Signal processing method and system of reconnaissance unmanned aerial vehicle

Abstract

The invention relates to the field of signal processing, and provides a signal processing method and system for a reconnaissance unmanned aerial vehicle, and the method comprises the steps: carrying out the signal decomposition of a first intermediate frequency signal obtained through the frequency mixing of an unmanned aerial vehicle signal, so as to obtain a set of signal component intrinsic mode functions; a signal analysis and processing algorithm based on deep learning is adopted to carry out waveform semantic feature extraction on the intrinsic mode function of each signal component, so that a filtered first intermediate frequency signal is automatically generated based on the significant dynamic aggregation representation of the waveform semantic feature of each signal component in a global domain; the method can dynamically adapt to the change of a signal environment and extract more representative features so as to improve the flexibility of signal processing. Compared with a traditional fixed parameter filter, the method can better deal with interference and noise in a complex environment, enables the final signal to be clearer, and improves the signal processing quality and automation level of the reconnaissance unmanned aerial vehicle.

Description

Technical Field

[0001] The present application relates to the field of signal processing, and more specifically, to a signal processing method and system for a reconnaissance drone. Background Art

[0002] With the rapid development of UAV technology, UAVs have been widely used in military reconnaissance, environmental monitoring, disaster relief and other fields. As one of the key technologies to ensure the realization of its functions, UAV signal processing plays a vital role in reconnaissance missions. In reconnaissance missions, the quality of the signal directly affects the accuracy and reliability of the data, and thus affects the effectiveness of decision-making. High-quality signal processing can ensure that the real-time transmitted images and data are clear and accurate, providing a reliable basis for command decision-making.

[0003] In traditional reconnaissance drone signal processing methods, fixed-parameter filters or preset thresholds are usually used to remove noise and unnecessary frequency components. However, in complex and changing signal environments, these filters are difficult to adapt to dynamic changes, resulting in poor signal processing. Once the signal environment changes, such as the emergence of new interference sources or changes in signal characteristics, the filter cannot effectively remove noise, but may lead to the loss of useful signals. In addition, relying on preset feature templates or thresholds makes them inflexible when facing diverse signal features and unable to capture subtle changes in the signal, especially in nonlinear or non-stationary signals, which often contain important information.

[0004] Therefore, an optimized signal processing solution for reconnaissance UAVs is needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides a signal processing method and system for a reconnaissance drone.

[0006] According to one aspect of the present application, a signal processing method for a reconnaissance drone is provided, comprising:

[0007] Acquire the drone signal collected by the antenna;

[0008] Preselecting the drone signal through a bandpass filter, and inputting the preselected drone signal into a low noise amplifier to obtain an amplified drone signal;

[0009] Performing a first mixing of the amplified UAV signal and a local oscillator signal generated by a first phase-locked loop to obtain a first intermediate frequency signal;

[0010] Performing filtering and amplification processing on the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal;

[0011] Performing a second mixing of the enhanced first intermediate frequency signal and a local oscillator signal generated by a second phase-locked loop to obtain a second intermediate frequency signal;

[0012] Performing filtering and amplification processing on the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal;

[0013] Performing a third mixing of the enhanced second intermediate frequency signal and a local oscillator signal generated by a direct digital synthesizer to obtain a third intermediate frequency signal;

[0014] The third intermediate frequency signal is input into an analog-to-digital converter to obtain a digital signal.

[0015] According to another aspect of the present application, a signal processing system for a reconnaissance drone is provided, comprising:

[0016] A drone signal data collection module is used to obtain drone signals collected by an antenna;

[0017] A drone signal preselection and amplification module, used to preselect the drone signal through a bandpass filter, and input the preselected drone signal into a low noise amplifier to obtain an amplified drone signal;

[0018] A drone signal primary mixing module, used for performing a first mixing of the amplified drone signal with a local oscillator signal generated by a first phase-locked loop to obtain a first intermediate frequency signal;

[0019] A first intermediate frequency signal enhancement module, configured to perform filtering and amplification processing on the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal;

[0020] A first enhanced intermediate frequency signal mixing module, used for performing a second mixing on the enhanced first intermediate frequency signal and a local oscillator signal generated by a second phase-locked loop to obtain a second intermediate frequency signal;

[0021] A second intermediate frequency signal enhancement module, configured to perform filtering and amplification processing on the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal;

[0022] A third intermediate frequency signal building module, used for performing a third mixing of the enhanced second intermediate frequency signal and the local oscillator signal generated by the direct digital synthesizer to obtain a third intermediate frequency signal;

[0023] The third intermediate frequency signal analog-to-digital conversion module is used to input the third intermediate frequency signal into an analog-to-digital converter to obtain a digital signal.

[0024] This application has significant technical effects due to the adoption of the above technical solutions:

[0025] Compared with the prior art, the signal processing method and system for the reconnaissance UAV provided in the present application performs signal decomposition on the first intermediate frequency signal obtained by mixing the UAV signal to obtain a set of signal component intrinsic mode functions, and uses a signal analysis and processing algorithm based on deep learning to extract waveform semantic features of the intrinsic mode functions of each signal component, thereby automatically generating a filtered first intermediate frequency signal based on the significant dynamic aggregation representation of the waveform semantic features of each signal component in the entire domain, and can dynamically adapt to changes in the signal environment and extract more representative features to improve the flexibility of signal processing. And compared with traditional fixed parameter filters, it can better cope with interference and noise in complex environments, making the final signal clearer, thereby improving the quality and automation level of reconnaissance UAV signal processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0027] Figure 1 Flow chart of a signal processing method for a reconnaissance drone according to an embodiment of the present application.

