frequency domain accumulation on a high frequency part signal of the crossing machine suspected signal, performing mean filtering processing on the accumulated signal in the time domain direction, identifying the filtered signal based on the pulse width, and obtaining a first crossing machine signal; acquiring a type of the crossing machine suspected signal, superimposing the crossing machine suspected signal in the time domain in a pulse period length based on a pulse period corresponding to the type of the crossing machine suspected signal, identifying the superimposed pulse signal based on the pulse width, and obtaining a second crossing machine signal; performing intersection processing on the first crossing machine signal and the second crossing machine signal, and completing crossing machine signal recognition. The application can efficiently recognize the crossing machine signal, significantly reduces the calculation complexity, and has an extremely low false alarm rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle supervision, and particularly relates to a multi-feature fusion crossing machine signal identification method and device, equipment and a medium. BACKGROUND

[0002] The crossing machine is widely used due to its characteristics of high-speed flight, lightweight body and super maneuverability. However, with the rapid development of the crossing machine, "black flight" and "wild flight" incidents occur frequently, which not only pose a serious threat to aviation safety and public order, but also derive new security risks such as invasion of personal privacy and interference with sensitive areas. The surge in safety accidents caused by the illegal use of crossing machines makes unmanned aerial vehicle supervision and defense an urgent problem to be solved.

[0003] In the prior art, when the crossing machine signal is demodulated for image reconstruction, real-time decoding of the video stream is required, but due to the large amount of calculation in the process, the processing efficiency is low, which is difficult to match the real-time tracking demand of the crossing machine. The traditional detection mechanism based on RSSI threshold has significant limitations, especially in complex electromagnetic environments. The same frequency band environmental interference signals will cause the false alarm rate to surge, resulting in a decrease in detection sensitivity. Although the intelligent detection model based on the deep learning framework can show advantages in specific scenarios, it is limited by the magnitude of the accurate annotation dataset, and the model reasoning relies on the GPU acceleration environment. At the same time, there is also the problem of a decrease in recognition accuracy caused by environmental mutations. SUMMARY

[0004] In view of the defects in the prior art, the present application provides a multi-feature fusion crossing machine signal identification method, device, equipment and medium.

[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present application is as follows:

[0006] On the one hand, the present application provides a multi-feature fusion crossing machine signal identification method, comprising the following steps:

[0007] Obtaining the time-frequency feature of the collected signal, and judging whether there is a crossing machine suspected signal based on the time-frequency feature;

[0008] Performing frequency domain accumulation on the high frequency part of the crossing machine suspected signal, and performing mean filtering processing on the time domain direction of the accumulated signal, identifying the filtered signal based on the pulse width, and obtaining the first crossing machine signal;

[0009] Obtaining the type of the crossing machine suspected signal, superimposing the pulse period length of the crossing machine suspected signal in the time domain based on the pulse period corresponding to the type of the crossing machine suspected signal, and identifying the superimposed pulse signal based on the pulse width to obtain the second crossing machine signal;

[0010] The first crossing machine signal and the second crossing machine signal are processed by intersection, and crossing machine signal recognition is completed.

[0011] Further, time-frequency characteristics of the collected signal are acquired, and crossing machine suspected signals are obtained based on the time-frequency characteristics, including:

[0012] The collected signal is converted to the frequency domain by Fourier transform, and the power spectral density of the signal is calculated according to the number of sampling points and the sampling rate.

[0013] Two frequency points with a power drop of 3dB on both sides of the crossing machine signal channel frequency are acquired, and the bandwidth difference between the two frequency points is calculated, if the bandwidth difference between the two frequency points is within 6MHz±2MHz, it is considered that there is a suspected signal in the crossing machine signal channel; if there is no frequency point meeting the requirements or the bandwidth difference between the two frequency points is not within 6MHz±2MHz, it is considered that there is no suspected signal in the crossing machine signal channel.

[0014] Whether there is a crossing machine suspected signal is determined according to the time characteristic.

[0015] Further, whether there is a crossing machine suspected signal is determined according to the time characteristic, including:

[0016] The suspected signal in the time dimension of the crossing machine signal channel is divided by two using the mean value method, and points greater than the division value are determined as signals, and points less than the division value are determined as noise.

[0017] The proportion of rising and falling edges of the suspected signal to the total signal is counted, if the proportion is greater than 0.25, it is considered that there is a crossing machine suspected signal; if the proportion is not greater than 0.25, it is considered that there is no crossing machine suspected signal.

