A drone detection method, device and equipment based on data-enhanced machine learning

Abstract

The embodiments of the present disclosure provide a method, device and equipment for detecting drones based on data-enhanced machine learning. The method includes: automatically collecting original drone signals; segmenting the collected continuous original drone signals; performing signal transformation on the segmented continuous signal segments; inputting the transformed signals into a neural network for training to obtain a decision network model; and sending the trained decision network model to a decision network module. The method also includes: segmenting the signal to be distinguished; inputting the signal segments into a decision network module to obtain a decision result of a single signal segment; and performing a joint decision of a sliding window on the decision results of continuous signal segments. The method simulates drone signals in multiple scenarios through signal transformation, thereby improving the accuracy of the decision network model. At the same time, the method uses a sliding window of continuous signal segments to achieve joint decision, thereby improving the accuracy of the drone signal detection results.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence and signal processing, and specifically to a drone detection method, device and equipment based on data-enhanced machine learning. Background Art

[0002] The drone industry has shown a very rapid development trend in recent years, and has gradually produced important applications in a wide range of industries such as police, urban management, agriculture, geology, meteorology, electricity, disaster relief, military, video shooting, etc. However, the development of drones will also bring a certain degree of security risks. Due to the concealment and convenience of drones themselves, if drones are used in the fields of illegal eavesdropping, illegal filming, illegal transportation of goods, etc., it will have a huge negative impact. Regulating the use of drones in daily life from a legislative level is one of the effective ways to prevent the illegal use of drones mentioned above. However, at the physical level, professional detection methods are still needed to detect and defend against illegal drone flights. Drone detection technology is the basis for establishing illegal drone flight defense and management.

[0003] In the current field of drone detection, many detection methods based on radar detectors, visual recognition, infrared spectrum recognition, acoustic array recognition, drone signal recognition and other technologies have been proposed. However, at a long distance, the accuracy of visual recognition detection and radar detection systems is limited, and it is often difficult to distinguish between birds and drones; the recognition method of acoustic arrays is also prone to the disadvantage of a narrow range of recognition; the infrared spectrum recognition method is very susceptible to interference from other heat sources. Overall, the method of detecting drones through drone signals can overcome geographical limitations to a large extent and has a certain degree of accuracy.

[0004] If you want to detect drone signals, it is very important to establish a suitable drone signal judgment network model. Machine learning is a research hotspot in the field of artificial intelligence, mainly extracting the features of input signals by training neural networks. Figure 1 This is a simple diagram of machine learning, where the circles represent the values ​​input to the neural network and the arrows represent the operations. The essence of the neural network training process is to find the optimal operation parameters (i.e. Figure 1 The training of neural network is of great significance to the establishment of the final judgment network model. The present invention is based on the technology of data-enhanced machine learning, and generates an optimal drone signal judgment network model by training the neural network, and then uses the model to detect and identify drones.

[0005] There are also methods and devices in the prior art that process drone signal data and then train it to obtain a better classification model, which is then used for drone detection and identification, such as patent application CN202110553890.9, which performs energy-time-frequency domain analysis on the normalized drone signal samples to obtain a time domain energy change spectrum; based on the time domain energy change spectrum, the drone signal samples are threshold sorted to obtain valid signal segments and noise segments; the valid signal segments are sliced ​​using a sliding window method to obtain multiple signal sequence segments of equal length; the feature data of each signal sequence segment is extracted, and the feature data is clustered and analyzed to obtain multiple unrelated signal subsets; the above steps are performed on the drone signal samples of each signal type, and the obtained signal subset is added to the feature library, and the classification model is trained using the feature library.

[0006] However, the drone identification device that currently stores the decision network model is still not very effective in identifying drone signals. There are two main reasons for this: First, the decision network currently mainly judges a single signal segment and outputs the final decision result. A single signal segment is more likely to be interfered with, which can easily lead to inaccurate final judgment results. Second, when training the decision network model, limited signal segments of a single scene are often input, which can easily lead to a decrease in the recognition accuracy of the decision network model for other scenes other than the single scene input during neural network training or complex scenes with multiple scenes superimposed. At the same time, collecting drone signal segments in many scenes requires a lot of time and energy, consumes a lot of costs, and the collected signal segments in different scenes are still relatively limited. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a drone detection method, device and equipment based on data enhanced machine learning. In the process of training the judgment network model, there is no need to actually collect drone signals in many scenarios. It is only necessary to collect drone signals in conventional scenarios. Then, through the signal transformation method, drone signals in many scenarios are obtained, and then each signal segment is input into the neural network for training.

