Interference source detection method based on frequency spectrograph and neural network, unmanned aerial vehicle and medium

Abstract

The invention discloses an interference source detection method based on a frequency spectrograph and a neural network, an unmanned aerial vehicle and a medium, and the method comprises the steps: controlling the unmanned aerial vehicle carrying the frequency spectrograph to inspect a target area, obtaining a frequency spectrum sequence which is collected by the frequency spectrograph and comprises multiple pieces of frequency spectrum information continuously collected according to a time sequence, and enabling a reference point to be a take-off origin of the unmanned aerial vehicle; and inputting the frequency spectrum sequence into a trained interference source detection model in the model module, and outputting an interference source detection result in the target area. The interference source is detected through the unmanned aerial vehicle with the frequency spectrograph and the interference source detection model, and compared with an existing scheme only capable of monitoring a fixed area, model computing power can be used for replacing hardware, on-demand mobility of the unmanned aerial vehicle can be used for replacing fixed coverage, and on the premise that positioning precision is not sacrificed, hardware, deployment and operation and maintenance cost is greatly reduced.

Description

Technical Field

[0001] This application relates to, but is not limited to, the field of wireless communication technology, and in particular to an interference source detection method based on a spectrum analyzer and neural network, an unmanned aerial vehicle (UAV), and a medium. Background Technology

[0002] Traditional methods for troubleshooting interference sources in wireless communication involve manual investigation, which is inaccurate and inefficient. To address this, a joint analysis method combining multi-spectral analyzers and antenna arrays arranged at specific distances has been proposed. However, this method requires significant equipment investment, is inflexible in deployment, may have blind spots, is costly, and cannot guarantee the accuracy of the interference source investigation results. Summary of the Invention

[0003] This application provides an interference source detection method, a drone, and a medium based on a spectrum analyzer and neural network, which ensures the accuracy of interference source detection results while reducing costs.

[0004] In a first aspect, embodiments of this application provide an interference source detection method based on a spectrum analyzer and a neural network, comprising: The system controls a drone equipped with a spectrum analyzer to inspect a target area and obtains a spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectrum information collected sequentially in time. Each spectrum information includes spectrum data, a collection timestamp corresponding to the spectrum data, the position information of the drone relative to a reference point at the collection timestamp, and the antenna orientation information of the spectrum analyzer at the collection timestamp. The drone is also equipped with a model module, and the reference point is the takeoff origin of the drone. The spectrum sequence is input into the interference source detection model trained in the model module, and the interference source detection result in the target region is output. The interference source detection result includes the target interference source type and the target interference source location.

[0005] In some embodiments, the interference source detection model includes a convolutional layer, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer. The spectral sequence is input into the pre-trained interference source detection model in the model module, and the output of the interference source detection result in the target region includes: The first fully connected layer encodes the acquisition timestamp, the location information, and the antenna orientation information in each of the spectrum information to obtain a first vector. The second vector is obtained by encoding the spectral data in each of the spectral information through the convolutional layer. The target vector is obtained by concatenating the first vector and the second vector. The target vector is processed by the Transformer encoder to extract spatiotemporal information; The spatiotemporal information is sequentially input into the convolutional residual network and the second fully connected layer, and multiple candidate interference source types are output. Each interference source type corresponds to a candidate position and an interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. The candidate interference source type corresponding to the highest probability of the interference source is determined as the target interference source type, and the candidate position corresponding to the target interference source type is determined as the target interference source position.

[0006] In some embodiments, the interference source detection model is trained according to the following steps: Obtain the training dataset; A joint constraint function is constructed by weighted summation of the cross-entropy loss function and the mean squared error loss function, wherein the cross-entropy loss function corresponds to the probability of any interference source of any interference source type, and the mean squared error loss function corresponds to the candidate position of any interference source type. The initial model is trained based on the training dataset and the joint constraint function to obtain the interference source detection model.