[0018] Further, the power spectral density of the signal is calculated according to the number of sampling points and the sampling rate:

[0019]

[0020] Wherein, is the power spectral density; is the number of sampling points; is the sampling rate; is the imaginary unit; is the time domain signal; is the signal frequency.

[0021] Further, the filtered signal is identified based on the pulse width, and the first crossing machine signal is obtained, including:

[0022] The filtered signal is segmented by threshold segmentation method;

[0023] Determine whether the signal with a pulse length within 18.4 ms ± 0.3 ms in every 20 ms from the start observation time is the first crossing machine signal.

[0024] Further, the superimposed pulse signal is identified based on the pulse width to obtain the second crossing machine signal, comprising:

[0025] The superimposed pulse signal is segmented by using a threshold segmentation method.

[0026] Determine whether the signal with a pulse length within 18.4 ms ± 0.3 ms in every 20 ms from the start observation time is the second crossing machine signal.

[0027] Further, the threshold segmentation method is performed according to the following formula:

[0028]

[0029] wherein, is a threshold value; is a maximum amplitude value; is a minimum amplitude value.

[0030] In another aspect, the present application provides a crossing machine signal recognition device based on multi-feature fusion, comprising:

[0031] A first module is configured to acquire time-frequency features of a collected signal, and determine whether there is a crossing machine suspected signal based on the time-frequency features;

[0032] A second module is configured to perform frequency domain accumulation on a high frequency part of the crossing machine suspected signal, perform mean filtering processing on the time domain direction of the accumulated signal, identify the filtered signal based on the pulse width, and obtain a first crossing machine signal;

[0033] A third module is configured to acquire a type of the crossing machine suspected signal, superimpose the crossing machine suspected signal in the time domain according to a pulse period corresponding to the type of the crossing machine suspected signal, identify the superimposed pulse signal based on the pulse width, and obtain a second crossing machine signal;

[0034] A fourth module is configured to perform intersection processing on the first crossing machine signal and the second crossing machine signal, and complete crossing machine signal recognition.

[0035] In another aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the crossing machine signal recognition method based on multi-feature fusion when executing the computer program.

[0036] In another aspect, the present application provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the multi-feature fusion crossing machine signal recognition method.

[0037] Compared with the prior art, the present application has the beneficial technical effects that:

[0038] The multi-feature fusion crossing machine signal recognition method, device, equipment and medium provided by the present application can realize preliminary screening of the collected signals by collecting the time-frequency features of the signals and judging whether there is a crossing machine suspected signal based on the time-frequency features, thereby reducing the data processing amount of subsequent signal recognition; the high-frequency part of the crossing machine suspected signal is subjected to frequency domain accumulation to eliminate field blanking signals, so that the accumulated pulse distribution of the high-frequency part of the signal is clearly displayed, and the accumulated signal is subjected to mean filtering processing in the time domain direction to increase the stability of the signal and suppress noise, so that the signal is smoothed, the accuracy and robustness of signal recognition are improved, and the filtered signal is recognized based on the pulse width to obtain a first crossing machine signal; the crossing machine suspected signal is subjected to pulse period length superposition in the time domain by using the pulse period corresponding to the type of the crossing machine suspected signal, so that the signal-to-noise ratio is superimposed to improve the detection capability, and the superimposed pulse signal is recognized based on the pulse width to obtain a second crossing machine signal. Finally, the first crossing machine signal and the second crossing machine signal are subjected to intersection processing to complete crossing machine signal recognition.

[0039] The multi-feature fusion recognition model constructed by comprehensively considering the time-frequency features, energy distribution characteristics, periodic pulse characteristics and pulse width characteristics of the crossing machine signal can efficiently recognize the crossing machine signal, significantly reduces the computational complexity, has an extremely low false alarm rate, improves the target recognition ability in a low signal-to-noise ratio scene, and is suitable for real-time scenes such as low-altitude security. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from the structures shown in the drawings without creative labor.

[0041] Figure 1 A multi-feature fusion crossing machine signal recognition method flowchart provided for an embodiment;

[0042] Figure 2 A high-frequency signal frequency domain accumulation filtering schematic diagram provided for an embodiment, wherein, Figure 2 (a) is a high-frequency signal frequency domain accumulation schematic diagram, Figure 2(b) is a schematic diagram of the accumulated signal after mean filtering in the time domain direction;

[0043] Figure 3 The low-frequency signal frequency domain accumulation filtering schematic diagram provided by an embodiment is shown in the figure, Figure 3 (a) is a low-frequency signal frequency domain accumulation schematic diagram, Figure 3 (b) is a schematic diagram of the accumulated signal after mean filtering in the time domain direction;

[0044] Figure 4 The PAL system signal provided by an embodiment is shown in the figure, in which the PAL system pulse period is superimposed in the time domain with the length of the pulse period.