[0008] After obtaining the trained decision network model, the collected signal to be distinguished is segmented, and the segmented signal segments are input into the decision network module containing the decision network model for judgment to obtain the judgment result of a single signal segment, and then the judgment results of continuous segments are jointly judged by a sliding window to obtain the final judgment result.

[0009] The specific technical solutions of the present invention are as follows:

[0010] In a first aspect, the present application provides a drone detection method based on data-enhanced machine learning, the method comprising:

[0011] Step 1: Use the drone signal acquisition device to automatically collect the original communication signal of the drone.

[0012] Step 2: segment the collected original signal of the drone to obtain multiple continuous signal segments with the same length.

[0013] Step 3, performing signal transformation on the signal segment obtained in step 2, inputting the transformed signal into a neural network for training, and obtaining a decision network model.

[0014] Step 4: segment the collected signal to be resolved to obtain multiple continuous signal segments with the same length.

[0015] Step 5: Input the segmented signal to be distinguished into the decision network model for decision, and obtain the decision result of the single signal segment.

[0016] Step 6: Perform joint decision of the sliding window according to the decision results of the continuous segments.

[0017] In the step 1, the original communication signal of the drone collected by the drone signal collection device includes the communication signal used when the drone communicates with the ground and the control end, and also includes the environmental signal.

[0018] In the step 1, after the original signal is collected, the collected original signal is added to the signal library.

[0019] In the step 3, the signal segment is transformed, which may include changing the transmission frequency of the signal, changing the signal-to-interference ratio of the signal, changing the signal-to-noise ratio of the signal, etc.

[0020] A wireless channel refers to a data signal transmission channel that uses wireless signals as the transmission medium. A drone usually has multiple wireless channels, corresponding to multiple transmission frequencies. By changing the transmission frequency of the signal, the signal in each wireless channel of the drone can be simulated, thereby increasing the input scenarios of the neural network.

[0021] The method for changing the transmission frequency of a signal includes: assuming that the original signal is {x[0], x[1], x[2], ..., x[N-1]}, where N is the length of the wireless signal. Assuming that the channel frequency of the original signal is f0, the set of all channel frequencies of the signal is {f0, f1, ..., f M}, where M is the total number of channels. Assume that the signal frequency point to be transformed in the data enhancement parameter is f m , the sampling frequency is f s , then the signal is transformed into:

[0022]

[0023] The signal-to-interference ratio conversion can simulate different scenarios in which a receiving end in a wireless communication system receives a useful signal and an interference signal at the same time when receiving a signal. The signal-to-interference ratio conversion is specifically:

[0024] Assume that the required signal-to-interference ratio is α, and the interference signal is {I[0],I[1],I[2],...,I[N-1]}, then

[0025]

[0026] in:

[0027] Then, the signal is transformed into:

[0028]

[0029] The signal-to-noise ratio conversion can simulate the attenuation effect of different degrees of attenuation due to the transmission distance when the wireless signal reaches the receiving end. The signal-to-noise ratio conversion is specifically:

[0030] Assume the noise power is P noise , the required signal-to-noise ratio is β, then:

[0031]

[0032] in:

[0033] Then, the signal is transformed into:

[0034]

[0035] in, The power is P noise is a cyclically symmetric zero-mean complex Gaussian random variable.

[0036] After the signal transformation in step 3, the segmented original signal is transformed into signals in multiple scenarios rather than signals in a single scenario, and then input into the neural network for training, so that the neural network can extract the drone signal features in multiple scenarios and improve the recognition accuracy of the judgment network model.

[0037] In step 6, the joint decision is specifically as follows: set the sliding window size to K, count the number of signal segments determined to contain the target drone in K consecutive signal segments, and record it as K pos If K posIf the value is greater than the threshold, the final judgment result is that there is a drone in the surrounding environment; otherwise, the final judgment result is that there is no drone in the surrounding environment.

[0038] The joint judgment of multiple signal segments in step 6 reduces the misjudgment caused by a single signal segment.