[0007] In some embodiments, obtaining the training dataset includes: A clean signal source is output using a first signal generator, wherein the clean signal source includes a 4G communication signal or a 5G communication signal; The second signal generator outputs an interference signal source, wherein the interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using the CMDA access method; The clean signal source and the interference signal source are subjected to attenuation, mixing and remote simulation processing to obtain the training dataset. The training dataset includes multiple basic signals, each of which carries a measurement point location and a label. The label includes the interference source type and the position of the interference source relative to the reference point.

[0008] Secondly, embodiments of this application provide a drone, including: A data acquisition unit is used to control a drone equipped with a spectrum analyzer to inspect a target area and obtain a spectrum sequence acquired by the spectrum analyzer. The spectrum sequence includes multiple spectrum information acquired sequentially in time. Each spectrum information includes spectrum data, a timestamp corresponding to the spectrum data, the position information of the drone relative to a reference point at the timestamp, and the antenna orientation information of the spectrum analyzer at the timestamp. The drone is also equipped with a model module, and the reference point is the takeoff origin of the drone. The interference source detection unit is used to input the spectrum sequence into the interference source detection model trained in the model module and output the interference source detection result in the target region, wherein the interference source detection result includes the target interference source type and the target interference source location.

[0009] In some embodiments, the interference source detection model includes a convolutional layer, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer; the interference source detection unit includes: The first data processing unit is configured to encode the acquisition timestamp, the location information, and the antenna orientation information in each of the spectrum information through the first fully connected layer to obtain a first vector; The second data processing unit is used to encode the spectral data in each of the spectral information through the convolutional layer to obtain a second vector; The third data processing unit is used to concatenate the first vector and the second vector to obtain the target vector; The fourth data processing unit is used to process the target vector through the Transformer encoder to extract spatiotemporal information; The fifth data processing unit is used to sequentially input the spatiotemporal information into the convolutional residual network and the second fully connected layer, and output multiple candidate interference source types. Each interference source type corresponds to a candidate position and an interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. The sixth data processing unit is used to determine the candidate interference source type corresponding to the highest probability of the interference source as the target interference source type, and to determine the candidate position corresponding to the target interference source type as the target interference source position.

[0010] In some embodiments, the interference source detection unit further includes: The data acquisition unit is used to acquire the training dataset. The constraint function construction unit is used to construct a joint constraint function based on the weighted summation of the cross-entropy loss function and the mean square error loss function, wherein the cross-entropy loss function corresponds to the probability of any interference source of any interference source type, and the mean square error loss function corresponds to the candidate position of any interference source type. The model training unit is used to train the initial model based on the training dataset and the joint constraint function to obtain the interference source detection model.

[0011] In some embodiments, the data acquisition unit includes: The first sub-data collection unit is used to output a clean signal source using a first signal generator, wherein the clean signal source includes a 4G communication signal or a 5G communication signal. The second sub-data collection unit is used to output an interference signal source using a second signal generator, wherein the interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using CMDA access. The seventh data processing unit is used to perform attenuation, mixing, and distance simulation processing on the clean signal source and the interference signal source to obtain the training dataset. The training dataset includes multiple basic signals, each of which carries a measurement point location and a label. The label includes the interference source type and the position of the interference source relative to the reference point.

[0012] Thirdly, embodiments of this application provide a drone, including at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor, which, when executed by the at least one control processor, enable the at least one control processor to perform the interference source detection method based on a spectrum analyzer and neural network as described in the first aspect.

[0013] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer-executable instructions for performing the interference source detection method based on a spectrum analyzer and neural network as described in the first aspect.

