Unmanned aerial vehicle remote controller signal source interference control system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an unmanned aerial vehicle remote controller signal source interference control system based on artificial intelligence. The system comprises a signal acquisition unit which captures a radio signal sent by an unmanned aerial vehicle remote controller through a radio receiving module; the signal processing unit is used for carrying out data processing on the radio signals through a feature extraction algorithm according to the radio signals captured by the signal acquisition unit; the identification and classification unit classifies the radio signals through an identification and classification model according to the radio signals processed by the signal processing unit; and the interference control unit is used for designing an interference control decision rule according to the classification result of the identification and classification unit. Through the identification and classification model, the processed radio signals can be accurately classified, and normal control signals and potential interference signals are distinguished.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based unmanned aerial vehicle remote controller signal source interference control system. Background Art

[0002] The AI-based drone remote control signal source jammer control system is an advanced technology solution designed to effectively counter unauthorized drone activities by intelligently identifying and precisely jamming drone remote control signals. The core of the system is the combination of modern AI algorithms and wireless communication technology, which can automatically detect and classify drone signals and quickly generate targeted jammer signals to interrupt the communication link between the drone and the ground control station, forcing the drone to enter a preset safety mode, such as return or landing.

[0003] In the existing AI-based UAV remote control signal source interference control system, the system that relies on AI algorithms requires a lot of data training to improve its accuracy and reliability. In the field of UAVs, obtaining high-quality training data is difficult and expensive. Due to the diversity and complexity of UAV remote control signals, the existing interference control system may be difficult to fully cover all types of signals, especially when facing new or unknown interference methods, the system's defense capability will be greatly reduced. Therefore, a UAV remote control signal source interference control system based on AI is designed. Summary of the invention

[0004] The purpose of the present invention is to provide an artificial intelligence-based UAV remote control signal source interference control system to solve the problem raised in the above background technology that in the existing artificial intelligence-based UAV remote control signal source interference control system, the system relying on artificial intelligence algorithm requires a large amount of data training to improve its accuracy and reliability, while in the field of UAVs, it is difficult and expensive to obtain high-quality training data. Due to the diversity and complexity of UAV remote control signals, the existing interference control system may be difficult to fully cover all types of signals, especially when facing new or unknown interference means, the system's defense capability will be greatly reduced.

[0005] To achieve the above-mentioned object, the present invention aims to provide an artificial intelligence-based UAV remote controller signal source interference control system, comprising a signal acquisition unit, which captures a radio signal emitted by the UAV remote controller through a radio receiving module; A signal processing unit, wherein the signal processing unit performs data processing on the radio signal through a feature extraction algorithm according to the radio signal captured by the signal acquisition unit; an identification and classification unit, wherein the identification and classification unit classifies the radio signal by using an identification and classification model according to the radio signal processed by the signal processing unit; An interference control unit designs interference control decision rules according to the classification result of the identification and classification unit.

[0006] As a further improvement of the technical solution: the signal acquisition unit includes a radio receiving module, a power identification module and a frequency identification module; Wherein, the radio receiving module collects radio signals through the SDR device; ; in, Indicates the received signal; represents the impulse response of the receiving device; represents the actual transmitted signal; Indicates the change of the signal on the time axis; Represents the relationship between signals at different time points; The power identification module is used to measure the power level of the received radio signal; The frequency identification module is used to identify the frequency used by the radio signal.

[0007] As a further improvement of the technical solution, the specific steps of the signal processing unit performing data processing on the radio signal captured by the signal acquisition unit based on the feature extraction algorithm are: S2.1, pre-processing the captured radio signals for denoising and normalization; S2.2, extract features from the preprocessed signal; S2.3. Combine the extracted features into a feature vector for subsequent signal classification and recognition.

[0008] As a further improvement of the technical solution, in S2.1, the specific process of preprocessing the captured radio signal for denoising and normalization is as follows: Use a low-pass filter to remove high-frequency noise; ; in, represents the filtered signal; represents the original signal; represents the impulse response of the low-pass filter; Represents the convolution operation; The convolution operation is expressed as: ; in, Represents the index in the convolution operation; Indicates infinite value; Indicates the input signal at time The value of represents the impulse response of the system, that is, at time Relative to The displacement value of Normalize the signal to a fixed range; ; in, represents the normalized signal; Represents the original signal The minimum value of Represents the original signal The maximum value of .