[0028] Figure 2 Schematic diagram of data flow of a signal processing method for a reconnaissance drone according to an embodiment of the present application.

[0029] Figure 3 This is a flowchart of filtering and amplifying the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal in a signal processing method for a reconnaissance drone according to an embodiment of the present application.

[0030] Figure 4 This is a flowchart of filtering the first intermediate frequency signal to obtain a filtered first intermediate frequency signal in a signal processing method for a reconnaissance drone according to an embodiment of the present application.

[0031] Figure 5 The present invention is a flowchart of performing global significant aggregation of feature distribution fields on the set of semantic feature vectors of the signal component waveforms in the signal processing method of the reconnaissance drone according to an embodiment of the present application to obtain a global significant aggregation representation of the signal component waveforms.

[0032] Figure 6 4 is a system block diagram of a signal processing system for a reconnaissance drone according to an embodiment of the present application.

[0033] Figure 7Schematic diagram of the overall design of a drone detection module according to other embodiments of the present application. DETAILED DESCRIPTION

[0034] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0035] With the rapid development of drone technology, drones have been widely used in many fields such as reconnaissance missions, environmental monitoring and disaster relief. Among them, drone signal processing, as one of the key technologies to ensure the realization of its various functions, plays a vital role in reconnaissance missions. For military reconnaissance, signal quality is related to the accuracy and reliability of data, which in turn affects the effectiveness of decision-making. High-quality signal processing can ensure that the real-time transmitted images and data are clear and accurate, providing a solid and reliable basis for command decision-making.

[0036] In traditional reconnaissance drone signal processing methods, fixed parameter filters or pre-set thresholds are often used to remove noise and unnecessary frequency components. However, in a complex signal environment, such filters are difficult to adapt to dynamic changes, resulting in poor signal processing results. In other words, once the signal environment changes, such as when a new interference source is generated or the signal characteristics change, the filter will not be able to effectively remove the noise, and may even cause the loss of useful signals. In addition, relying on pre-set feature templates or thresholds makes them lack sufficient flexibility in dealing with diverse signal features, and unable to capture subtle changes in the signal, especially in nonlinear or non-stationary signals, which often contain important information.

[0037] Based on this, this application proposes an optimized signal processing method for reconnaissance drones. Figure 1 Flow chart of a signal processing method for a reconnaissance drone according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the signal processing method of the reconnaissance drone according to the embodiment of the present application. Figure 1 and Figure 2As shown, the signal processing method of the reconnaissance drone according to the embodiment of the present application includes: S110, acquiring the drone signal collected by the antenna; S120, preselecting the drone signal through a bandpass filter, and inputting the preselected drone signal into a low noise amplifier to obtain an amplified drone signal; S130, mixing the amplified drone signal with the local oscillator signal generated by the first phase-locked loop for the first time to obtain a first intermediate frequency signal; S140, filtering and amplifying the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal; S150, mixing the enhanced first intermediate frequency signal with the local oscillator signal generated by the second phase-locked loop for the second time to obtain a second intermediate frequency signal; S160, filtering and amplifying the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal; S170, mixing the enhanced second intermediate frequency signal with the local oscillator signal generated by the direct digital synthesizer for the third time to obtain a third intermediate frequency signal; S180, inputting the third intermediate frequency signal into an analog-to-digital converter to obtain a digital signal.

[0038] In step S110, the drone signal collected by the antenna is obtained. Specifically, the antenna is first used to receive the electromagnetic signal generated by the drone during flight. The obtained signal usually contains target information, environmental noise and some interference signals.

[0039] In step S120, the drone signal is preselected through a bandpass filter, and the preselected drone signal is input into a low noise amplifier to obtain an amplified drone signal. Specifically, the bandpass filter is used to select signals within a specific frequency band and remove noise below and above the frequency band. It can ensure that the subsequently processed signals are mainly concentrated in the frequency range of interest, thereby improving the signal-to-noise ratio of the signal. Among them, pay attention to selecting a suitable center frequency and bandwidth, which is conducive to effectively filtering useless signals. Considering that the signal passing through the bandpass filter is usually very weak, it needs to be amplified by a low noise amplifier (LNA). The design of the LNA is to minimize the introduction of additional noise to maintain the clarity of the signal. The amplified drone signal will have a higher amplitude, which is convenient for subsequent processing.

[0040] In step S130, the amplified drone signal is mixed for the first time with the local oscillator signal generated by the first phase-locked loop to obtain a first intermediate frequency signal. Specifically, a phase-locked loop (PLL) is a circuit that can generate a stable frequency and phase. The PLL output frequency range is wide, the frequency is stable, and there is less spurious. By mixing the local oscillator signal generated by the first PLL with the amplified drone signal, the signal is converted from the original frequency to the intermediate frequency (IF) range. Intermediate frequency processing helps to further reduce noise and interference, and facilitates subsequent filtering and amplification processing.

[0041] In step S140, the first intermediate frequency signal is filtered and amplified to obtain an enhanced first intermediate frequency signal. Specifically, Figure 3 This is a flow chart of filtering and amplifying the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal in a signal processing method for a reconnaissance drone according to an embodiment of the present application. Figure 3 As shown, filtering and amplifying the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal includes: S141, filtering the first intermediate frequency signal to obtain a filtered first intermediate frequency signal; S142, amplifying the filtered first intermediate frequency signal to obtain the enhanced first intermediate frequency signal.