[0039] In a second aspect, the present application provides a drone detection device based on data-enhanced machine learning, the device comprising:

[0040] The UAV signal acquisition module automatically collects the communication signals of the UAV to be identified.

[0041] The signal segment module is connected to the drone signal acquisition module to segment the collected drone communication signal to be identified to obtain multiple continuous signal segments.

[0042] The decision network module is connected to the signal segment module and is used to make a decision on the segmented signal to obtain a decision result of a single signal segment. The decision network module includes a trained decision network model.

[0043] The trained decision network model is obtained by segmenting the collected original signal fragments of the drone, performing signal transformation, and inputting the transformed signals into the neural network for training.

[0044] The signal transformation includes: changing the transmission frequency of the signal, and / or changing the signal-to-interference ratio of the signal, and / or changing the signal-to-noise ratio of the signal.

[0045] The sliding window joint decision module is connected to the decision network module and performs a sliding window joint decision according to the decision results of the continuous segments.

[0046] The communication signals of the drone collected by the drone signal collection module include the communication signals used by the drone to communicate with the ground and the control terminal, and also include environmental signals. The method and steps of the automatic collection are the same as those described above and will not be repeated here.

[0047] The specific joint decision is as follows: set the sliding window size to K, count the number of signal segments that are judged to contain the target drone in K consecutive signal segments, and record it as K pos If K pos If the value is greater than the threshold, the final judgment result is that there is a drone in the surrounding environment; otherwise, the final judgment result is that there is no drone in the surrounding environment.

[0048] The joint judgment of multiple signal segments reduces the misjudgment caused by a single signal segment.

[0049] In a third aspect, the present application also provides a drone detection device based on data-enhanced machine learning, wherein the device includes a memory and a processor.

[0050] The memory is used to store program codes and transmit the program codes to the processor.

[0051] The processor is used to execute the drone detection method described in the first aspect according to the instructions in the program code.

[0052] In a fourth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the post-processing method for drone detection and identification based on data-enhanced machine learning and the steps of the drone detection method described in the first aspect.

[0053] The beneficial effects of this application include:

[0054] 1. When training a neural network to establish a decision network model, the present application performs arbitrary signal transformation on the collected original drone signal to transform it into signals for multiple scenarios rather than signals for a single scenario. Changing the transmission frequency of the signal can simulate the signals in each wireless channel of the drone. Using signal-to-interference ratio transformation for data transformation can simulate the actual situation in which the receiving end in a wireless communication system receives both useful signals and interference signals when receiving signals. Using signal-to-noise ratio transformation for data transformation can simulate the attenuation effect of different degrees of attenuation due to the transmission distance when the wireless signal transmission reaches the receiving end. The signals in multiple scenarios after signal transformation are input into the neural network, so that the neural network can extract the drone signal features in multiple scenarios, thereby improving the accuracy of the decision network model in identifying drone signals.

[0055] 2. When the judgment network of the present application makes a judgment on the signal to be distinguished, a joint judgment of the sliding window is performed based on the judgment results of the continuous segments to obtain the final judgment result, thereby avoiding the problem of misjudgment caused by judging a single signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly and clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. The drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0057] Figure 1 A schematic diagram of the structure of a neural network model in a drone detection device based on data-enhanced machine learning;

[0058] Figure 2 A flowchart of a drone detection method based on data-enhanced machine learning provided in an embodiment of the present application;

[0059] Figure 3 A schematic diagram of the structure of a drone detection device based on data-enhanced machine learning provided in an embodiment of the present application;

[0060] Figure 4 A schematic diagram of a sliding window of a drone detection device based on data-enhanced machine learning provided in an embodiment of the present application.

[0061] Figure 5 A schematic diagram of the structure of a drone detection device based on data-enhanced machine learning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0063] In the current drone detection based on the radio signals of drones, after the training of the neural network based on the radio signals of drones is completed, that is, in the actual drone identification process using the drone identification device storing the trained decision network model, the recognition effect is still not ideal due to the following two reasons: First, currently in the decision network, judgment is mainly made on a single signal segment and the final judgment result is output. The single signal segment is more likely to be interfered with, which easily leads to inaccurate judgment results output by the decision network. Second, when the decision network model is currently trained, a limited single scene signal segment is often input, which easily leads to a decrease in the accuracy of the decision network model for drone signal detection in scenes other than the single scene input during neural network training or complex scenes with multiple scenes superimposed.