[0009] As a further improvement of the technical solution, in S2.2, the specific process of extracting features from the preprocessed signal is as follows: The features are divided into three categories for extraction: Time domain characteristics: Compute the mean of a signal: ; in, represents the average value of the signal; Indicates the length of the signal; Represents an index variable; Compute the variance of a signal: ; in, represents the variance of the signal; Calculate the difference between the maximum and minimum values ​​of a signal: ; in, Indicates the maximum value of the signal With minimum difference; Count the number of times a signal crosses zero: ; in, Indicates the number of times the signal crosses zero; represents a symbolic function; express The next signal data point of Frequency domain characteristics: Convert a time domain signal to a frequency domain signal: ; in, represents a frequency domain signal; represents the frequency index; represents the scale coefficient; Compute the total energy of the spectrum: ; in, Represents the total energy of the spectrum; Calculate the entropy of a spectrum: ; in, represents the entropy of the spectrum; No. The normalized energy of the frequency components; Wavelet transform features: Decompose the signal into wavelet coefficients of different scales: ; in, represents the coefficients after wavelet transformation; represents the wavelet basis function; Compute the energy of wavelet coefficients at each scale: ; in, Indicates The energy of the wavelet coefficients at scale.

[0010] As a further improvement of the present technical solution, the recognition and classification unit constructs a recognition and classification model based on the SVM algorithm, and the recognition and classification model construction process is: S3.1, creating a data set based on the feature vector of the signal processing unit; S3.2, divide the data set into training set and test set; S3.3, train the recognition and classification model through the SVM algorithm; S3.4. Use the test set to test and evaluate the performance of the model.

[0011] As a further improvement of the technical solution, in S3.3, the specific training steps of training the recognition classification model by the SVM algorithm are: S3.31, Use Gaussian kernel function to deal with nonlinear classification problems; Among them, the specific expression of Gaussian kernel function is: ; in, represents the Gaussian kernel function; Indicates The feature vector of samples; Indicates The feature vector of samples; Represents the parameters of the kernel function, controlling the width of the Gaussian kernel; represents the natural exponential function; S3.32. Construct an optimization problem to maximize the interval between two types of data; The optimization problem is expressed as: ; ; ; in, represents the weight vector; represents the bias term; represents the slack variable; represents the regularization parameter; represents the total number of samples; Indicates The labels of the samples; S3.33, transform the optimization problem into a dual problem; The dual problem is expressed as: ; ; ; in, Indicates Lagrange multiplier of samples; Indicates Lagrange multiplier of samples; Indicates The labels of the samples; S3.34. Use the SMO algorithm to solve the dual problem and obtain the optimal Lagrange multiplier .

[0012] As a further improvement of the technical solution, in S3.34, the SMO algorithm is used to solve the dual problem to obtain the optimal Lagrange multiplier The specific steps are: Initialize all Lagrange multipliers and select two Lagrange multipliers and Optimize During the optimization process, add random perturbations: ; ; in, and ; represents the standard deviation, which is used to control the intensity of the disturbance; calculate The border and : ; ; Calculate the error between two samples and : ; ; renew : ; in, Denotes the new Lagrange multiplier ; represents the second-order derivative of the kernel function; Will Adjust to Bounds and between: ; renew : ; in, Denotes the new Lagrange multiplier ; Update Threshold : ; ; in, represents the Lagrange multiplier The threshold value of represents the Lagrange multiplier The threshold value of represents the original threshold; Select a new threshold : ; in, represents the new threshold of the Lagrange multiplier; Check whether the convergence condition is met by checking whether the change of the Lagrange multiplier is less than the threshold. If the convergence condition is not met, repeat the above steps; After the convergence condition is met, the optimal Lagrange multiplier is obtained by substituting the solution of the dual problem into .

[0013] As a further improvement of the technical solution, in S3.4, the specific steps of using the test set to test the performance of the evaluation model are: S3.41. According to the optimal Lagrange multiplier , calculate the weight vector and bias term; Among them, the specific expressions for calculating the weight vector and bias term are: ; ; in, represents the support vector; Labels representing support vectors; S3.42, predict new samples; Among them, the specific expression for predicting new samples is: ; ; in, Represents the new sample feature vector; represents the decision function; Represents the predicted label of the new sample; S3.43. Use the test set to evaluate the performance of the model and obtain the evaluation indicators.