[0014] This application provides an interference source detection method, a drone, and a medium based on a spectrum analyzer and a neural network. The method includes: controlling a drone equipped with a spectrum analyzer to inspect a target area and obtaining a spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectrum information collected sequentially in time. Each spectrum information includes spectrum data, a collection timestamp corresponding to the spectrum data, the position information of the drone relative to a reference point at the collection timestamp, and the antenna orientation information of the spectrum analyzer at the collection timestamp. The drone is also equipped with a model module, and the reference point is the takeoff origin of the drone. The spectrum sequence is input into a pre-trained interference source detection model in the model module, and the interference source detection result in the target area is output. The interference source detection result includes the type and location of the target interference source. According to the solution provided in the embodiments of this application, interference sources are detected by using a drone equipped with a spectrum analyzer and a preset interference source detection model. Compared with the existing interference source detection solution that can only monitor fixed areas by combining multiple spectrum analyzers and fixed antenna arrays, this solution can replace hardware with model computing power and fixed coverage with the on-demand mobility of the drone. Without sacrificing positioning accuracy, it significantly reduces hardware, deployment and maintenance costs. Attached Figure Description

[0015] Figure 1 This is a flowchart of the steps of an interference source detection method based on a spectrum analyzer and neural network provided in one embodiment of this application; Figure 2 This is a schematic diagram of a drone module provided in another embodiment of this application; Figure 3 This is a structural diagram of a drone provided in another embodiment of this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0017] It is understandable that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, or the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0018] Traditional methods for troubleshooting interference sources in wireless communication involve manual investigation, which is inaccurate and inefficient. To address this, a joint analysis method combining multi-spectral analyzers and antenna arrays arranged at specific distances has been proposed. However, this method requires significant equipment investment, is inflexible in deployment, may have blind spots, is costly, and cannot guarantee the accuracy of the interference source investigation results.

[0019] To address the aforementioned problems, this application provides an interference source detection method, a drone, and a medium based on a spectrum analyzer and a neural network. The method includes: controlling a drone equipped with a spectrum analyzer to inspect a target area and obtaining a spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectral information items collected sequentially over time. Each spectral information item includes spectrum data, a timestamp corresponding to the spectrum data, the drone's position relative to a reference point at the timestamp, and the antenna orientation information of the spectrum analyzer at the timestamp. The drone is also equipped with a model module, and the reference point is the drone's takeoff origin. The spectrum sequence is input into a pre-trained interference source detection model in the model module, and the interference source detection result in the target area is output. The interference source detection result includes the target interference source type and the target interference source location. According to the solution provided in the embodiments of this application, interference sources are detected by using a drone equipped with a spectrum analyzer and a preset interference source detection model. Compared with the existing interference source detection solution that can only monitor fixed areas by combining multiple spectrum analyzers and fixed antenna arrays, this solution can replace hardware with model computing power and fixed coverage with the on-demand mobility of the drone. Without sacrificing positioning accuracy, it significantly reduces hardware, deployment and maintenance costs.

[0020] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0021] refer to Figure 1 , Figure 1 This is a flowchart illustrating the steps of an interference source detection method based on a spectrum analyzer and neural network according to an embodiment of this application. This application provides an interference source detection method based on a spectrum analyzer and neural network, which includes, but is not limited to, the following steps: Step S10: Control the UAV equipped with a spectrum analyzer to inspect the target area and obtain the spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectrum information collected continuously in chronological order. Each spectrum information includes spectrum data, the acquisition timestamp corresponding to the spectrum data, the position information of the UAV at the acquisition timestamp relative to the reference point, and the antenna orientation information of the spectrum analyzer at the acquisition timestamp. The UAV is also equipped with a model module, and the reference point is the takeoff origin of the UAV.

[0022] It is understood that this embodiment controls a drone equipped with a spectrum analyzer to conduct inspections in the target area, while simultaneously collecting spectrum data in real time to form a spectrum sequence, thereby providing an effective data foundation for the drone to use the interference source detection model built into the model module for subsequent interference source detection.

[0023] Specifically, the antenna orientation information includes the angle relative to due north, the angle relative to due east, and the angle relative to due up; any spectrum data includes multiple sets of data including frequency points, amplitude, and phase, wherein the frequency point intervals are 1k, 2k, 5k, 1m, 10m, 20m, or 50m.