[0042] In step S141, the first intermediate frequency signal is filtered and amplified to obtain an enhanced first intermediate frequency signal. Specifically, in order to remove unnecessary frequency components in the first intermediate frequency signal, the first intermediate frequency signal needs to be filtered, and the first intermediate frequency signal after filtering eliminates unnecessary noise interference.

[0043] Specifically, Figure 4 This is a flow chart of filtering the first intermediate frequency signal to obtain a filtered first intermediate frequency signal in a signal processing method for a reconnaissance drone according to an embodiment of the present application. Figure 4 As shown, filtering the first intermediate frequency signal to obtain a filtered first intermediate frequency signal, including: S1411, performing signal decomposition based on empirical mode decomposition technology on the first intermediate frequency signal to obtain a set of signal component intrinsic mode functions; S1412, performing signal component waveform feature extraction on each signal component intrinsic mode function in the set of signal component intrinsic mode functions to obtain a set of signal component waveform semantic feature vectors; S1413, performing global significant aggregation of feature distribution fields on the set of signal component waveform semantic feature vectors to obtain a signal component waveform global significant aggregation representation; S1414, obtaining the filtered first intermediate frequency signal based on the signal component waveform global significant aggregation representation.

[0044] In step S1411, the first intermediate frequency signal is subjected to signal decomposition based on the empirical mode decomposition technology to obtain a set of signal component intrinsic mode functions. Accordingly, considering that the first intermediate frequency signal exhibits different characteristics at different local times, which contains key characteristic components but has noise components, in order to more centrally analyze and process the features in different signal region segments in a more detailed manner, in the technical solution of the present application, the first intermediate frequency signal is subjected to signal decomposition based on the empirical mode decomposition technology to obtain a set of signal component intrinsic mode functions, which can decompose the complex signal into a set of relatively simple intrinsic mode functions, each of which represents the fluctuation of the signal within a specific frequency range, and provides a basis for subsequent feature extraction from the signal.

[0045] In step S1412, signal component waveform feature extraction is performed on each signal component intrinsic modal function in the set of signal component intrinsic modal functions to obtain a set of signal component waveform semantic feature vectors. Specifically, in an embodiment of the present application, signal component waveform feature extraction is performed on each signal component intrinsic modal function in the set of signal component intrinsic modal functions to obtain a set of signal component waveform semantic feature vectors, including: inputting each signal component intrinsic modal function in the set of signal component intrinsic modal functions into a signal component waveform feature extractor based on a 1D-CNN model to obtain a set of signal component waveform semantic feature vectors. Specifically, each signal component intrinsic modal function in the set of signal component intrinsic modal functions contains complex waveform feature information, and there is semantic information of waveform features in different local ranges. Therefore, in the technical solution of the present application, each signal component intrinsic modal function in the set of signal component intrinsic modal functions is input into a signal component waveform feature extractor based on a 1D-CNN model to effectively identify and capture the local waveform semantic information in each function, and obtain a set of signal component waveform semantic feature vectors.

[0046] In step S1413, the set of the signal component waveform semantic feature vectors is subjected to global significant aggregation of the feature distribution field to obtain a signal component waveform global significant aggregation representation. Accordingly, it is considered that the importance and interaction relationship of each signal component waveform semantic feature in the set of the signal component waveform semantic feature vectors in the entire time domain are different. Therefore, in order to focus on and significantly highlight the key features to achieve global optimization and aggregation of signal features, in the technical solution of the present application, the set of the signal component waveform semantic feature vectors is subjected to global significant aggregation of the feature distribution field to obtain a signal component waveform global significant aggregation representation vector and use it as a signal component waveform global significant aggregation representation.

[0047] In detail, first, the static energy factor of each signal component waveform semantic feature vector is calculated to obtain a set of signal component waveform semantic static energy factors. Specifically, the static energy factor can be regarded as an indicator to measure the strength or significance of the feature vector, that is, by evaluating the weight and role of each feature vector in the data based on the metric function, so as to provide data support for the selection of subsequent aggregation centers. Then, based on the set of signal component waveform semantic static energy factors, the set of signal component waveform semantic feature vectors is subjected to force field-guided feature dynamic aggregation to obtain a signal component waveform global significant aggregation representation vector. Specifically, the force field is the interaction force (attraction or repulsion) between objects in physics that determines their position and movement. In signal processing, by simulating such a force field, the feature vectors can find the optimal layout in the high-dimensional space, so as to better reflect the intrinsic connection between them. This processing not only takes into account the spatial relationship of each feature, but also combines the properties of each vector itself (static energy factor) and the distribution of the entire data, so as to ensure that the final aggregation representation can better reflect the global significant feature distribution. Specifically, the signal component waveform semantic feature vector corresponding to the largest signal component waveform semantic static energy factor is selected as the initial center vector of the signal component waveform semantic clustering to represent the most representative and holistic features in the data set. Then, the dynamic aggregation energy factor of each signal component waveform semantic feature vector is calculated based on the static energy factor of each signal component waveform semantic feature vector and the initial center vector of the clustering, and the spatial span between each signal component waveform semantic feature vector and the initial center vector of the clustering to obtain a set of signal component waveform semantic dynamic aggregation energy factors. Specifically, feature vectors with higher static energy factors tend to be more significant, indicating that they play a more important role in the signal. The spatial span reflects the distance or similarity between feature vectors. Feature vectors with a closer distance may represent similar signal features. Therefore, by combining the static energy factor and the spatial span, the importance and positional relationship of feature vectors in the dynamic aggregation process can be comprehensively evaluated, so as to more accurately determine which feature vectors should be aggregated together. Subsequently, each dynamic aggregation energy factor is input into the gated mask unit for weight screening and adaptive adjustment to significantly enhance key features and suppress irrelevant features. Finally, each masked factor is used as a weight to perform weighted summation on the set of semantic feature vectors of the signal component waveform to obtain a global significant aggregation representation vector of the signal component waveform, so as to comprehensively and adaptively adjust the importance of each feature.