[0014] As a further improvement of the technical solution, the interference control unit designs an interference control decision rule according to the classification result of the identification and classification unit, that is, the interference control decision rule is specifically: S4.1. Adjust the transmission power of the interference signal according to the transmission power of the UAV; Among them, the basis for adjusting the transmission power of the interference signal according to the transmission power of the drone is: ; in, Indicates the transmission power of the interference signal; Indicates the maximum transmit power; Indicates the minimum generated power; Indicates the UAV transmission power; Indicates high drone transmit power; Indicates low to medium UAV transmission power; S4.2. Adjust the frequency of the interference signal according to the UAV radio wave frequency; Among them, the basis for adjusting the frequency of the interference signal according to the UAV radio wave frequency is: ; in, Indicates the frequency of the interference signal; Indicates the new frequency; Indicates the current frequency; Indicates the frequency of radio waves; Indicates medium and high drone radio wave frequencies; Indicates low drone radio wave frequency; S4.3, generating a pseudo-random sequence as an interference signal through LFSR; ; in, represents the interference signal sequence; represents the feedback coefficient; Indicates the length of the register; Represents the modulo operation.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This AI-based drone remote control signal source interference control system is equipped with high-performance radio receiving equipment, which can adapt to different frequency bands and modulation methods to ensure the integrity and accuracy of the signal. By preprocessing and extracting features from the captured radio signal, noise and redundant information can be removed, highlighting the key features of the signal.

[0016] 2. In this AI-based UAV remote control signal source interference control system, the processed radio signals can be accurately classified through the identification and classification model to distinguish normal control signals from potential interference signals. Based on the results of the identification and classification unit, an interference control strategy is formulated and implemented to protect the UAV from malicious interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is the overall flow chart of the present invention; The meaning of each number in the figure is: 1. Signal acquisition unit; 2. Signal processing unit; 3. Identification and classification unit; 4. Interference control unit. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example

[0020] See also Figure 1 As shown, a UAV remote controller signal source interference control system based on artificial intelligence is provided, comprising a signal acquisition unit 1, the signal acquisition unit 1 captures the radio signal emitted by the UAV remote controller through a radio receiving module; In this example, the signal acquisition unit 1 includes a radio receiving module, a power identification module and a frequency identification module; Wherein, the radio receiving module collects radio signals through the SDR device; ; in, Indicates the received signal; represents the impulse response of the receiving device; represents the actual transmitted signal; Indicates the change of the signal on the time axis; Represents the relationship between signals at different time points; The power identification module is used to measure the power level of the received radio signal; The frequency identification module is used to identify the frequency used by the radio signal.

[0021] Specifically, SDR (Software Defined Radio) is a radio communication technology whose core idea is to use software as much as possible to implement radio functions that are traditionally performed by hardware.

[0022] The SDR device efficiently collects the radio signals emitted by the drone remote control, provides high-fidelity and wide-bandwidth raw data, adapts to complex electromagnetic environments and has anti-interference capabilities, and provides a reliable foundation for subsequent signal processing, feature extraction, and intelligent classification, ensuring real-time monitoring and rapid response of the system, thereby effectively ensuring the safe and stable flight of the drone.

[0023] In the UAV remote control signal source interference control system, the power identification module can be used to determine which direction or which signal may be the main control signal, because usually the legitimate remote control signal will be stronger than the background noise or other interference signals; in the UAV remote control signal source interference control system, the frequency identification module helps the system accurately locate and identify the specific frequency used by the UAV remote control, thereby providing the necessary information for subsequent signal processing and interference.

[0024] The artificial intelligence-based UAV remote control signal source interference control system also includes a signal processing unit 2, which performs data processing on the radio signal through a feature extraction algorithm according to the radio signal captured by the signal acquisition unit 1; In this example, the specific steps of the signal processing unit 2 performing data processing on the radio signal captured by the signal acquisition unit 1 based on the feature extraction algorithm are as follows: S2.1, pre-processing the captured radio signals for denoising and normalization; In this example, the specific process of preprocessing the captured radio signal for denoising and normalization is as follows: Use a low-pass filter to remove high-frequency noise; ; in, represents the filtered signal; represents the original signal; represents the impulse response of the low-pass filter; Represents the convolution operation; The convolution operation is expressed as: ; in, Represents the index in the convolution operation; Indicates infinite value; Indicates the input signal at time The value of represents the impulse response of the system, that is, at time Relative to The displacement value of Normalize the signal to a fixed range; ; in, represents the normalized signal; Represents the original signal The minimum value of Represents the original signal The maximum value of .