[0045] Figure 5 The PAL system signal provided by an embodiment is shown in the figure, in which the PAL system pulse period is superimposed in the time domain with the length of the pulse period. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0047] Referring to Figure 1 An embodiment provides a multi-feature fusion crossing machine signal recognition method, including the following steps:

[0048] Obtain the time-frequency feature of the collected signal, and determine whether there is a crossing machine suspected signal based on the time-frequency feature;

[0049] Perform frequency domain accumulation on the high-frequency part of the crossing machine suspected signal, and perform mean filtering in the time domain direction on the accumulated signal, recognize the filtered signal based on the pulse width, and obtain a first crossing machine signal;

[0050] Obtain the system of the crossing machine suspected signal, superimpose the pulse period of the crossing machine suspected signal in the time domain with the pulse period corresponding to the system of the crossing machine suspected signal, recognize the superimposed pulse signal based on the pulse width, and obtain a second crossing machine signal;

[0051] Perform intersection processing on the first crossing machine signal and the second crossing machine signal, and complete the crossing machine signal recognition.

[0052] In the method, time-frequency characteristics of the collected signal are acquired, and whether the crossing machine suspected signal exists is determined based on the time-frequency characteristics, so as to realize preliminary screening of the collected signal, and reduce the data processing amount of subsequent signal identification; the high-frequency part signal of the crossing machine suspected signal is accumulated in the frequency domain, so as to eliminate the field blanking signal, thereby clearly showing the accumulated pulse distribution of the high-frequency part of the signal, and the mean filtering processing in the time domain is performed on the accumulated signal, so as to increase the stability of the signal, suppress the noise, smooth the signal, improve the accuracy and robustness of signal identification, and then the filtered signal is identified based on the pulse width, so as to obtain the first crossing machine signal; the pulse cycle length of the crossing machine suspected signal in the time domain is superimposed by taking the pulse cycle corresponding to the type of the crossing machine suspected signal, so as to superimpose the signal-to-noise ratio and improve the detection capability, and then the superimposed pulse signal is identified based on the pulse width, so as to obtain the second crossing machine signal. Finally, the intersection processing is performed on the obtained first crossing machine signal and second crossing machine signal, and the crossing machine signal identification is completed.

[0053] In a preferred embodiment, time-frequency characteristics of the collected signal are acquired, and the crossing machine suspected signal is screened based on the time-frequency characteristics, including:

[0054] The collected signal is converted to the frequency domain by Fourier transform, and the power spectral density of the signal is calculated according to the sampling point number and the sampling rate;

[0055] Two frequency points with a power drop of 3dB on both sides of the crossing machine signal channel frequency are acquired, and the bandwidth difference value between the two frequency points is calculated, if the bandwidth difference value between the two frequency points is within 6MHz±2MHz, it is considered that the suspected signal exists in the crossing machine signal channel; if there is no frequency point meeting the requirements or the bandwidth difference value between the two frequency points is not within 6MHz±2MHz, it is considered that the suspected signal does not exist in the crossing machine signal channel;

[0056] Whether the crossing machine suspected signal exists is determined according to the time characteristic.

[0057] Firstly, the power spectral density of the signal is calculated based on the frequency domain characteristics of the collected signal, so as to find two frequency points with a power drop of 3dB on both sides of the crossing machine signal channel frequency, and calculate the bandwidth difference value between the two frequency points, thereby screening the collected signal and determining whether the suspected signal exists in the crossing machine signal channel; since the bandwidth of the crossing machine analog image signal is mostly 6MHz, usually within 5MHz-8MHz, 6MHz±2MHz is taken as the bandwidth difference determination range; if the suspected signal exists, the signal is determined based on the time characteristic, and the crossing machine suspected signal is screened out. Through twice screening, a large number of samples can be quickly distinguished, the sample quantity is reduced, the calculation amount is reduced, the calculation time is saved, and the calculation complexity is reduced.

[0058] In a preferred embodiment, judging whether there is a suspected signal of the crossing machine according to the time characteristic comprises:

[0059] The suspected signal of the crossing machine signal channel time dimension is divided into two parts by using the mean value method, and the points greater than the divided value are determined as signals, and the points less than the divided value are determined as noises.