[0048] Specifically, Figure 5The present invention is a flow chart of performing global significant aggregation of feature distribution field on the set of semantic feature vectors of signal component waveforms to obtain global significant aggregation representation of signal component waveforms in the signal processing method of the reconnaissance drone according to the embodiment of the present application. Figure 5 As shown, the set of signal component waveform semantic feature vectors is subjected to global significant aggregation of feature distribution field to obtain a global significant aggregation representation of the signal component waveform, including: S14131, calculating the static energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors to obtain a set of signal component waveform semantic static energy factors; S14132, based on the set of signal component waveform semantic static energy factors, performing force field-guided feature dynamic aggregation on the set of signal component waveform semantic feature vectors to obtain a global significant aggregation representation of the signal component waveform.

[0049] More specifically, the static energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors is calculated to obtain a set of signal component waveform semantic static energy factors, including: calculating the mean and variance of the signal component waveform semantic feature vectors to obtain the signal component waveform semantic mean and the signal component waveform semantic variance; subtracting the signal component waveform semantic feature vector from the signal component waveform semantic mean by position, and calculating the fourth power of each position of the feature vector after subtraction to obtain the signal component waveform semantic difference modulation vector; calculating the expected value of the signal component waveform semantic difference modulation vector to obtain the signal component waveform semantic expected value; dividing the signal component waveform semantic expected value by the square of the signal component waveform semantic variance, and inputting the divided value into the sigmoid function to obtain the signal component waveform semantic static energy factor.

[0050] More specifically, based on the set of signal component waveform semantic static energy factors, the set of signal component waveform semantic feature vectors is subjected to force field-guided feature dynamic aggregation to obtain a global significant aggregation representation of the signal component waveform, including: selecting the signal component waveform semantic feature vector corresponding to the largest signal component waveform semantic static energy factor in the set of signal component waveform semantic static energy factors as the initial center vector of signal component waveform semantic clustering; based on the spatial span between each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors and the initial center vector of signal component waveform semantic clustering and the static energy of each signal component waveform semantic feature vector The energy factor and the static energy factor of the initial center vector of the signal component waveform semantic clustering are used to calculate the dynamic aggregation energy factor of each signal component waveform semantic feature vector in the set of the signal component waveform semantic feature vectors to obtain a set of signal component waveform semantic dynamic aggregation energy factors; the set of signal component waveform semantic dynamic aggregation energy factors is input into a gated mask unit to obtain a set of signal component waveform semantic dynamic aggregation weight factors; the weighted sum of the set of signal component waveform semantic feature vectors is calculated using the set of signal component waveform semantic dynamic aggregation weight factors to obtain a signal component waveform global significant aggregation representation vector and use it as the signal component waveform global significant aggregation representation.

[0051] More specifically, in an embodiment of the present application, based on the spatial span between each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors and the initial center vector of the signal component waveform semantic clustering, as well as the static energy factor of each signal component waveform semantic feature vector and the static energy factor of the initial center vector of the signal component waveform semantic clustering, the dynamic aggregation energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors is calculated to obtain a set of signal component waveform semantic dynamic aggregation energy factors, including: multiplying the static energy factor of the signal component waveform semantic feature vector, the static energy factor of the initial center vector of the signal component waveform semantic clustering, and a first weighting parameter to obtain a first signal component waveform semantic dynamic aggregation energy factor; multiplying the square of the spatial span between the signal component waveform semantic feature vector and the initial center vector of the signal component waveform semantic clustering by a second weighting parameter to obtain a second signal component waveform semantic dynamic aggregation energy factor; dividing the first signal component waveform semantic dynamic aggregation energy factor by the second signal component waveform semantic dynamic aggregation energy factor to obtain the signal component waveform semantic dynamic aggregation energy factor.

[0052] More specifically, in an embodiment of the present application, the set of signal component waveform semantic dynamic aggregation energy factors is input into a gated mask unit to obtain a set of signal component waveform semantic dynamic aggregation weight factors, including: normalizing the set of signal component waveform semantic dynamic aggregation energy factors to obtain a set of normalized signal component waveform semantic dynamic aggregation energy factors; and masking the set of normalized signal component waveform semantic dynamic aggregation energy factors to obtain a set of signal component waveform semantic dynamic aggregation weight factors.

[0053] More specifically, in an embodiment of the present application, the set of signal component waveform semantic dynamic aggregation energy factors is normalized to obtain a set of normalized signal component waveform semantic dynamic aggregation energy factors, including: calculating an exponential function value with the natural constant e as the base and the negative of each signal component waveform semantic dynamic aggregation energy factor in the set of signal component waveform semantic dynamic aggregation energy factors as the exponent to obtain a set of signal component waveform semantic dynamic aggregation energy index factors; after positionally adding the set of signal component waveform semantic dynamic aggregation energy index factors to the constant one, the inverse of each signal component waveform semantic dynamic aggregation energy index modulation factor in the obtained set of signal component waveform semantic dynamic aggregation energy index modulation factors is calculated to obtain the set of normalized signal component waveform semantic dynamic aggregation energy factors.