[0025] S2.2, extract features from the preprocessed signal; In this example, the specific process of extracting features from the preprocessed signal is as follows: The features are divided into three categories for extraction: Time domain characteristics: Compute the mean of a signal: ; in, represents the average value of the signal; Indicates the length of the signal; Represents an index variable; Compute the variance of a signal: ; in, represents the variance of the signal; Calculate the difference between the maximum and minimum values ​​of a signal: ; in, Indicates the maximum value of the signal With minimum difference; Count the number of times a signal crosses zero: ; in, Indicates the number of times the signal crosses zero; represents a symbolic function; express The next signal data point of Specifically, the symbolic function expression is: ; Frequency domain characteristics: Convert a time domain signal to a frequency domain signal: ; in, represents a frequency domain signal; represents the frequency index; represents the scale coefficient; Compute the total energy of the spectrum: ; in, Represents the total energy of the spectrum; Calculate the entropy of a spectrum: ; in, represents the entropy of the spectrum; No. The normalized energy of the frequency components; Specifically, The expression is: ; Wavelet transform features: Decompose the signal into wavelet coefficients of different scales: ; in, represents the coefficients after wavelet transformation; represents the wavelet basis function; Compute the energy of wavelet coefficients at each scale: ; in, Indicates The energy of the wavelet coefficients at scale.

[0026] S2.3, select the feature that best represents the characteristics of the drone signal from the extracted features; S2.4. Combine the selected features into a feature vector for subsequent signal classification and recognition.

[0027] Specifically, denoising, normalizing and feature extraction of the captured radio signals can effectively remove noise, retain the key features of the signal, and generate high-quality feature vectors. These feature vectors provide reliable input for subsequent recognition and classification units, helping to improve the recognition accuracy and anti-interference ability of the system, thereby ensuring the safety and stability of the drone remote control signal.

[0028] The artificial intelligence-based UAV remote control signal source interference control system also includes an identification and classification unit 3, which classifies the radio signal through an identification and classification model according to the radio signal processed by the signal processing unit 2; In this example, the recognition classification unit 3 constructs a recognition classification model based on the SVM algorithm, and the recognition classification model construction process is: S3.1, creating a data set according to the feature vector of signal processing unit 2; Specifically, the data set includes the signal type; Signal types include FHSS signal, QPSK signal, DSSS signal, time hopping signal, and AM signal; FHSS (Frequency-Hopping Spread Spectrum) frequency hopping spread spectrum signal: Features: FHSS technology transmits data by quickly switching between multiple predetermined frequencies, which can reduce the risk of interference and eavesdropping and improve the security and reliability of communications. In each frequency hopping cycle, the signal is transmitted on a different carrier frequency.

[0029] Power: Typically between tens of milliwatts to several watts, depending on application needs and regulatory restrictions.

[0030] Frequency: Common frequency hopping frequency ranges include the 2.4GHz ISM band (2.400 GHz to 2.4835 GHz) and the 900MHz band (902 MHz to 928 MHz).

[0031] QPSK (Quadrature Phase Shift Keying) quadrature phase shift keying signal: Features: QPSK is a modulation technique that transmits two bits of information simultaneously on two orthogonal carriers, so that each symbol can represent four possible states. This modulation method has high spectral efficiency.

[0032] Power: Typically between a few milliwatts to several hundred milliwatts, depending on transmission distance and channel conditions.

[0033] Frequency: Widely used in various wireless communication systems, including satellite communications, mobile communications (such as LTE), Wi-Fi, etc. Common frequency ranges include 2.4 GHz, 5 GHz, and higher millimeter wave bands.

[0034] DSSS (Direct Sequence Spread Spectrum) direct sequence spread spectrum signal: Features: DSSS technology expands the signal bandwidth by mixing the data signal with a high-frequency pseudo-random code (PN code), thereby improving anti-interference and concealment. The receiving end recovers the original data through the same PN code.

[0035] Power: Typically between a few milliwatts to several hundred milliwatts, depending on application needs.

[0036] Frequency: Common frequency ranges include the 2.4 GHz ISM band and the 900 MHz band.

[0037] Frequency-Hopping Spread Spectrum with Time Hopping: Features: Time hopping technology transmits data by jumping in time and frequency at the same time, further improving the security and anti-interference ability of communication. The time point and frequency of each transmission are random.

[0038] Power: Usually between a few milliwatts to several hundred milliwatts.

[0039] Frequency: Common frequency ranges include the 2.4 GHz ISM band and the 900 MHz band.

[0040] AM (Amplitude Modulation) amplitude modulation signal: Features: AM is a basic modulation technique that represents information by changing the amplitude of the carrier signal. AM signals are simple and easy to implement, but are more sensitive to noise and interference.