[0024] Step S20: Input the spectrum sequence into the trained interference source detection model in the model module, and output the interference source detection result in the target region. The interference source detection result includes the target interference source type and the target interference source location.

[0025] Specifically, in some embodiments, the interference source detection model includes a convolutional layer, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer. Figure 1 Step S20 includes, but is not limited to, the following steps: Step S21: Encode the acquisition timestamp, location information, and antenna orientation information in each spectrum information through the first fully connected layer to obtain the first vector; Step S22: Encode the spectral data in each spectral information through a convolutional layer to obtain the second vector; Step S23: Concatenate the first vector and the second vector to obtain the target vector; Step S24: Process the target vector using a Transformer encoder to extract spatiotemporal information; Step S25: The spatiotemporal information is sequentially input into the convolutional residual network and the second fully connected layer, and multiple candidate interference source types are output. Each interference source type has a candidate position and an interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. Step S26: Determine the candidate interference source type corresponding to the interference source with the highest numerical interference source probability as the target interference source type, and determine the candidate position corresponding to the target interference source type as the target interference source position.

[0026] It is understood that in this embodiment, the spectrum sequence input to the interference source detection model includes multiple temporally continuous spectrum information. The spectrum information collected at each different time includes one-dimensional data and two-dimensional data. The one-dimensional data includes the collection timestamp, location information, and antenna orientation information. The dimensions of these information are fixed in each frame, so they can be encoded by the first fully connected layer to obtain the corresponding first vector. For the two-dimensional data, i.e., the spectrum data collected by the spectrum analyzer, a convolutional layer is needed to receive and process it to obtain the corresponding second vector. Finally, the first vector and the second vector corresponding to the same collection timestamp are concatenated to obtain the target vector, thus completing the reception of the spectrum information of a single frame. Then, through a multi-head attention mechanism and a transformer... The Transformer encoder in the encoding layer processes the target vector to extract spatiotemporal information. This spatiotemporal information is then sequentially input into a convolutional residual network and a second fully connected layer (i.e., the skeleton network layer) for multi-layer extraction, focusing on both the overall and local details of the spatiotemporal information. Multiple sets of data are output, each including candidate interference source types, corresponding candidate locations, and interference source probabilities. The candidate location is the position of the interference source associated with the corresponding interference source type relative to the reference point. Next, the candidate interference source type corresponding to the highest interference source probability is determined as the target interference source type, and the candidate location corresponding to the target interference source type is determined as the target interference source location. Thus, this embodiment of the application achieves interference source detection using a UAV equipped with a spectrum analyzer and a pre-set interference source detection model. Compared to existing interference source detection schemes that can only monitor fixed areas and combine multiple spectrum analyzers and fixed antenna arrays, this approach trades hardware for model computing power and fixed coverage for on-demand mobility of the UAV, significantly reducing hardware, deployment, and maintenance costs without sacrificing positioning accuracy.

[0027] In addition, in some embodiments, the training method of the interference source detection model of this application includes, but is not limited to, the following steps: Step S31: Obtain the training dataset; Step S32: Construct a joint constraint function based on the weighted summation of the cross-entropy loss function and the mean square error loss function, where the cross-entropy loss function corresponds to the probability of any type of interference source, and the mean square error loss function corresponds to the candidate position of any type of interference source. Step S33: Train the initial model based on the training dataset and joint constraint function to obtain the interference source detection model.

[0028] It is understood that in this embodiment, the interference source detection model trained and put into use is constructed according to the following methods: obtaining a training dataset; constructing a joint constraint function based on the weighted summation of the cross-entropy loss function and the mean square error loss function, wherein the cross-entropy loss function corresponds to the interference source probability of any interference source type, and the mean square error loss function corresponds to the candidate position of any interference source type. The candidate position, that is, the position of the interference source relative to the takeoff origin (reference point), is a three-dimensional position information. Therefore, in the training process, this embodiment converts the candidate position into calculating the Euclidean distance between the interference source and the reference point, using the Euclidean distance as input, and then using the mean square error loss function as the loss function. The interference source probability is used to indicate the probability that the currently detected interference source is a certain interference source type.