[0054] More specifically, in an embodiment of the present application, the set of normalized signal component waveform semantic dynamic aggregation energy factors is masked to obtain a set of signal component waveform semantic dynamic aggregation weight factors, including: in response to each normalized signal component waveform semantic dynamic aggregation energy factor in the set of normalized signal component waveform semantic dynamic aggregation energy factors being greater than a predetermined threshold, setting the normalized signal component waveform semantic dynamic aggregation energy factor to the original value, and setting the rest to zero, so as to obtain the set of signal component waveform semantic dynamic aggregation weight factors.

[0055] In the embodiment of the present application, specifically, the set of the waveform semantic feature vectors of the signal component is subjected to significant aggregation of the feature distribution field in the whole domain, which can be expressed by the formula:

[0056] X={x1,x2,...,x i ,...,x n}

[0057]

[0058] x c =x m

[0059]

[0060] w si =mask(w i )

[0061]

[0062] Where X is the set of semantic feature vectors of the signal component waveform, x1, x2, ..., x i ,...,x n are respectively the 1st, 2nd, ..., ith, ..., nth signal component waveform semantic feature vectors in the set of signal component waveform semantic feature vectors, is the eigenvalue of each position in the semantic feature vector of the waveform of the i-th signal component, E(·) is the expected value, μ i and σ i 2 x i The mean and variance of , sigmoid(·) is the sigmoid function, is x i The corresponding signal component waveform semantic static energy factor, arg max j (·) is the j value corresponding to the maximum value, m is the maximum matching value, x c is the initial center vector of the semantic clustering of signal component waveforms, is x m The corresponding signal component waveform semantic static energy factor, Count(x i →x m ) represents x i With x m The spatial span between them, γ and δ are weighting parameters, For x i The corresponding signal component waveform semantic dynamic aggregation energy factor, w i is x i The corresponding normalized signal component waveform semantic dynamic aggregation energy factor, mask(·) is the masking process, θ is the predetermined threshold, w si is x i The corresponding signal component waveform semantic dynamic aggregation weight factor, n is the number of vectors in the set of signal component waveform semantic feature vectors, x f is the global significant aggregation representation vector of the signal component waveform.

[0063] In step S1414, the filtered first intermediate frequency signal is obtained based on the global significant aggregation representation of the signal component waveform. Specifically, in an embodiment of the present application, the filtered first intermediate frequency signal is obtained based on the global significant aggregation representation of the signal component waveform, including: inputting the global significant aggregation representation vector of the signal component waveform into a decoder-based signal restorer to obtain the filtered first intermediate frequency signal. By performing decoding processing on the global significant aggregation representation of the signal component waveform obtained by performing significant aggregation on the set of semantic feature vectors of the signal component waveform, the filtered first intermediate frequency signal is automatically generated, which can dynamically adapt to changes in the signal environment and extract more representative features to improve the flexibility of signal processing. And compared with traditional fixed parameter filters, it can better cope with interference and noise in complex environments, making the final signal clearer, thereby improving the quality and automation level of reconnaissance drone signal processing.

[0064] In particular, considering that each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors respectively represents the waveform semantic features of the intrinsic modal functions of each signal component obtained by performing empirical mode decomposition on the first intermediate frequency signal, when dynamically aggregating the feature distribution field based on the imitation field, the waveform semantic expression differences of the intrinsic modal functions of each signal component will also make the frequency domain fusion of the signal component waveform global domain representation vector represent the richness of multi-band feature aggregation coding, resulting in repetition of unstructured probability mapping, affecting the accuracy of the decoding results.

[0065] Preferably, in an example of the present application, inputting the global significant aggregation representation vector of the signal component waveform into a decoder-based signal restorer to obtain a filtered first intermediate frequency signal comprises:

[0066] Determine the number of zero eigenvalues ​​n in the global significant aggregation representation vector of the signal component waveform and the signal component waveform global significant aggregation zero differentiable value (n-1) obtained by subtracting one from the number of zero eigenvalues ​​n;

[0067] The global significant aggregation zero differentiable value (n-1) of the signal component waveform is multiplied and divided by the number of zero eigenvalues ​​n to obtain the global significant aggregation field fitting value (n-1)×n of the first signal component waveform and the global significant aggregation field fitting value (n-1) / n of the second signal component waveform;

[0068] Calculate the square root of the sum of squares of all eigenvalues ​​of the global significant aggregation representation vector of the signal component waveform to obtain the global significant aggregation mode representation value of the signal component waveform Among them, v i Each eigenvalue of the vector representing the global significant aggregation of the signal component waveform is represented, Indicates the global significant aggregation mode characterization value of the signal component waveform;

[0069] Each eigenvalue v of the global significant aggregation representation vector of the signal component waveform is calculated using the zero differentiable value (n-1) of the global significant aggregation of the signal component waveform as an index. i The power function v i n-1 , and multiply it by the first signal component waveform global significant aggregation field fitting value (n-1)×n to obtain the first signal component waveform global significant aggregation intermediate representation value v 1i ;

[0070] Calculate each eigenvalue v of the global significant aggregation representation vector of the signal component waveform i The second signal component waveform global significant aggregation field fitting value (n-1) / n and the signal component waveform global significant aggregation mode characterization value The product of the second signal component waveform is obtained by the global significant aggregation intermediate representation value v 2i ;

[0071] The first signal component waveform is globally significantly aggregated into an intermediate representation value v 1i The intermediate representation value v is significantly aggregated with the second signal component waveform over the entire domain 2i The weighted difference w1v 1i -w2v 2i An optimized signal component waveform global significant aggregation representation vector is formed, wherein w1 and w2 represent weighted hyperparameters; finally, the optimized signal component waveform global significant aggregation representation vector is input into the decoder-based signal restorer to obtain the filtered first intermediate frequency signal.