[0064] To this end, the present application provides a method, device and equipment for detecting drones based on data-enhanced machine learning. In drone identification, there is no need to actually provide drone signals in many scenarios. It is only necessary to collect drone signals in conventional scenarios, and then obtain drone signals in multiple scenarios through signal transformation. The transformed signal segments are input into the neural network for training to obtain a decision network model, and the trained decision network model is sent to the decision network module. After the sending is completed, the signal to be distinguished is segmented, and the segmented continuous signal segments are input into the decision network module for judgment. The judgment result of a single signal segment is obtained, and the judgment result of the continuous signal segment is jointly judged by the sliding window to obtain the final judgment result.

[0065] The following specifically introduces an embodiment of a data enhancement method in signal machine learning and a drone detection method provided by this application, with reference to Figure 2 , is a flow chart of a method provided in an embodiment of the present application, the method specifically includes:

[0066] S101: Use the drone signal acquisition device to collect the original communication signal of the drone.

[0067] In the embodiment of the present application, the original communication signal of the drone collected by the drone signal collection device includes the communication signal used by the drone to communicate with the ground and the control end, and also includes the environmental signal. In addition, after the signal is collected, the collected signal is added to the signal library.

[0068] S102: Segment the collected original drone signal to obtain multiple continuous signal segments with the same length.

[0069] S103: Perform signal transformation on the multiple continuous signal segments, and use the transformed signals as inputs of the decision network.

[0070] S104: Input the transformed signal into a neural network for training to obtain a decision network model, and send the trained decision network model to a decision network module.

[0071] S201: Using drone signal acquisition equipment, the communication signal of the drone to be identified is collected.

[0072] S202: Using a signal segment module, the collected drone signal to be identified is segmented to obtain a plurality of continuous signal segments of the same length.

[0073] S203: using the decision network module to make a decision on the segmented signal to obtain a decision result of a single signal segment.

[0074] S204: Using the sliding window joint decision module, perform sliding window joint decision according to the decision results of the continuous segments to obtain a final decision result.

[0075] In S103, signal transformation is performed on the signal segment, including changing the transmission frequency of the signal, changing the signal-to-interference ratio of the signal, changing the signal-to-noise ratio of the signal, etc.

[0076] A wireless channel refers to a data signal transmission channel that uses wireless signals as the transmission medium. A drone usually has multiple wireless channels. Channelization conversion is to convert the wireless signal on a certain channel into the signal on any other channel.

[0077] The method for changing the transmission frequency of the signal is as follows: Assume that the original signal is {x[0], x[1], x[2], ..., x[N-1]}, where N is the length of the wireless signal. Assume that the channel frequency of the original signal is f0, and the set of all channel frequencies of the signal is {f0, f1, ..., f M}, where M is the total number of channels. Assume that the signal frequency point to be transformed in the data enhancement parameter is f m , the sampling frequency is f s , then the signal is transformed into:

[0078]

[0079] The signal-to-interference ratio conversion is to simulate the actual situation that the receiving end in the wireless communication system receives both the useful signal and the interference signal when receiving the signal. The signal-to-interference ratio conversion is specifically:

[0080] Assume that the required signal-to-interference ratio is α, and the interference signal is {I[0],I[1],I[2],...,I[N-1]}, then

[0081]

[0082] in:

[0083] Then, the signal is transformed into:

[0084]

[0085] The signal-to-noise ratio conversion is to simulate the attenuation effect of different degrees of attenuation due to the transmission distance when the wireless signal reaches the receiving end. The signal-to-noise ratio conversion is specifically:

[0086] Assume the noise power is P noise , the required signal-to-noise ratio is β, then:

[0087]

[0088] in:

[0089] Then, the signal is transformed into:

[0090]

[0091] in, The power is P noise is a cyclically symmetric zero-mean complex Gaussian random variable.

[0092] The specific joint decision is as follows: set the sliding window size to K, count the number of signal segments that are judged to contain the target drone in K consecutive signal segments, and record it as K pos If K pos If the value is greater than the threshold, the final judgment result is that there is a drone in the surrounding environment; otherwise, the final judgment result is that there is no drone in the surrounding environment.

