A method and device for intelligently receiving communication signals

Abstract

The present invention discloses a method and device for intelligently receiving communication signals. The method comprises: constructing a collection of communication data from a communication area; performing cluster analysis on the collection of communication data to obtain a data classification model and a communication data set for each communication type; utilizing the communication data set for each communication type to obtain a signal demodulation and detection model corresponding to each communication type; utilizing the data classification model to classify the location information of a communication receiver and collected environmental electromagnetic signals to obtain the communication type of the communication receiver; utilizing the signal demodulation and detection model corresponding to the communication type of the communication receiver to demodulate and detect the communication signal received by the communication receiver, the location information of the communication receiver, and the environmental electromagnetic signals to obtain a detection signal. The present invention solves the problem of achieving highly reliable information transmission and accurate information reception in dense multipath and complex electromagnetic scenarios.

Description

Technical Field

[0001] The present invention relates to the fields of communications and artificial intelligence, and in particular to a method and device for intelligently receiving communication signals. Background Art

[0002] With the advancement of wireless communication technology in the IoT era, more and more communication terminals will be widely used. Communication in dense multipath scenarios, such as indoor environments, will become increasingly important. The ubiquity of wireless devices has led to an increasingly complex electromagnetic environment for communication. Achieving highly reliable information transmission and accurate reception in dense multipath and complex electromagnetic environments is an urgent challenge. Summary of the Invention

[0003] The present invention mainly solves the problem of how to achieve highly reliable information transmission and accurate information reception in dense multipath and complex electromagnetic scenarios. The present invention discloses a method and device for intelligent reception of communication signals.

[0004] In a first aspect of an embodiment of the present application, a method for intelligently receiving a communication signal is disclosed, comprising:

[0005] S1, constructing a communication data set of a communication area; the communication data set includes communication node data information; the communication node data information includes location information, standard signals, received signals, and environmental electromagnetic signals of the communication nodes; the communication nodes are nodes that receive communication signals within the communication area; the standard signals in the data information of all communication nodes are the same;

[0006] S2, performing cluster analysis on the total communication data set to obtain a data classification model and a communication data set for each communication type;

[0007] S3, using the communication data set of each communication type, training the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type;

[0008] S4, using a data classification model, classifying and processing the location information of the communication receiver and the collected environmental electromagnetic signals to obtain a communication type of the communication receiver;

[0009] S5, using a signal demodulation detection model corresponding to the communication type of the communication receiver, demodulate and detect the communication signal received by the communication receiver, the position information of the communication receiver, and the environmental electromagnetic signal to obtain a detection signal.

[0010] The cluster analysis process is performed on the total communication data set to obtain a data classification model and a communication data set of each communication type, including:

[0011] Representing the location information of the communication node and the environmental electromagnetic signal of each communication node data information in the communication data collection as a node vector; the elements of the node vector are the location information and the environmental electromagnetic signal of the communication node;

[0012] Perform cluster analysis on all node vectors to obtain the type information of node vectors and the node vectors contained in each type;

[0013] Using the communication node data information corresponding to all node vectors of the same type, a communication data set of the corresponding communication type is constructed;

[0014] For all node vectors of the same type, center point calculation and boundary value calculation processing are performed respectively to obtain the midpoint value and boundary value of the communication type corresponding to the type;

[0015] A data classification model is constructed using the midpoint values and boundary values of all communication types.

[0016] The expression for calculating the center point is:

[0017]

[0018] in, is the jth element of the i-th node vector of the same type, N is the number of node vectors contained in the type, and are respectively a preset first weighting factor and a preset second weighting factor, The jth element of the center point of the communication type corresponding to the type;

[0019] The boundary value calculation expression is:

[0020] ,

[0021] in, is the range value of the jth element of all node vectors of the same type, 、 and are the mean, variance and median values of the jth element of all node vectors of the same type, is the boundary value of the j-th element of the node vector of the communication type corresponding to the type.