[0041] Power: The power of a radio station's AM transmitter can range from a few kilowatts to tens of kilowatts.

[0042] Frequency: The frequency range commonly used for AM broadcasting is 530 kHz to 1710 kHz (the medium wave band).

[0043] S3.2, divide the data set into training set and test set; S3.3, train the recognition and classification model through the SVM algorithm; In this example, the specific training steps for training the recognition classification model through the SVM algorithm are: S3.31, Use Gaussian kernel function to deal with nonlinear classification problems; Among them, the specific expression of Gaussian kernel function is: ; in, Represents the Gaussian kernel function, which is used to calculate the similarity between two samples in high-dimensional space; Indicates The feature vector of samples; Indicates The feature vector of samples; Represents the parameters of the kernel function, controlling the width of the Gaussian kernel; represents the natural exponential function; Specifically, in practical applications, the signal of the drone remote controller may not be linearly separable, which means that a straight line (or hyperplane, in multidimensional space) cannot be used to clearly divide different signal types. Therefore, using the Gaussian kernel function can map the original feature space to a higher-dimensional space where the signal may be linearly separable. This enables SVM to handle nonlinear classification problems.

[0044] S3.32. Construct an optimization problem to maximize the interval between two types of data; The optimization problem is expressed as: ; ; ; in, represents the weight vector; represents the bias term; represents the slack variable; represents the regularization parameter; represents the total number of samples; Indicates The labels of the samples; Specifically, the core idea of ​​SVM is to find a hyperplane with the largest margin to separate data of different categories while ensuring correct classification. The margin here refers to the distance from the nearest data point (support vector) to the hyperplane. In order to allow a certain degree of misclassification (because real-world data is often imperfect), a slack variable is introduced. , and through the regularization parameter To balance the relationship between maximizing the margin and minimizing the misclassification.

[0045] S3.33, transform the optimization problem into a dual problem; The dual problem is expressed as: ; ; ; in, Indicates Lagrange multiplier of samples; Indicates Lagrange multiplier of samples; Indicates The labels of the samples; Specifically, it is difficult to directly solve the above optimization problem, especially when it comes to nonlinear classification. By mathematically transforming the original problem, it can be converted into a problem that is easier to solve - the dual problem. In this process, kernel functions (such as Gaussian kernel functions) are used to calculate the similarity between different samples.

[0046] S3.34. Use the SMO algorithm to solve the dual problem and obtain the optimal Lagrange multiplier .

[0047] Specifically, by using the Sequential Minimal Optimization (SMO) algorithm, this dual problem can be efficiently solved to find the optimal Lagrange multiplier These multipliers determine which samples are support vectors and their role in determining the final classification boundary.

[0048] In this example, the SMO algorithm is used to solve the dual problem and obtain the optimal Lagrange multiplier The specific steps are: Initialize all Lagrange multipliers and select two Lagrange multipliers and Optimize During the optimization process, add random perturbations: ; ; in, and ; represents the standard deviation, which is used to control the intensity of the disturbance; calculate The border and : ; ; Calculate the error between two samples and : ; ; renew : ; in, Denotes the new Lagrange multiplier ; represents the second-order derivative of the kernel function; Will Adjust to Bounds and between: ; renew : ; in, Denotes the new Lagrange multiplier ; Update Threshold : ; ; in, represents the Lagrange multiplier The threshold value of represents the Lagrange multiplier The threshold value of represents the original threshold; Select a new threshold : ; in, represents the new threshold of the Lagrange multiplier; Check whether the convergence condition is met by checking whether the change of the Lagrange multiplier is less than the threshold. If the convergence condition is not met, repeat the above steps; After the convergence condition is met, the optimal Lagrange multiplier is obtained by substituting the solution of the dual problem into .