[0029] Specifically, in this embodiment, considering that the interference type must be accurate for the corresponding location information to be meaningful, more than 70% of the weight is allocated to the cross-entropy loss function associated with the probability of the interference source in the joint constraint function.

[0030] Specifically, this embodiment utilizes a supervised learning method, combining a training dataset and a joint constraint function to train the model and the joint constraint function.

[0031] Specifically, in some embodiments, step S31 includes, but is not limited to, the following steps: Step S311: Use the first signal generator to output a clean signal source, wherein the clean signal source includes a 4G communication signal or a 5G communication signal; Step S312: The second signal generator outputs an interference signal source, wherein the interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using the CMDA access method. Step S313: Perform attenuation, mixing, and distance simulation processing on the clean signal source and the interference signal source to obtain a training dataset. The training dataset includes multiple basic signals, each carrying the measurement point location and label. The label includes the interference source type and the position of the interference source relative to the reference point.

[0032] It is understood that the frequency harmonic interference signal of the interference signal source in this embodiment is a continuous frequency harmonic signal interference (such as transformer, construction site tower crane radio, FM, AM radio, etc.), multipath propagation interference signal (such as multipath propagation interference caused by base station signals due to diffraction, reflection, etc.), and communication interference signals using CMDA access method include privately set 4G or 5G signal sources, different network 3G, 4G, 5G signals, Wi-Fi 7 signals, etc. In addition, it may also include co-frequency spoofing signaling, airport radar signals, weather radar signals, maritime radar signals, and repeater interference signals emitted by repeaters illegally used in urban villages and other areas.

[0033] Thus, this embodiment constructs a training dataset by mixing clean and interference signals, and each sample is accompanied by a precise label, including the type of interference and the location of the interference source, which can cover diverse interference patterns and provide an effective data foundation for improving model robustness and the accuracy of interference source detection.

[0034] Additionally, in some embodiments, such as Figure 2 As shown in the embodiments, this application also discloses a drone 200, which includes: The data acquisition unit is used to control the UAV equipped with a spectrum analyzer to inspect the target area and obtain the spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectrum information collected continuously in time order. Each spectrum information includes spectrum data, the acquisition timestamp corresponding to the spectrum data, the position information of the UAV relative to the reference point at the acquisition timestamp, and the antenna orientation information of the spectrum analyzer at the acquisition timestamp. The UAV is also equipped with a model module, and the reference point is the takeoff origin of the UAV. The interference source detection unit is used to input the spectrum sequence into the interference source detection model trained in the model module and output the interference source detection results in the target region. The interference source detection results include the type and location of the target interference source.

[0035] In some embodiments, the interference source detection model includes a convolutional layer, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer. The interference source detection unit includes: The first data processing unit is used to encode the acquisition timestamp, location information and antenna orientation information in each spectrum information through the first fully connected layer to obtain the first vector; The second data processing unit is used to encode and process the spectral data in each spectral information through a convolutional layer to obtain the second vector. The third data processing unit is used to concatenate the first vector and the second vector to obtain the target vector; The fourth data processing unit is used to process the target vector through the Transformer encoder to extract spatiotemporal information; The fifth data processing unit is used to sequentially input spatiotemporal information into the convolutional residual network and the second fully connected layer, and output multiple candidate interference source types. Each interference source type has a corresponding candidate position and interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. The sixth data processing unit is used to determine the candidate interference source type corresponding to the highest interference source probability as the target interference source type, and to determine the candidate position corresponding to the target interference source type as the target interference source position.

[0036] In some embodiments, the interference source detection unit further includes: The data acquisition unit is used to acquire the training dataset. The constraint function construction unit is used to construct a joint constraint function based on the weighted summation of the cross-entropy loss function and the mean square error loss function. The cross-entropy loss function corresponds to the probability of any interference source type, and the mean square error loss function corresponds to the candidate position of any interference source type. The model training unit is used to train the initial model based on the training dataset and joint constraint functions to obtain the interference source detection model.