[0022] The discriminant expression of the data classification model is:

[0023] ,

[0024] in, is the jth element of the node vector to be classified, M is the total number of elements contained in the node vector, is the discriminant expression corresponding to the j-th element of the node vector;

[0025] When each element of the node vector to be classified satisfies the corresponding discriminant expression, it is determined that the node vector to be classified belongs to the communication type corresponding to the midpoint value and the boundary value.

[0026] The signal demodulation detection model includes: a signal feature extraction network, a signal feature detection network and a signal feature fusion network;

[0027] The signal feature extraction network includes a first input module, a first two-dimensional convolution module, a first activation module, a first residual module, and a first normalization module; the input end of the first input module is the input end of the signal feature extraction network, and is used to receive input data; in the signal feature extraction network, the first input module, the first two-dimensional convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence; the output end of the first normalization module is the output end of the signal feature extraction network, and is connected to the input end of the second input module;

[0028] The signal feature detection network includes a second input module, a random permutation module, a second two-dimensional convolution module, a second activation module, and a second residual module; the input end of the signal feature detection network is the input end of the second input module; the output end of the signal feature detection network is the output end of the second residual module, which is connected to the input end of the third input module; in the signal feature detection network, the second input module, the random permutation module, the second two-dimensional convolution module, the second activation module, and the second residual module are connected in sequence;

[0029] The signal feature fusion network includes: a third input module, a first convolution module, a second convolution module, a third convolution module, a first pooling module, a fourth convolution module, a first fully connected module, and a second fully connected module;

[0030] The input end of the third input module of the signal feature fusion network is connected to the output end of the second residual module; the output end of the third input module of the signal feature fusion network is connected to the input end of the first convolution module of the signal feature fusion network; the output end of the first convolution module of the signal feature fusion network is connected to the input end of the second convolution module of the signal feature fusion network; the output end of the second convolution module of the signal feature fusion network is connected to the input end of the third convolution module of the signal feature fusion network; the output end of the third convolution module of the signal feature fusion network is connected to the input end of the first pooling module of the signal feature fusion network; the output end of the first pooling module of the signal feature fusion network is connected to the input end of the fourth convolution module of the signal feature fusion network; the output end of the fourth convolution module of the signal feature fusion network is connected to the input end of the first fully connected module of the signal feature fusion network; the output end of the first fully connected module of the signal feature fusion network is connected to the input end of the second fully connected module of the signal feature fusion network; the output end of the second fully connected module is used to output a detection signal for the input signal.

[0031] The method of training a preset signal demodulation detection model using a communication data set of each communication type to obtain a signal demodulation detection model corresponding to each communication type includes:

[0032] Initializing a training iteration count for a communication data set of a communication type; during the training process, using the location information, received signal, and environmental electromagnetic signal of the communication data set corresponding to the communication type as training data; and using the standard signal corresponding to the training data as label information;

[0033] Using the training data in the communication data set as input data, inputting it into a signal demodulation detection model corresponding to the communication type;

[0034] Processing the input data using the signal demodulation detection model to obtain a detection signal;

[0035] Performing difference calculation processing on the obtained detection signal and the label information corresponding to the input data to obtain a difference value;

[0036] Determine whether the difference value meets the convergence condition, and obtain a first determination result;

[0037] When the first judgment result is no, determining whether the number of training iterations is equal to a training number threshold, and obtaining a second judgment result;

[0038] When the second judgment result is no, determining that the model training state does not meet the training termination condition;

[0039] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;

[0040] When the first judgment result is yes, determining that the model training state satisfies the training termination condition;

[0041] When the model training state does not meet the training termination condition, the signal demodulation detection model is updated with the parameter update model, the number of training iterations is increased by 1, and the training data in the communication data set is triggered to be input into the signal demodulation detection model as input data;

[0042] When the model training state satisfies the training termination condition, the training process of the signal demodulation detection model is completed to obtain a trained signal demodulation detection model;

[0043] The communication data set of each communication type is used to train the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type.