[0049] Specifically, the implementation process is: alpha = np.random.uniform(0, C, M) while not converged: indices = np.random.permutation(M) for i in indices: # Randomly select another variable j j = randint(0, M-1) while j == i: j = randint(0, M-1) # Calculate the current error Ei = error(i) Ej = error(j) # Check whether the optimization conditions are met if (yi * Ei < -tol and alpha[i] < C) or (yi * Ei > tol andalpha[i] > 0): # Calculate the boundaries L, H = compute_bounds(alpha[i], alpha[j], yi, yj) # If the boundaries are the same, skip if L == H: continue # Calculate the standard step size eta = 2.0 * K[i, j] - K[i, i] - K[j, j] if eta >= 0: continue # Randomly adjust the step size epsilon = np.random.uniform(-delta, delta) eta_prime = eta * (1 + epsilon) # Update alpha[j] alpha_j_old = alpha[j] alpha[j] -= yi * (Ei - Ej) / eta_prime alpha[j] = clip(alpha[j], L, H) # If alpha[j] changes very little, skip if abs(alpha[j] - alpha_j_old) < 1e-5: continue # Update alpha[i] alpha[i] += yi * yj * (alpha_j_old - alpha[j]) # Add random perturbations alpha_i_perturbation = np.random.normal(0, sigma) alpha_j_perturbation = np.random.normal(0, sigma) alpha[i] += alpha_i_perturbation alpha[j] += alpha_j_perturbation # Check if the new solution is better if new_solution_is_better(): accept_new_solution() else: reject_new_solution() S3.4. Use the test set to test and evaluate the performance of the model.

[0050] In this example, the specific steps to use the test set to test the performance of the evaluation model are: S3.41. According to the optimal Lagrange multiplier , calculate the weight vector and bias term; Among them, the specific expressions for calculating the weight vector and bias term are: ; ; in, represents the support vector; Labels representing support vectors; S3.42, predict new samples; Among them, the specific expression for predicting new samples is: ; ; in, Represents the new sample feature vector; represents the decision function; Represents the predicted label of the new sample; S3.43. Use the test set to evaluate the performance of the model and obtain the evaluation indicators.

[0051] Specifically, the evaluation indicators include accuracy, precision, recall, and F1 score; Among them, the accuracy is: ; in, Indicates accuracy; Represents true positive examples, that is, the number of samples that are actually positive and predicted to be positive; Represents true negative examples, that is, the number of samples that are actually negative and predicted to be negative; Represents false positive examples, that is, the number of samples that are actually negative but predicted to be positive; Represents false negative examples, that is, the number of samples that are actually positive but predicted to be negative; The accuracy is: ; in, Indicates accuracy; The recall rate is: ; in, represents the recall rate; The F1 score is: ; in, Denotes the F1 score.

[0052] Specifically, the SVM model is constructed and trained to effectively classify the drone remote control signal and interference signal. The model uses the Gaussian kernel function to handle nonlinear classification problems, and the solution of the optimization problem ensures the robustness and generalization ability of the model. Finally, the performance of the model is evaluated through the test set to ensure the reliability and effectiveness of the model in practical applications. This provides strong technical support for the protection and interference suppression of drone remote control signals, and ensures the safe and stable flight of drones.

[0053] The artificial intelligence-based UAV remote control signal source interference control system also includes an interference control unit 4, which designs interference control decision rules according to the classification results of the identification and classification unit 3.

[0054] Specifically, FHSS signals, QPSK signals, DSSS signals, time-hopping signals, and AM signals are classified according to frequency and power; power: Low power: FHSS signal, QPSK signal, DSSS signal, time hopping signal; Medium power: FHSS signal, QPSK signal; High power: AM signal; frequency: Low frequency: no above signal; Medium frequency: AM signal; High frequency: FHSS signal, QPSK signal, DSSS signal, time hopping signal.

[0055] In this example, the interference control unit 4 designs an interference control decision rule according to the classification result of the identification and classification unit 3, that is, the interference control decision rule is specifically: S4.1. Adjust the transmission power of the interference signal according to the transmission power of the UAV; Among them, the basis for adjusting the transmission power of the interference signal according to the transmission power of the drone is: ; in, Indicates the transmission power of the interference signal; Indicates the maximum transmit power; Indicates the minimum generated power; Indicates the UAV transmission power; Indicates high drone transmit power; Indicates low to medium UAV transmission power; Specifically, if a drone is detected transmitting power, the power of the interference signal is increased; otherwise, the power of the interference signal is reduced. This ensures that when a drone signal is detected, the strength of the interference signal is large enough to effectively interfere with the normal communication of the drone.

[0056] S4.2. Adjust the frequency of the interference signal according to the UAV radio wave frequency; Among them, the basis for adjusting the frequency of the interference signal according to the UAV radio wave frequency is: ; in, Indicates the frequency of the interference signal; Indicates the new frequency; Indicates the current frequency; Indicates the frequency of radio waves; Indicates medium and high drone radio wave frequencies; Indicates low drone radio wave frequency; Specifically, if the drone radio wave frequency is detected, the frequency of the interference signal is changed; otherwise, the current frequency is kept unchanged. In this way, when the drone signal is detected, its communication can be interfered with by changing the frequency, thus avoiding the normal transmission of the drone signal.