[0072] That is, based on the vector field fitting law of high-dimensional manifold, the isolated zero in the global significant aggregation representation vector of the signal component waveform is used as the differentiable dimension of the high-dimensional manifold, so as to fix the normal direction pointing to the distribution center near the local coordinates represented by the eigenvalue of the global significant aggregation representation vector of the signal component waveform, thereby avoiding the invalid repetition of the probability mapping process caused by the unstructured characteristics of the interactively rich distribution of the global significant aggregation representation vector of the signal component waveform by realizing the effective alignment of the characteristic distribution pattern of the global significant aggregation representation vector of the signal component waveform to the probability density differentiable field, improving the signal quality of the filtered first intermediate frequency signal obtained by the decoder-based signal restorer, being able to dynamically adapt to changes in the signal environment and extract more representative features to improve the flexibility of signal processing. And compared with the traditional fixed parameter filter, it can better cope with interference and noise in complex environments, making the final signal clearer, thereby improving the quality and automation level of reconnaissance drone signal processing.

[0073] In step S142, the filtered first intermediate frequency signal is amplified to obtain the enhanced first intermediate frequency signal. Specifically, in order to improve the strength of the first intermediate frequency signal, the filtered first intermediate frequency signal needs to be amplified to obtain the enhanced first intermediate frequency signal. Specifically, a low-noise amplifier with bypass mode can be used to amplify the filtered first intermediate frequency signal.

[0074] In step S150, the enhanced first intermediate frequency signal is mixed for a second time with the local oscillator signal generated by the second phase-locked loop to obtain a second intermediate frequency signal. Specifically, the local oscillator signal generated by the second phase-locked loop and the enhanced first intermediate frequency signal are mixed to convert the signal from the first intermediate frequency range to the second intermediate frequency range for more sophisticated processing. In the present application, the second intermediate frequency signal needs to use a lower intermediate frequency, which can effectively reduce the performance pressure of the subsequent DDS and AD sampling chips.

[0075] In step S160, the second intermediate frequency signal is filtered and amplified to obtain an enhanced second intermediate frequency signal. Specifically, the second intermediate frequency signal is filtered and amplified, similar to the processing of the first intermediate frequency signal. This step ensures the quality of the second intermediate frequency signal, removes excess noise and enhances the amplitude of the signal, in preparation for subsequent mixing and digital conversion.

[0076] In step S170, the enhanced second intermediate frequency signal is mixed for a third time with the local oscillator signal generated by the direct digital synthesizer to obtain a third intermediate frequency signal. Specifically, a direct digital synthesizer (DDS) is a digital circuit that can generate high-precision, programmable frequency and phase. The DDS frequency settling time is fast and meets the frequency hopping requirements. In this step, the local oscillator signal generated by the DDS is mixed with the enhanced second intermediate frequency signal to convert the signal from the second intermediate frequency range to the third intermediate frequency range. This step may help adjust the signal to a frequency band that is more suitable for subsequent processing.

[0077] In step S180, the third intermediate frequency signal is input into an analog-to-digital converter to obtain a digital signal. Specifically, an analog-to-digital converter (ADC) is a circuit that can convert an analog signal into a digital signal. In this step, the third intermediate frequency signal is input into the ADC for conversion to obtain a digital signal. Digital signals can be stored, analyzed and transmitted more conveniently.

[0078] In summary, the signal processing method of the reconnaissance drone based on the embodiment of the present application is explained, which performs signal decomposition on the first intermediate frequency signal obtained by mixing the drone signal to obtain a set of signal component intrinsic mode functions, and uses a signal analysis and processing algorithm based on deep learning to extract waveform semantic features of the intrinsic mode functions of each signal component, thereby automatically generating a filtered first intermediate frequency signal based on the significant dynamic aggregation representation of the waveform semantic features of each signal component in the entire domain, which can dynamically adapt to changes in the signal environment and extract more representative features to improve the flexibility of signal processing. And compared with traditional fixed parameter filters, it can better cope with interference and noise in complex environments, making the final signal clearer, thereby improving the quality and automation level of reconnaissance drone signal processing.

[0079] Figure 6 FIG. 1 is a system block diagram of a signal processing system for a reconnaissance drone according to an embodiment of the present application. Figure 6 As shown, according to the signal processing system 100 of the reconnaissance drone of the embodiment of the present application, it includes: a drone signal data collection module 110, which is used to obtain the drone signal collected by the antenna; a drone signal pre-selection and amplification module 120, which is used to pre-select the drone signal through a bandpass filter, and input the pre-selected drone signal into a low noise amplifier to obtain an amplified drone signal; a drone signal primary mixing module 130, which is used to perform a first mixing on the amplified drone signal and the local oscillator signal generated by the first phase-locked loop to obtain a first intermediate frequency signal; a first intermediate frequency signal enhancement module 140, which is used to filter and amplify the first intermediate frequency signal to obtain to an enhanced first intermediate frequency signal; a first enhanced intermediate frequency signal mixing module 150, used to perform a second mixing of the enhanced first intermediate frequency signal with a local oscillator signal generated by a second phase-locked loop to obtain a second intermediate frequency signal; a second intermediate frequency signal enhancing module 160, used to perform filtering and amplifying processing on the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal; a third intermediate frequency signal constructing module 170, used to perform a third mixing of the enhanced second intermediate frequency signal with a local oscillator signal generated by a direct digital synthesizer to obtain a third intermediate frequency signal; a third intermediate frequency signal analog-to-digital conversion module 180, used to input the third intermediate frequency signal into an analog-to-digital converter to obtain a digital signal.