[0060] The proportion of the rising and falling edges of the suspected signal in the total signal is counted, if the proportion is greater than 0.25, it is considered that there is a suspected signal of the crossing machine, if the proportion is not greater than 0.25, it is considered that there is no suspected signal of the crossing machine.

[0061] For digital signals, the signal is an on / off mode, and common ones are square waves, so the rising and falling edges of the digital signal after division are relatively few, and the proportion in the total signal is low. Through the above setting, the digital signal and the analog signal in the suspected signal of the crossing machine signal channel time dimension are effectively distinguished, and the existence of the suspected signal of the crossing machine is determined.

[0062] The power spectral density of the signal is calculated according to the number of sampling points and the sampling rate as follows:

[0063]

[0064] Among them, is the power spectral density; is the number of sampling points; is the sampling rate; is the imaginary unit; is the time domain signal; is the signal frequency.

[0065] For the acquisition of the first crossing machine signal, specifically, the crossing machine analog image signal uses NTSC and PAL (phase alternating line) two systems, considering that the frequency of the field blanking signal in the two systems is in the low frequency part of the baseband signal, therefore, the high frequency part signal of the crossing machine suspected signal is selected for frequency domain accumulation to eliminate the field blanking signal, so as to clearly show the cumulative pulse distribution of the high frequency part of the signal.

[0066] In an embodiment, the results of frequency domain accumulation of the high frequency part and the low frequency part of the crossing machine suspected signal and the mean value filtering processing in the time domain direction are respectively shown, and the results are shown in Figure 2 , Figure 3 From Figure 2 (a), it can be seen that after the frequency domain accumulation of the high frequency part of the crossing machine suspected signal, the cumulative pulse distribution of the high frequency part of the signal can be clearly seen, Figure 2 (b) can be seen, after the mean value filtering processing in the time domain direction, the signal curve becomes smooth, which can more accurately identify the signal. From Figure 3It can be seen that the low frequency part will have a certain impact on the detection sensitivity due to the influence of the line synchronization signal; specifically, because of the existence of the synchronization signal, the noise part of the signal will be raised, thereby reducing the detection sensitivity; and the field blanking signal pulse is short in time and narrow in frequency bandwidth, and the signal recognition error rate is high; specifically, the narrow frequency domain accumulation range is easy to miss the signal, and the wide frequency domain accumulation range averages the signal energy, thereby leading to low sensitivity, so the low frequency part is not selected for processing and subsequent identification.

[0067] In an embodiment, the PAL system crossing machine image signal is identified, and the filtered signal is identified based on the pulse width to obtain a first crossing machine signal, including:

[0068] The filtered signal is segmented by using a threshold segmentation method.

[0069] It is determined that, in the segmented signal, the signal with a pulse length of 18.4ms±0.3ms within 20ms from the start observation time is the first crossing machine signal.

[0070] The pulse length and the period are determined by superimposing the threshold segmentation and the period (the pulse period corresponding to the system) to determine the signal with the pulse length and the period meeting the above setting as the first crossing machine signal.

[0071] The threshold segmentation method is performed according to the following formula:

[0072]

[0073] wherein, is a threshold value; is a maximum amplitude value; is a minimum amplitude value.

[0074] For obtaining the second crossing machine signal, specifically, the crossing machine analog image signal uses NTSC and PAL two systems, the NTSC system has a refresh rate of 60Hz and a pulse period of 16.67ms; the PAL system has a refresh rate of 50Hz and a pulse period of 20ms; considering that the pulse characteristics of the crossing machine are consistent in each period, the pulse signal time domain is superimposed according to the period length to maintain the original pulse characteristics. Therefore, the crossing machine suspected signal is superimposed in the time domain according to the pulse period corresponding to the crossing machine suspected signal system, that is, 20ms-40ms data is superimposed on 0ms-20ms signal, and then 40ms-60ms data is superimposed on the last superimposed signal, and the processing is performed in this way to obtain a complete period of pulse signal, and the superimposition operation can superimpose the signal-to-noise ratio to improve the detection capability.