[0037] In some embodiments, the data acquisition unit includes: The first sub-data collection unit is used to output a clean signal source using the first signal generator, wherein the clean signal source includes 4G communication signal or 5G communication signal; The second sub-data collection unit is used to output an interference signal source using the second signal generator. The interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using the CMDA access method. The seventh data processing unit is used to perform attenuation, mixing, and distance simulation processing on clean signal sources and interference signal sources to obtain a training dataset. The training dataset includes multiple basic signals, each carrying the location of a measurement point and a label. The label includes the type of interference source and the position of the interference source relative to the reference point.

[0038] It should be noted that the specific implementation of the UAV 200 is basically the same as the specific implementation of the interference source detection method based on the spectrum analyzer and neural network described above, and will not be repeated here.

[0039] like Figure 3 As shown, Figure 3 This is a structural diagram of a control device provided in one embodiment of this application. The present invention also provides a control device 300, comprising: The processor 310 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 320 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 320 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 320 and is called and executed by the processor 310 to implement the interference source detection method based on spectrum analyzer and neural network of the embodiments of this application. Input / output interface 330 is used to realize information input and output; The communication interface 340 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 350 transmits information between various components of the device (e.g., processor 310, memory 320, input / output interface 330, and communication interface 340); The processor 310, memory 320, input / output interface 330 and communication interface 340 are connected to each other within the device via bus 350.

[0040] In addition, this application also provides an electronic device, including the control device 300 described in the above embodiments.

[0041] In addition, this application embodiment also provides a storage medium, which is a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the above-described interference source detection method based on a spectrum analyzer and a neural network.

[0042] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof. The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separate, and may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0043] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0044] The above provides a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A method for detecting interference sources based on a spectrum analyzer and a neural network, characterized in that, include: The system controls a drone equipped with a spectrum analyzer to inspect a target area and obtains a spectrum sequence collected by the spectrum analyzer. The spectrum sequence includes multiple spectrum information collected sequentially in time. Each spectrum information includes spectrum data, a collection timestamp corresponding to the spectrum data, the position information of the drone relative to a reference point at the collection timestamp, and the antenna orientation information of the spectrum analyzer at the collection timestamp. The drone is also equipped with a model module, and the reference point is the takeoff origin of the drone. The spectrum sequence is input into the interference source detection model trained in the model module, and the interference source detection result in the target region is output. The interference source detection result includes the target interference source type and the target interference source location.

2. The interference source detection method based on a spectrum analyzer and neural network according to claim 1, characterized in that, The interference source detection model includes convolutional layers, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer. The spectral sequence is input into the pre-trained interference source detection model in the model module, and the output is the interference source detection result in the target region, including: The first fully connected layer encodes the acquisition timestamp, the location information, and the antenna orientation information in each of the spectrum information to obtain a first vector. The second vector is obtained by encoding the spectral data in each of the spectral information through the convolutional layer. The target vector is obtained by concatenating the first vector and the second vector. The target vector is processed by the Transformer encoder to extract spatiotemporal information; The spatiotemporal information is sequentially input into the convolutional residual network and the second fully connected layer, and multiple candidate interference source types are output. Each interference source type corresponds to a candidate position and an interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. The candidate interference source type corresponding to the highest interference source probability is determined as the target interference source type, and the candidate position corresponding to the target interference source type is determined as the target interference source position.

3. The interference source detection method based on a spectrum analyzer and neural network according to claim 2, characterized in that, The interference source detection model is trained according to the following steps: Obtain the training dataset; A joint constraint function is constructed by weighted summation of the cross-entropy loss function and the mean squared error loss function, wherein the cross-entropy loss function corresponds to the probability of any interference source of any interference source type, and the mean squared error loss function corresponds to the candidate position of any interference source type. The initial model is trained based on the training dataset and the joint constraint function to obtain the interference source detection model.