[0057] S4.3, generating a pseudo-random sequence as an interference signal through LFSR; ; in, represents the interference signal sequence; represents the feedback coefficient; Indicates the length of the register; Represents the modulo operation.

[0058] Specifically, the basic structure of LFSR includes a shift register and a feedback mechanism. The shift register consists of multiple registers (or storage cells), each of which stores one bit of data. The feedback mechanism generates new input values ​​based on the register values ​​at certain specific positions. These values ​​are combined through an exclusive OR (XOR) operation and then sent to the first bit of the shift register.

[0059] By generating a pseudo-random sequence, the complexity and uncertainty of the jamming signal can be increased, making it difficult for drones to predict and avoid the jamming signal, thereby improving the effectiveness of the jamming.

[0060] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. An artificial intelligence-based UAV remote control signal source interference control system, characterized by: include A signal acquisition unit (1), wherein the signal acquisition unit (1) captures a radio signal emitted by a remote controller of the drone via a radio receiving module; A signal processing unit (2), wherein the signal processing unit (2) performs data processing on the radio signal captured by the signal acquisition unit (1) by using a feature extraction algorithm; An identification and classification unit (3), wherein the identification and classification unit (3) classifies the radio signal by using an identification and classification model according to the radio signal processed by the signal processing unit (2); An interference control unit (4), wherein the interference control unit (4) designs an interference control decision rule according to the classification result of the identification and classification unit (3).

2. The artificial intelligence-based UAV remote control signal source interference control system according to claim 1 is characterized by: The signal acquisition unit (1) comprises a radio receiving module, a power identification module and a frequency identification module; Wherein, the radio receiving module collects radio signals through the SDR device; ； in, Indicates the received signal; represents the impulse response of the receiving device; represents the actual transmitted signal; Indicates the change of the signal on the time axis; Represents the relationship between signals at different time points; The power identification module is used to measure the power level of the received radio signal; The frequency identification module is used to identify the frequency used by the radio signal.

3. The artificial intelligence-based UAV remote control signal source interference control system according to claim 1 is characterized by: The specific steps of the signal processing unit (2) performing data processing on the radio signal captured by the signal acquisition unit (1) based on the feature extraction algorithm are as follows: S2.1, pre-processing the captured radio signals for denoising and normalization; S2.2, extract features from the preprocessed signal; S2.

3. Combine the extracted features into a feature vector for subsequent signal classification and recognition.

4. The artificial intelligence-based UAV remote control signal source interference control system according to claim 3 is characterized by: In S2.1, the specific process of preprocessing the captured radio signal for denoising and normalization is as follows: Use a low-pass filter to remove high-frequency noise; ； in, represents the filtered signal; represents the original signal; represents the impulse response of the low-pass filter; Represents the convolution operation; The convolution operation is expressed as: ； in, Represents the index in the convolution operation; Indicates infinite value; Indicates the input signal at time The value of represents the impulse response of the system, that is, at time Relative to The displacement value of Normalize the signal to a fixed range; ； in, represents the normalized signal; Represents the original signal The minimum value of Represents the original signal The maximum value of .

5. The artificial intelligence-based UAV remote control signal source interference control system according to claim 3 is characterized by: In S2.2, the specific process of extracting features from the preprocessed signal is as follows: The features are divided into three categories for extraction: Time domain characteristics: Compute the mean of a signal: ； in, represents the average value of the signal; Indicates the length of the signal; Represents an index variable; Compute the variance of a signal: ； in, represents the variance of the signal; Calculate the difference between the maximum and minimum values ​​of a signal: ； in, Indicates the maximum value of the signal With minimum difference; Count the number of times a signal crosses zero: ； in, Indicates the number of times the signal crosses zero; represents a symbolic function; express The next signal data point of Frequency domain characteristics: Convert a time domain signal to a frequency domain signal: ； in, represents a frequency domain signal; represents the frequency index; represents the scale coefficient; Compute the total energy of the spectrum: ； in, Represents the total energy of the spectrum; Calculate the entropy of a spectrum: ； in, represents the entropy of the spectrum; No. The normalized energy of the frequency components; Wavelet transform features: Decompose the signal into wavelet coefficients of different scales: ； in, Represents the coefficients after wavelet transformation; represents the wavelet basis function; Compute the energy of wavelet coefficients at each scale: ； in, Indicates The energy of the wavelet coefficients at scale.