[0075] In an embodiment, a processing result after superimposing the period length corresponding to the PAL system and the period length corresponding to the non-system is shown, and the result is shown in FIGS. Figure 4 , Figure 5 Figure 4 A signal diagram for superimposing the pulse period length in the time domain with the PAL system signal using the PAL system pulse period can be seen that the superimposed signal forms a complete period, and compared with Figure 4 and Figure 2 It can be seen that the crossing machine signal can be more obviously identified, and the detection ability is improved. Figure 5 A signal diagram for superimposing the pulse period length in the time domain with the PAL system signal using the NTSC system pulse period can be seen from the figure that the period length corresponding to the non-system is superimposed after multiple signal pulses appear, which does not conform to the signal characteristics and is not convenient for identifying the crossing machine signal.

[0076] In an embodiment, the PAL system crossing machine image transmission signal is identified, the superimposed pulse signal is identified based on the pulse width, and the second crossing machine signal is obtained, including:

[0077] The superimposed pulse signal is segmented by using a threshold segmentation method;

[0078] It is determined that in the segmented signal, from the start observation time, the signal with a pulse length of 18.4ms±0.3ms within 20ms is the second crossing machine signal.

[0079] In an embodiment, a crossing machine signal identification device with multi-feature fusion is provided, including:

[0080] The first module is used for acquiring the time-frequency features of the collected signal, and determining whether there is a crossing machine suspected signal based on the time-frequency features;

[0081] The second module is used for accumulating the high-frequency part signal of the crossing machine suspected signal in the frequency domain, and performing mean filtering processing in the time domain on the accumulated signal, identifying the filtered signal based on the pulse width, and obtaining the first crossing machine signal;

[0082] The third module is used for acquiring the system of the crossing machine suspected signal, superimposing the pulse period length in the time domain of the crossing machine suspected signal with the pulse period corresponding to the system of the crossing machine suspected signal, and identifying the superimposed pulse signal based on the pulse width to obtain the second crossing machine signal;

[0083] The fourth module is used for performing intersection processing on the first crossing machine signal and the second crossing machine signal to complete the crossing machine signal identification.

[0084] ​In another aspect, the present application provides a computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the multi-feature fusion crossing machine signal recognition method provided in any of the above embodiments when executing the computer program. The computer device can be a server. The computer device comprises a processor, a memory, a network interface and a database connected by a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store sample data. The network interface of the computer device is configured to communicate with an external terminal through a network connection.

[0085] In another aspect, the present application provides a computer readable storage medium storing a computer program, the computer program being executed by a processor to implement the steps of the multi-feature fusion crossing machine signal recognition method provided in any of the above embodiments.

[0086] It is understood by those skilled in the art that all or part of the processes of the above embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM).

[0087] The details of the present application are as described above.

[0088] Any technical features in the above-described embodiments can be combined in any manner, and for the sake of brevity, not all possible combinations are described, however, any combination of the technical features is considered to be within the scope of the present specification.

[0089] The above-described embodiments are merely illustrative for the present application and are not used to limit the present application. It should be pointed out that, for those skilled in the art, some modifications and improvements can be made without departing from the concept of the present application, and these should be included in the protection scope of the present application.

[0090] The above-described embodiments are merely illustrative for the present application and are not used to limit the present application. It should be pointed out that, for those skilled in the art, some modifications and improvements can be made without departing from the concept of the present application, and these should be included in the protection scope of the present application. The above-described embodiments are merely illustrative for the present application and are not used to limit the present application. It should be pointed out that, for those skilled in the art, some modifications and improvements can be made without departing from the concept of the present application, and these should be included in the protection scope of the present application.

Claims

1. A multi-feature fusion crossing machine signal recognition method, characterized in that, The method comprises the following steps: Obtaining time-frequency characteristics of the collected signal, and determining whether there is a suspected signal of the crossing machine based on the time-frequency characteristics, comprising: Converting the collected signal to the frequency domain through Fourier transform, and calculating the power spectral density of the signal according to the number of sampling points and the sampling rate; Obtaining two frequency points on both sides of the frequency of the crossing machine signal channel with a power drop of 3dB, and calculating the bandwidth difference between the two frequency points, if the bandwidth difference between the two frequency points is within 6MHz±2MHz, it is considered that there is a suspected signal in the crossing machine signal channel; if there is no frequency point meeting the requirements or the bandwidth difference between the two frequency points is not within 6MHz±2MHz, it is considered that there is no suspected signal in the crossing machine signal channel; Determining whether there is a suspected signal of the crossing machine according to the time characteristics, comprising: Dividing the suspected signal of the crossing machine signal channel in the time dimension by two using the mean value method, and determining that the points greater than the division value are signals and the points less than the division value are noises; Statistically analyzing the proportion of the rising and falling edges of the suspected signal in the total signal, if the proportion is greater than 0.25, it is considered that there is a suspected signal of the crossing machine; if the proportion is not greater than 0.25, it is considered that there is no suspected signal of the crossing machine; Performing frequency domain accumulation on the high frequency part of the suspected signal of the crossing machine, and performing mean value filtering processing on the accumulated signal in the time domain direction, identifying the filtered signal based on the pulse width to obtain a first crossing machine signal; Obtaining the mode of the suspected signal of the crossing machine, superimposing the pulse period length of the suspected signal of the crossing machine in the time domain based on the pulse period corresponding to the mode of the suspected signal of the crossing machine, and identifying the superimposed pulse signal based on the pulse width to obtain a second crossing machine signal; Performing intersection processing on the first crossing machine signal and the second crossing machine signal to complete the identification of the crossing machine signal.