[0044] The parameter update model is:

[0045] ,

[0046] ;

[0047] Where, is the difference value calculated for the i-th training data in the communication data set, Update the value for the parameter, are the parameters of the signal demodulation detection model, is the initial parameter learning rate, is the momentum angle parameter, , Indicates that for variables Find the partial derivative.

[0048] In a second aspect of an embodiment of the present application, a smart receiving device for communication signals is disclosed, the device comprising:

[0049] a memory storing executable program code;

[0050] a processor coupled to the memory;

[0051] The processor calls the executable program code stored in the memory to execute the intelligent receiving method of the communication signal.

[0052] In a third aspect of an embodiment of the present application, a computer-storable medium is disclosed, wherein the computer-storable medium stores computer instructions. When the computer instructions are called by a computer, they are used to execute the intelligent receiving method of the communication signal.

[0053] In a fourth aspect of the embodiments of the present application, an information data processing terminal is disclosed, which is used to implement the intelligent receiving method of the communication signal.

[0054] The beneficial effects of the present invention are:

[0055] The present invention discloses an intelligent receiving method and device for communication signals, which solves the problem of how to achieve highly reliable information transmission and accurate information reception in dense multipath and complex electromagnetic scenarios.

[0056] Before conducting formal communication, the present invention first constructs a total set of communication data in the communication area, obtains the electromagnetic signal of each communication node and the received signal for the standard signal; by classifying the received signal and environmental signal of each node, a communication classification model is established. A corresponding signal demodulation and detection model is constructed for each type of scenario to improve the matching of demodulation and scenario.

[0057] The present invention establishes a signal demodulation detection model and utilizes three networks, namely a signal feature extraction network, a signal feature detection network and a signal feature fusion network, to realize signal demodulation detection, thereby achieving accuracy and real-time performance of demodulation detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 4 is an implementation flow chart of the method of the present invention. DETAILED DESCRIPTION

[0059] In order to better understand the content of the present invention, an embodiment is given here.

[0060] Figure 1 4 is an implementation flow chart of the method of the present invention.

[0061] In a first aspect of an embodiment of the present application, a method for intelligently receiving a communication signal is disclosed, comprising:

[0062] S1, constructing a communication data collection of a communication area; the communication data collection includes communication node data information; the communication node data information includes location information, standard signal, received signal, and environmental electromagnetic signal of the communication node; the communication node is a node that receives communication signals within the communication area; the standard signal is a signal sent to a communication receiver at the communication node location; the received signal is a signal corresponding to the standard signal received by the communication receiver at the communication node location; the electromagnetic signal is an environmental electromagnetic signal collected by the communication receiver at the communication node location; the standard signal in all communication node data information is the same; the environmental electromagnetic signal is an electromagnetic signal of the background environment collected by the communication receiver of the communication node when no signal is received;

[0063] The standard signal in the communication data centralization is a unified information signal sent by nodes within the communication area;

[0064] The received signals in the communication data set are received signals corresponding to the standard signals of the nodes in the communication area.

[0065] S2, performing cluster analysis on the total communication data set to obtain a data classification model and a communication data set for each communication type;

[0066] S3, using the communication data set of each communication type, training the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type;

[0067] S4, using a data classification model, classifying and processing the location information of the communication receiver and the collected environmental electromagnetic signals to obtain a communication type of the communication receiver;

[0068] S5, using a signal demodulation detection model corresponding to the communication type of the communication receiver, performing demodulation detection processing on the communication signal received by the communication receiver to obtain a detection signal;

[0069] The cluster analysis process is performed on the total communication data set to obtain a data classification model and a communication data set of each communication type, including:

[0070] Representing the location information of the communication node and the environmental electromagnetic signal of each communication node data information in the communication data collection as a node vector; the elements of the node vector are the location information and the environmental electromagnetic signal of the communication node;