6. The artificial intelligence-based UAV remote control signal source interference control system according to claim 1 is characterized by: The recognition and classification unit (3) constructs a recognition and classification model based on the SVM algorithm. The recognition and classification model construction process is: S3.1, creating a data set according to the feature vector of the signal processing unit (2); S3.2, divide the data set into training set and test set; S3.3, train the recognition and classification model through the SVM algorithm; S3.

4. Use the test set to test and evaluate the performance of the model.

7. The artificial intelligence-based UAV remote control signal source interference control system according to claim 6 is characterized by: In S3.3, the specific training steps of training the recognition classification model by the SVM algorithm are: S3.31, Use Gaussian kernel function to deal with nonlinear classification problems; Among them, the specific expression of Gaussian kernel function is: ； in, represents the Gaussian kernel function; Indicates The feature vector of samples; Indicates The feature vector of samples; Represents the parameters of the kernel function, controlling the width of the Gaussian kernel; represents the natural exponential function; S3.

32. Construct an optimization problem to maximize the interval between two types of data; The optimization problem is expressed as: ； ； ； in, represents the weight vector; represents the bias term; represents the slack variable; represents the regularization parameter; represents the total number of samples; Indicates The labels of the samples; S3.33, transform the optimization problem into a dual problem; The dual problem is expressed as: ； ； ； in, Indicates Lagrange multiplier of samples; Indicates Lagrange multiplier of samples; Indicates The labels of the samples; S3.

34. Use the SMO algorithm to solve the dual problem and obtain the optimal Lagrange multiplier .

8. The artificial intelligence-based UAV remote controller signal source interference control system according to claim 7 is characterized by: In S3.34, the SMO algorithm is used to solve the dual problem and obtain the optimal Lagrange multiplier The specific steps are: Initialize all Lagrange multipliers and select two Lagrange multipliers and Optimize During the optimization process, add random perturbations: ； ； in, and ; represents the standard deviation, which is used to control the intensity of the disturbance; calculate The border and : ； ； Calculate the error between two samples and : ； ； renew : ； in, Denotes the new Lagrange multiplier ; represents the second-order derivative of the kernel function; Will Adjust to Bounds and border between: ； renew : ； in, Denotes the new Lagrange multiplier ; Update Threshold : ； ； in, represents the Lagrange multiplier The threshold value of represents the Lagrange multiplier The threshold value of represents the original threshold; Select a new threshold : ； in, represents the new threshold of the Lagrange multiplier; Check whether the convergence condition is met by checking whether the change of the Lagrange multiplier is less than the threshold. If the convergence condition is not met, repeat the above steps; After the convergence condition is met, the optimal Lagrange multiplier is obtained by substituting the solution to the dual problem .

9. The artificial intelligence-based UAV remote controller signal source interference control system according to claim 7, characterized in that: In S3.4, the specific steps of using the test set to test the performance of the evaluation model are: S3.

41. According to the optimal Lagrange multiplier , calculate the weight vector and bias term; Among them, the specific expressions for calculating the weight vector and bias term are: ； ； in, represents the support vector; Labels representing support vectors; S3.42, predict new samples; Among them, the specific expression for predicting new samples is: ； ； in, Represents the new sample feature vector; represents the decision function; Represents the predicted label of the new sample; S3.

43. Use the test set to evaluate the performance of the model and obtain the evaluation indicators.

10. The artificial intelligence-based UAV remote control signal source interference control system according to claim 1, characterized in that: The interference control unit (4) designs an interference control decision rule according to the classification result of the identification and classification unit (3), that is, the interference control decision rule is specifically: S4.

1. Adjust the transmission power of the interference signal according to the transmission power of the UAV; Among them, the basis for adjusting the transmission power of the interference signal according to the transmission power of the drone is: ； in, Indicates the transmission power of the interference signal; Indicates the maximum transmit power; Indicates the minimum generated power; Indicates the UAV transmission power; Indicates high drone transmit power; Indicates low to medium UAV transmission power; S4.

2. Adjust the frequency of the interference signal according to the UAV radio wave frequency; Among them, the basis for adjusting the frequency of the interference signal according to the frequency of the drone radio wave is: ； in, Indicates the frequency of the interference signal; Indicates the new frequency; Indicates the current frequency; Indicates the frequency of radio waves; Indicates medium and high drone radio wave frequencies; Indicates low drone radio wave frequency; S4.3, generating a pseudo-random sequence as an interference signal through LFSR; ； in, represents the interference signal sequence; represents the feedback coefficient; Indicates the length of the register; Represents the modulo operation.

Images