4. The interference source detection method based on a spectrum analyzer and neural network according to claim 3, characterized in that, Obtain the training dataset, including: A clean signal source is output using a first signal generator, wherein the clean signal source includes a 4G communication signal or a 5G communication signal; The second signal generator outputs an interference signal source, wherein the interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using the CMDA access method; The clean signal source and the interference signal source are subjected to attenuation, mixing and remote simulation processing to obtain the training dataset. The training dataset includes multiple basic signals, each of which carries a measurement point location and a label. The label includes the interference source type and the position of the interference source relative to the reference point.

5. A drone, characterized in that, include: A data acquisition unit is used to control a drone equipped with a spectrum analyzer to inspect a target area and obtain a spectrum sequence acquired by the spectrum analyzer. The spectrum sequence includes multiple spectrum information acquired sequentially in time. Each spectrum information includes spectrum data, a timestamp corresponding to the spectrum data, the position information of the drone relative to a reference point at the timestamp, and the antenna orientation information of the spectrum analyzer at the timestamp. The drone is also equipped with a model module, and the reference point is the takeoff origin of the drone. The interference source detection unit is used to input the spectrum sequence into the interference source detection model trained in the model module and output the interference source detection result in the target region, wherein the interference source detection result includes the target interference source type and the target interference source location.

6. The UAV according to claim 5, characterized in that, The interference source detection model includes a convolutional layer, a first fully connected layer, a Transformer encoder, a convolutional residual network, and a second fully connected layer. The interference source detection unit includes: The first data processing unit is configured to encode the acquisition timestamp, the location information, and the antenna orientation information in each of the spectrum information through the first fully connected layer to obtain a first vector; The second data processing unit is used to encode the spectral data in each of the spectral information through the convolutional layer to obtain a second vector; The third data processing unit is used to concatenate the first vector and the second vector to obtain the target vector; The fourth data processing unit is used to process the target vector through the Transformer encoder to extract spatiotemporal information; The fifth data processing unit is used to sequentially input the spatiotemporal information into the convolutional residual network and the second fully connected layer, and output multiple candidate interference source types. Each interference source type corresponds to a candidate position and an interference source probability. The candidate position is the position of the interference source associated with the corresponding interference source type relative to the reference point. The sixth data processing unit is used to determine the candidate interference source type corresponding to the highest probability of the interference source as the target interference source type, and to determine the candidate position corresponding to the target interference source type as the target interference source position.

7. The UAV according to claim 6, characterized in that, The interference source detection unit further includes: The data acquisition unit is used to acquire the training dataset. The constraint function construction unit is used to construct a joint constraint function based on the weighted summation of the cross-entropy loss function and the mean square error loss function, wherein the cross-entropy loss function corresponds to the probability of any interference source of any interference source type, and the mean square error loss function corresponds to the candidate position of any interference source type. The model training unit is used to train the initial model based on the training dataset and the joint constraint function to obtain the interference source detection model.

8. The UAV according to claim 7, characterized in that, The data acquisition unit includes: The first sub-data collection unit is used to output a clean signal source using a first signal generator, wherein the clean signal source includes a 4G communication signal or a 5G communication signal. The second sub-data collection unit is used to output an interference signal source using a second signal generator, wherein the interference signal source includes at least a frequency harmonic interference signal, a multipath propagation interference signal, and a communication interference signal using CMDA access. The seventh data processing unit is used to perform attenuation, mixing, and distance simulation processing on the clean signal source and the interference signal source to obtain the training dataset. The training dataset includes multiple basic signals, each of which carries a measurement point location and a label. The label includes the interference source type and the position of the interference source relative to the reference point.

9. A drone, characterized in that, It includes at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor, which, when executed by the at least one control processor, enable the at least one control processor to perform the interference source detection method based on a spectrum analyzer and neural network as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to perform the interference source detection method based on a spectrum analyzer and neural network as described in any one of claims 1 to 4.

Images