2. The multi-feature fused crossing machine signal identification method of claim 1, wherein, The power spectral density of the signal calculated according to the number of sampling points and the sampling rate is: wherein, is the power spectral density; is the number of samples; is the sampling rate; is the imaginary unit; is the time domain signal; is the signal frequency.

3. The multi-feature fused crossing machine signal identification method of claim 1, wherein, Identifying the filtered signal based on the pulse width to obtain a first crossing machine signal, comprising: Segmenting the filtered signal by threshold segmentation; Determining that the signal in the segmented signal, starting from the initial observation time, within every 20ms, the pulse length within 18.4ms±0.3ms is the first crossing machine signal.

4. The multi-feature fused crossing machine signal identification method of claim 1, wherein, Identifying the superimposed pulse signal based on the pulse width to obtain a second crossing machine signal, comprising: Segmenting the superimposed pulse signal by threshold segmentation; Determining that the signal in the segmented signal, starting from the initial observation time, within every 20ms, the pulse length within 18.4ms±0.3ms is the second crossing machine signal.

5. The multi-feature fused crossing-machine signal identification method of any one of claims 3 or 4, wherein, The threshold segmentation method is according to the following formula: wherein is a threshold value; is a maximum amplitude value; is a minimum amplitude value.

6. A multi-feature fusion crossing machine signal recognition device, characterized in that, Comprising: The first module is used for obtaining the time-frequency characteristics of the collected signal, and determining whether there is a suspected signal of the crossing machine based on the time-frequency characteristics, comprising: Converting the collected signal to the frequency domain through Fourier transform, and calculating the power spectral density of the signal according to the number of sampling points and the sampling rate; Two frequency points are acquired at which the power of the signal channel of the crossing machine decreases by 3dB on both sides of the frequency, and the bandwidth difference between the two frequency points is calculated. If the bandwidth difference between the two frequency points is within 6MHz±2MHz, it is considered that there is a suspected signal in the signal channel of the crossing machine. If there is no frequency point meeting the requirements or the bandwidth difference between the two frequency points is not within 6MHz±2MHz, it is considered that there is no suspected signal in the signal channel of the crossing machine. According to the time characteristics, it is judged whether there is a suspected signal of the crossing machine, including: The suspected signal of the time dimension of the signal channel of the crossing machine is divided into two parts by using the mean value method, and the points greater than the division value are determined as signals, and the points less than the division value are determined as noises. The proportion of the rising and falling edges of the suspected signal to the total signal is counted. If the proportion is greater than 0.25, it is considered that there is a suspected signal of the crossing machine. If the proportion is not greater than 0.25, it is considered that there is no suspected signal of the crossing machine. The second module is used for frequency domain accumulation of the high frequency part of the suspected signal of the crossing machine, and mean value filtering processing of the accumulated signal in the time domain direction, identification of the filtered signal based on the pulse width, and obtaining of the first crossing machine signal. The third module is used for acquiring the type of the suspected signal of the crossing machine, superimposing the pulse period length of the suspected signal of the crossing machine in the time domain based on the pulse period corresponding to the type of the suspected signal of the crossing machine, and identifying the superimposed pulse signal based on the pulse width to obtain the second crossing machine signal. The fourth module is used for intersection processing of the first crossing machine signal and the second crossing machine signal to complete the crossing machine signal identification. 7.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-6 when the computer program is executed by the processor. The processor executes the computer program to realize the multi-feature fusion crossing machine signal identification method of any one of claims 1 to 4.

8. A computer-readable storage medium, characterized in that, The computer program is stored on the processor and executed by the processor to realize the multi-feature fusion crossing machine signal identification method of any one of claims 1 to 4.

Images