[0071] Perform cluster analysis on all node vectors to obtain the type information of node vectors and the node vectors contained in each type;

[0072] Using the communication node data information corresponding to all node vectors of the same type, a communication data set of the corresponding communication type is constructed;

[0073] For all node vectors of the same type, center point calculation and boundary value calculation processing are performed respectively to obtain the midpoint value and boundary value of the communication type corresponding to the type;

[0074] Using the midpoint values and boundary values of all communication types, a data classification model is constructed;

[0075] The expression for calculating the center point is:

[0076]

[0077] in, is the jth element of the i-th node vector of the same type, N is the number of node vectors contained in the type, and are respectively a preset first weighting factor and a preset second weighting factor, The jth element of the center point of the communication type corresponding to the type;

[0078] The center point calculation expression combines the geometric mean and weighted logarithm, taking into account the overall product and relative proportions of elements in the same node vector type. For complex data such as communication node location information and environmental electromagnetic signals, it can extract key features from multiple dimensions, avoiding the loss of important information from a single calculation method and more accurately depicting the core trends of the data distribution.

[0079] The center point calculation expression, through the preset first and second weighting factors, can flexibly adjust the contribution of the two calculation methods to the center point calculation based on the needs of actual communication scenarios. For example, in scenarios with drastic changes in the electromagnetic environment, the second weighting factor can be increased to highlight the sensitivity of the logarithmic operation to data fluctuations, making the center point more representative of the characteristics of the ambient electromagnetic signal. Conversely, when location information significantly affects signal reception, the first weighting factor can be increased to emphasize the geometric mean's comprehensive consideration of location distribution.

[0080] The logarithmic operation in the center point calculation expression introduces a denominator for normalization, effectively suppressing the interference of abnormal data on the center point calculation. In complex electromagnetic environments, even if anomalous electromagnetic signal data collected by individual communication nodes is present, the characteristics of logarithmic operation in this expression can reduce the impact of outliers, ensuring the stability and reliability of the center point calculation results, thereby improving the accuracy of the data classification model.

[0081] The boundary value calculation expression is:

[0082] ,

[0083] in, is the range value of the jth element of all node vectors of the same type, 、 and are the mean, variance and median values of the jth element of all node vectors of the same type, The boundary value of the j-th element of the node vector of the communication type corresponding to the type;

[0084] The boundary value calculation expression combines the data's range, mean, variance, and the maximum deviation of a single data point from a reference value, comprehensively describing the distribution boundaries of node vector elements of the same type from multiple perspectives. The range reflects the degree of data dispersion, the mean and variance reflect the data's central tendency and fluctuation, and the maximum deviation captures the data's extreme changes. This multi-indicator approach can accurately define the boundary range of communication data and avoid misclassification.

[0085] The boundary value calculation expression incorporates the data mean and variance ratio, allowing it to adaptively adjust the boundary value based on the data's volatility. In complex and changing electromagnetic environments, when the variance increases, indicating increased data fluctuations, the mean and variance ratio adjust accordingly, allowing the boundary value to dynamically expand and accommodate more signal variations. Conversely, when the data is relatively stable, the boundary value can be tightened, improving classification accuracy and enabling the data classification model to better adapt to diverse communication environments. By taking the maximum deviation value in the calculation, the boundary value ensures that all possible extreme data conditions are covered. In dense multipath and complex electromagnetic scenarios, even rare signal interference or abnormal received data can be accommodated by the boundary value determined by this expression, preventing these extreme conditions from leading to misclassification of communication types, effectively enhancing the robustness and fault tolerance of the data classification model.