[0080] The specific functions and operations of the various units and modules in the signal processing system 100 of the above-mentioned reconnaissance drone have been referred to above. Figures 1 to 5 The signal processing method of the reconnaissance drone has been introduced in detail, and therefore, its repeated description will be omitted.

[0081] In summary, the signal processing system 100 of the reconnaissance drone based on the embodiment of the present application is explained, which performs signal decomposition on the first intermediate frequency signal obtained by mixing the drone signal to obtain a set of signal component intrinsic mode functions, and uses a signal analysis and processing algorithm based on deep learning to extract waveform semantic features of the intrinsic mode functions of each signal component, thereby automatically generating a filtered first intermediate frequency signal based on the significant dynamic aggregation representation of the waveform semantic features of each signal component in the entire domain, and can dynamically adapt to changes in the signal environment and extract more representative features to improve the flexibility of signal processing. And compared with traditional fixed parameter filters, it can better cope with interference and noise in complex environments, making the final signal clearer, thereby improving the quality and automation level of reconnaissance drone signal processing.

[0082] In other embodiments of the present application, a drone detection module is also provided. Figure 7 FIG. 1 is a schematic diagram of the overall design of a drone detection module according to other embodiments of the present application. Figure 7 As shown, the UAV detection module adopts a superheterodyne triple frequency conversion structure, and the signals in each passband are band-pass filtered, low-noise amplified, mixed to intermediate frequency by the first and second frequency mixing, and filtered and amplified at the intermediate frequency. After the three mixing, AD sampling is performed, and then the single-chip microcomputer performs detection. According to the detection results, the display screen and the sound and light alarm are controlled to interact. In detail, according to the different frequencies of the received signals, the radio frequency is divided into 4 channels, and the architecture of each channel is consistent. The filtering and amplification parameters are fine-tuned according to the different signal frequencies. The received signal at the antenna is filtered and amplified, and then mixed with a local oscillator to generate an intermediate frequency signal. After filtering and amplification, it is mixed with the second local oscillator to generate a second local oscillator signal. After processing, the third mixing is performed to generate a signal, which is then sampled by AD and sent to the microcontroller (MCU) for processing. The MCU performs human-computer interaction according to the detection situation. Specifically, if the UAV frequency hopping signal needs to be further processed later, the first and second local oscillators can be set to fixed frequencies within the passband, then the first and second intermediate frequencies are frequency hopping signals within the band, and the intermediate frequency filtering is changed to low pass; by controlling the three local oscillator DDS, the frequency hopping signal can be captured to the three intermediate frequencies for fixed frequency processing to meet the frequency hopping requirements.

Claims

1. A signal processing method for a reconnaissance drone, characterized in that: include: Acquire the drone signal collected by the antenna; Preselecting the drone signal through a bandpass filter, and inputting the preselected drone signal into a low noise amplifier to obtain an amplified drone signal; Performing a first mixing of the amplified UAV signal and a local oscillator signal generated by a first phase-locked loop to obtain a first intermediate frequency signal; Performing filtering and amplification processing on the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal; Performing a second mixing of the enhanced first intermediate frequency signal and a local oscillator signal generated by a second phase-locked loop to obtain a second intermediate frequency signal; Performing filtering and amplification processing on the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal; Performing a third mixing of the enhanced second intermediate frequency signal and a local oscillator signal generated by a direct digital synthesizer to obtain a third intermediate frequency signal; The third intermediate frequency signal is input into an analog-to-digital converter to obtain a digital signal.

2. The signal processing method of a reconnaissance drone according to claim 1, characterized in that: The method of filtering and amplifying the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal includes: Performing filtering processing on the first intermediate frequency signal to obtain a filtered first intermediate frequency signal; The filtered first intermediate frequency signal is amplified to obtain the enhanced first intermediate frequency signal.

3. The signal processing method of the reconnaissance drone according to claim 2, characterized in that: Filtering the first intermediate frequency signal to obtain a filtered first intermediate frequency signal includes: Performing signal decomposition based on empirical mode decomposition technology on the first intermediate frequency signal to obtain a set of intrinsic mode functions of signal components; Performing signal component waveform feature extraction on each signal component intrinsic modal function in the set of signal component intrinsic modal functions to obtain a set of signal component waveform semantic feature vectors; The set of the signal component waveform semantic feature vectors is subjected to global significant aggregation of feature distribution fields to obtain a global significant aggregation representation of the signal component waveform, including: calculating the static energy factor of each signal component waveform semantic feature vector in the set of the signal component waveform semantic feature vectors to obtain a set of signal component waveform semantic static energy factors; based on the set of signal component waveform semantic static energy factors, the set of the signal component waveform semantic feature vectors is subjected to force field-guided feature dynamic aggregation to obtain a global significant aggregation representation of the signal component waveform; Based on the global significant aggregation representation of the signal component waveforms, the filtered first intermediate frequency signal is obtained.

4. The signal processing method of the reconnaissance drone according to claim 3 is characterized in that: Performing signal component waveform feature extraction on each signal component intrinsic modal function in the set of signal component intrinsic modal functions to obtain a set of signal component waveform semantic feature vectors, including: inputting each signal component intrinsic modal function in the set of signal component intrinsic modal functions into a signal component waveform feature extractor based on a 1D-CNN model to obtain the set of signal component waveform semantic feature vectors.