[0093] When training the neural network, the present invention transforms the original drone signal into signals in multiple scenarios rather than signals in a single scenario, so that the decision network model can more accurately identify the drone signal. At the same time, multiple signal segments are used for joint decision making for the signal to be distinguished, reducing misjudgment caused by a single signal segment.

[0094] Corresponding to the above method embodiments, the present application also provides a drone detection device based on data enhanced machine learning, see Figure 3 , the device comprises:

[0095] Signal acquisition module, which automatically collects the drone signals to be identified.

[0096] The signal segment module is connected to the signal acquisition module and segments the collected UAV communication signal to be identified to obtain multiple continuous signal segments.

[0097] The decision network module is connected to the signal segment module and is used to make a decision on the segmented signal to obtain a decision result of a single signal segment. The decision network module includes a trained decision network model.

[0098] The decision network model is obtained by segmenting the collected original signal fragments of the drone and then performing signal transformation, and inputting the transformed data into the neural network for training.

[0099] The signal transformation includes changing the transmission frequency of the signal, and / or changing the signal-to-interference ratio of the signal, and / or changing the signal-to-noise ratio of the signal, which will not be described in detail here.

[0100] The sliding window joint decision module is connected to the decision network module, and performs a sliding window joint decision according to the decision results of the continuous segments to obtain a final decision result.

[0101] The embodiment of the present invention also provides a sliding window schematic diagram of a drone detection device based on data enhanced machine learning, see Figure 4 As shown:

[0102] Set the sliding window size to K, and count the number of signal segments that are determined to contain the target drone in K consecutive signal segments, denoted as K pos If K posIf the value is greater than the threshold, the final judgment result is that there is a drone in the surrounding environment; otherwise, the final judgment result is that there is no drone in the surrounding environment.

[0103] exist Figure 4 In the illustrated exemplary embodiment, the sliding window size K=3. It should be understood that a larger or smaller sliding window may be used. For example, a sliding window with a window size K=2, 4, 5, 6, etc. may be used. Here, it is sufficient that K is a natural number greater than or equal to 2.

[0104] The embodiment of the present invention also provides a drone detection device based on data enhanced machine learning, see Figure 5 As shown, it may include:

[0105] Processor 501, memory 502, input device 503 and output device 504. The number of processors 501 in the drone detection device based on data enhanced machine learning can be one or more. Figure 4 In some embodiments of the present invention, the processor 501, the memory 502, the input device 503 and the output device 504 may be connected via a bus or other means, wherein: Figure 4 The example of connecting through bus is taken in the following.

[0106] The memory 502 can be used to store software programs and modules. The processor 501 executes various functional applications and data processing of the drone detection device based on data-enhanced machine learning by running the software programs and modules stored in the memory 502. The memory 502 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, at least one application required for a function, and the like. In addition, the memory 502 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. The input device 503 may be used to receive input digital or character information, and to generate signal input related to user settings and function control of the drone detection device based on data-enhanced machine learning.

[0107] Specifically in this embodiment, the processor 501 will load the executable files corresponding to the processes of one or more applications into the memory 502 according to the following instructions, and the processor 501 will run the applications stored in the memory 502, thereby realizing the various functions of the above-mentioned drone detection method based on data enhanced machine learning.

[0108] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0109] It should be noted that, in this article, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or apparatus.