[0086] The discriminant expression of the data classification model is:

[0087] ,

[0088] in, is the jth element of the node vector to be classified, M is the total number of elements contained in the node vector, is the discriminant expression corresponding to the j-th element of the node vector; when each element of the node vector to be classified satisfies the corresponding discriminant expression, it is determined that the node vector to be classified belongs to the communication type corresponding to the midpoint value and the boundary value;

[0089] The signal demodulation detection model includes: a signal feature extraction network, a signal feature detection network and a signal feature fusion network;

[0090] The signal feature extraction network includes a first input module, a first two-dimensional convolution module, a first activation module, a first residual module, and a first normalization module; the input end of the first input module is the input end of the signal feature extraction network, which is used to receive input data; in the signal feature extraction network, the first input module, the first two-dimensional convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence; the output port of the first normalization module is the output port of the signal feature extraction network;

[0091] The signal feature detection network includes a second input module, a random permutation module, a second two-dimensional convolution module, a second activation module, and a second residual module; the input of the signal feature detection network is the input end of the second input module; the output of the signal feature detection network is the output end of the second residual module; in the signal feature detection network, the second input module, the random permutation module, the second two-dimensional convolution module, the second activation module, and the second residual module are connected in sequence;

[0092] The signal feature fusion network includes: a third input module, a first convolution module, a second convolution module, a third convolution module, a first pooling module, a fourth convolution module, a first fully connected module, and a second fully connected module;

[0093] The input end of the third input module of the signal feature fusion network is connected to the output end of the second residual module; the output end of the third input module of the signal feature fusion network is connected to the input end of the first convolution module of the signal feature fusion network; the output end of the first convolution module of the signal feature fusion network is connected to the input end of the second convolution module of the signal feature fusion network; the output end of the second convolution module of the signal feature fusion network is connected to the input end of the third convolution module of the signal feature fusion network; the output end of the third convolution module of the signal feature fusion network is connected to the input end of the first pooling module of the signal feature fusion network; the output end of the first pooling module of the signal feature fusion network is connected to the input end of the fourth convolution module of the signal feature fusion network; the output end of the fourth convolution module of the signal feature fusion network is connected to the input end of the first fully connected module of the signal feature fusion network; the output end of the first fully connected module of the signal feature fusion network is connected to the input end of the second fully connected module of the signal feature fusion network. The output end of the second fully connected module is used to output a detection signal of the input signal;

[0094] The cluster analysis may adopt density-based clustering, graph clustering, spectral clustering, etc.

[0095] The activation module is used to implement an activation function; the activation function may be a sigmoid function;

[0096] The residual module can be implemented using a Resnet network;

[0097] The normalization module can be implemented using the Quantile Normalization function in php.

[0098] The random permutation module is implemented using the shuffle function in Python.

[0099] The ninth activation module is implemented using a tanh function;

[0100] The calculation formula of the quantization module is:

[0101] ,

[0102] Wherein, X is the input of the quantization module.

[0103] The method of training a preset signal demodulation detection model using a communication data set of each communication type to obtain a signal demodulation detection model corresponding to each communication type includes:

[0104] Initializing a training iteration count for a communication data set of a communication type; during the training process, using the location information, received signal, and environmental electromagnetic signal of the communication data set as training data; and using the standard signal corresponding to the training data as label information;

[0105] Using the training data in the communication data set as input data, inputting it into a signal demodulation detection model corresponding to the communication type;

[0106] Processing the input data using the signal demodulation detection model to obtain a detection signal;

[0107] Performing difference calculation processing on the obtained detection signal and the label information corresponding to the input data to obtain a difference value;

[0108] Determine whether the difference value meets the convergence condition, and obtain a first determination result;

[0109] When the first judgment result is no, determining whether the number of training iterations is equal to a training number threshold, and obtaining a second judgment result;

[0110] When the second judgment result is no, determining that the model training state does not meet the training termination condition;

[0111] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;

[0112] When the first judgment result is yes, determining that the model training state satisfies the training termination condition;

[0113] When the model training state does not meet the training termination condition, the signal demodulation detection model is updated with the parameter update model, the number of training iterations is increased by 1, and the training data in the communication data set is triggered to be input into the signal demodulation detection model as input data;

[0114] When the model training state satisfies the training termination condition, the training process of the signal demodulation detection model is completed to obtain a trained signal demodulation detection model.