5. The signal processing method of a reconnaissance drone according to claim 4, characterized in that: Calculating the static energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors to obtain a set of signal component waveform semantic static energy factors, including: Calculating the mean and variance of the signal component waveform semantic feature vector to obtain the signal component waveform semantic mean and the signal component waveform semantic variance; Subtracting the signal component waveform semantic feature vector from the signal component waveform semantic mean by position, and calculating the fourth power of each position of the feature vector after the subtraction to obtain a signal component waveform semantic difference modulation vector; Calculating the expected value of the signal component waveform semantic difference modulation vector to obtain the signal component waveform semantic expected value; The signal component waveform semantic expected value is divided by the square of the signal component waveform semantic variance, and the value obtained by the division is input into a sigmoid function to obtain the signal component waveform semantic static energy factor.

6. The signal processing method of a reconnaissance drone according to claim 5, characterized in that: Based on the set of semantic static energy factors of the signal component waveform, a force field-guided feature dynamic aggregation is performed on the set of semantic feature vectors of the signal component waveform to obtain a global significant aggregation representation of the signal component waveform, including: Selecting a signal component waveform semantic feature vector corresponding to the largest signal component waveform semantic static energy factor in the set of signal component waveform semantic static energy factors as an initial center vector of signal component waveform semantic clustering; Based on the spatial span between each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors and the initial center vector of the signal component waveform semantic clustering, as well as the static energy factor of each signal component waveform semantic feature vector and the static energy factor of the initial center vector of the signal component waveform semantic clustering, the dynamic aggregation energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors is calculated to obtain a set of signal component waveform semantic dynamic aggregation energy factors; Inputting the set of signal component waveform semantic dynamic aggregation energy factors into a gated mask unit to obtain a set of signal component waveform semantic dynamic aggregation weight factors; The weighted sum of the set of signal component waveform semantic dynamic aggregation weight factors is calculated to obtain a signal component waveform global significant aggregation representation vector as the signal component waveform global significant aggregation representation.

7. The signal processing method of a reconnaissance drone according to claim 6, characterized in that: Based on the spatial span between each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors and the initial center vector of the signal component waveform semantic clustering, as well as the static energy factor of each signal component waveform semantic feature vector and the static energy factor of the initial center vector of the signal component waveform semantic clustering, the dynamic aggregation energy factor of each signal component waveform semantic feature vector in the set of signal component waveform semantic feature vectors is calculated to obtain a set of signal component waveform semantic dynamic aggregation energy factors, including: Multiplying the static energy factor of the signal component waveform semantic feature vector, the static energy factor of the signal component waveform semantic clustering initial center vector and a first weighting parameter to obtain a first signal component waveform semantic dynamic aggregation energy factor; Multiplying the square of the spatial span between the signal component waveform semantic feature vector and the signal component waveform semantic clustering initial center vector by a second weighting parameter to obtain a second signal component waveform semantic dynamic aggregation energy factor; The first signal component waveform semantic dynamic aggregation energy factor is divided by the second signal component waveform semantic dynamic aggregation energy factor to obtain the signal component waveform semantic dynamic aggregation energy factor.

8. The signal processing method of a reconnaissance drone according to claim 7, characterized in that: Inputting the set of signal component waveform semantic dynamic aggregation energy factors into a gated mask unit to obtain a set of signal component waveform semantic dynamic aggregation weight factors, including: Normalizing the set of signal component waveform semantic dynamic aggregation energy factors to obtain a set of normalized signal component waveform semantic dynamic aggregation energy factors; The set of normalized signal component waveform semantic dynamic aggregation energy factors is subjected to masking processing to obtain the set of signal component waveform semantic dynamic aggregation weight factors.

9. The signal processing method of a reconnaissance drone according to claim 8, characterized in that: Based on the global significant aggregation representation of the signal component waveform, the filtered first intermediate frequency signal is obtained, including: inputting the global significant aggregation representation vector of the signal component waveform into a decoder-based signal restorer to obtain the filtered first intermediate frequency signal.

10. A signal processing system for a reconnaissance drone, characterized in that: include: A drone signal data collection module is used to obtain drone signals collected by an antenna; A drone signal preselection and amplification module, used to preselect the drone signal through a bandpass filter, and input the preselected drone signal into a low noise amplifier to obtain an amplified drone signal; A drone signal primary mixing module, used for performing a first mixing of the amplified drone signal with a local oscillator signal generated by a first phase-locked loop to obtain a first intermediate frequency signal; A first intermediate frequency signal enhancement module, configured to perform filtering and amplification processing on the first intermediate frequency signal to obtain an enhanced first intermediate frequency signal; A first enhanced intermediate frequency signal mixing module, used for performing a second mixing on the enhanced first intermediate frequency signal and a local oscillator signal generated by a second phase-locked loop to obtain a second intermediate frequency signal; A second intermediate frequency signal enhancement module, configured to perform filtering and amplification processing on the second intermediate frequency signal to obtain an enhanced second intermediate frequency signal; A third intermediate frequency signal building module, used for performing a third mixing of the enhanced second intermediate frequency signal and the local oscillator signal generated by the direct digital synthesizer to obtain a third intermediate frequency signal; The third intermediate frequency signal analog-to-digital conversion module is used to input the third intermediate frequency signal into an analog-to-digital converter to obtain a digital signal.

Images