[0110] The above is a detailed introduction to a drone detection method, device and equipment based on data-enhanced machine learning provided in the embodiments of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the ideas of the present application, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A drone detection method based on data-enhanced machine learning, characterized in that: The method comprises: Automatically collect the original signal of the drone; Segmenting the collected original signal of the drone to obtain a plurality of continuous signal segments of the same length; Performing signal transformation on the plurality of signal segments, and using the transformed data as input for neural network training; wherein the signal transformation includes one or more of the following operations: changing the transmission frequency of the signal, changing the signal-to-interference ratio of the signal, and changing the signal-to-noise ratio of the signal; Inputting the transformed data into a neural network for training to obtain a decision network model; and Sending the trained decision network model to the decision network module for drone detection; The method further comprises: Segment the collected drone signals to be identified to obtain multiple continuous signal segments with the same length; Inputting multiple signal segments of the drone signal to be distinguished into the judgment network module for judgment, so as to obtain multiple judgment results for the single signal segment; and Performing sliding window joint decision according to the multiple decision results for the single signal segment to obtain a final decision result; The operation of changing the transmission frequency of the signal includes: For the signal {x[0], x[1], x[2], ..., x[N-1]}, transform it into {y1[0], y1[1], y1[2], ..., y1[N-1]}; Among them, y1[n] satisfies the following equation: Wherein, N is the signal length; f0 is one of the channel frequencies of the signal, and the set of all channel frequencies of the signal is {f0, f1, ..., f M }, where M is the total number of channels; f m is the channel frequency point that needs to be transformed in the data enhancement parameters; f s is the sampling frequency; The operation of changing the signal-to-interference ratio of the signal includes: For the signal y1[n], transform it into y2[n]; Among them, y2[n] satisfies the following equation: in, α is the target signal-to-interference ratio, I[n] is the interference signal; x[n] is the original signal; The operation of changing the signal-to-noise ratio of the signal comprises: For the signal y2[n], transform it to y3[n]: Among them, y3[n] satisfies the following equation: in, β is the target signal-to-noise ratio, P noise is the noise power, The power is P noise is a cyclically symmetric zero-mean complex Gaussian random variable.

2. The method according to claim 1, characterized in that The sliding window joint decision is specifically: Set the sliding window size to K, where K is a natural number greater than or equal to 2; Counting the number Kpos of signal segments determined to contain target UAV signals among K consecutive signal segments; and In response to Kpos being greater than a threshold, it is determined that a target drone exists.

3. A drone detection device based on data-enhanced machine learning, characterized in that: The device comprises: UAV signal acquisition module, which automatically collects the UAV signals to be identified; A signal segment module, connected to the drone signal acquisition module, segments the acquired drone signal to be identified to obtain a plurality of continuous signal segments of the same length; a decision network module, connected to the signal segment module, and configured to make decisions on the multiple segmented signal segments respectively, so as to obtain multiple decision results for the single signal segment; and A sliding window joint decision module, connected to the decision network module, for performing a sliding window joint decision according to the plurality of decision results for a single signal segment to obtain a final decision result; The decision network module includes a trained decision network model, and the decision network model is trained through the following steps: Automatically collect the original signal of the drone; Segmenting the collected original signal of the drone to obtain a plurality of continuous signal segments of the same length; Performing signal transformation on the plurality of signal segments, and using the transformed data as input for neural network training; wherein the signal transformation includes one or more of the following operations: changing the transmission frequency of the signal, changing the signal-to-interference ratio of the signal, and changing the signal-to-noise ratio of the signal; Inputting the transformed data into a neural network for training to obtain the decision network model; The operation of changing the transmission frequency of the signal includes: For the signal {x[0], x[1], x[2], ..., x[N-1]}, transform it into {y1[0], y1[1], y1[2], ..., y1[N-1]}; Among them, y1[n] satisfies the following equation: Wherein, N is the signal length; f0 is one of the channel frequencies of the signal, and the set of all channel frequencies of the signal is {f0, f1, ..., f M }, where M is the total number of channels; f m is the channel frequency point that needs to be transformed in the data enhancement parameters; f s is the sampling frequency; The operation of changing the signal-to-interference ratio of the signal includes: For the signal y1[n], transform it into y2[n]; Among them, y2[n] satisfies the following equation: in, α is the target signal-to-interference ratio, I[n] is the interference signal; x[n] is the original signal; The operation of changing the signal-to-noise ratio of the signal comprises: For the signal y2[n], transform it to y3[n]: Among them, y3[n] satisfies the following equation: in, β is the target signal-to-noise ratio, P noise is the noise power, The power is P noise is a cyclically symmetric zero-mean complex Gaussian random variable.

4. The device according to claim 3, characterized in that The sliding window joint decision is specifically: Set the sliding window size to K, where K is a natural number greater than or equal to 2; Counting the number Kpos of signal segments determined to contain target UAV signals among K consecutive signal segments; and In response to Kpos being greater than a threshold, it is determined that a target drone exists.

5. A drone detection device based on data-enhanced machine learning, characterized in that: The device comprises a memory and a processor; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the method according to any one of claims 1 to 2 according to the instructions in the program code.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.

Images