[0115] The communication data set of each communication type is used to train the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type.

[0116] The difference value satisfies the convergence condition, which means that the difference value is less than a preset convergence threshold, such as 0.5; the difference value does not satisfy the convergence condition, which means that the difference value is not less than the preset convergence threshold.

[0117] The difference calculation process can be implemented using a loss function.

[0118] The loss function may be a cross entropy loss function.

[0119] The parameter update model is:

[0120] ,

[0121] ,

[0122] Where, is the difference value calculated for the i-th training data in the communication data set, Update the value for the parameter, are the parameters of the signal demodulation detection model, is the initial parameter learning rate, is the momentum angle parameter, , Indicates that for variables Find the partial derivative.

[0123] The data classification model is used to classify the location information of the communication receiver and the collected environmental electromagnetic signals to obtain the communication type of the communication receiver, including:

[0124] The position information of the communication receiver and the collected environmental electromagnetic signal are expressed as a node vector; the elements of the node vector are the position information of the communication node and the environmental electromagnetic signal;

[0125] The node vectors are classified using a data classification model to obtain the communication type of the communication receiver.

[0126] The method utilizes a signal demodulation detection model corresponding to the communication type of the communication receiver to demodulate and detect the communication signal received by the communication receiver, the position information of the communication receiver, and the environmental electromagnetic signal to obtain a detection signal, including:

[0127] Obtaining a trained signal demodulation detection model corresponding to the communication type of the communication receiver;

[0128] The signal demodulation detection model is used to perform demodulation detection processing on the communication signal received by the communication receiver, the position information of the communication receiver and the environmental electromagnetic signal to obtain a detection signal.

[0129] The range value is the difference between the maximum value and the minimum value;

[0130] For the communication node data information corresponding to all node vectors of each type, the communication data set of the corresponding communication type is constructed respectively, thereby obtaining the communication data set of all communication types;

[0131] In a second aspect of an embodiment of the present application, a smart receiving device for communication signals is disclosed, the device comprising:

[0132] a memory storing executable program code;

[0133] a processor coupled to the memory;

[0134] The processor calls the executable program code stored in the memory to execute the intelligent receiving method of the communication signal.

[0135] In a third aspect of an embodiment of the present application, a computer-storable medium is disclosed, wherein the computer-storable medium stores computer instructions. When the computer instructions are called by a computer, they are used to execute the intelligent receiving method of the communication signal.

[0136] In a fourth aspect of the embodiments of the present application, an information data processing terminal is disclosed, which is used to implement the intelligent receiving method of the communication signal.

[0137] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for intelligently receiving communication signals, characterized in that: include: S1, construct the total communication data set of the communication area; The communication data collection includes communication node data information; The communication node data information includes the location information, standard signal, received signal, and environmental electromagnetic signal of the communication node; the communication node is a node that receives the communication signal within the communication area; The standard signal in the data information of all communication nodes is the same; S2, performing cluster analysis on the total communication data set to obtain a data classification model and a communication data set for each communication type; S3, using the communication data set of each communication type, training the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type; S4, using a data classification model, classifying and processing the location information of the communication receiver and the collected environmental electromagnetic signals to obtain a communication type of the communication receiver; S5, using a signal demodulation detection model corresponding to the communication type of the communication receiver, demodulating and detecting the communication signal received by the communication receiver, the position information of the communication receiver, and the environmental electromagnetic signal to obtain a detection signal; The cluster analysis process is performed on the total communication data set to obtain a data classification model and a communication data set of each communication type, including: Representing the location information of the communication node and the environmental electromagnetic signal of each communication node data information in the communication data collection as a node vector; the elements of the node vector are the location information and the environmental electromagnetic signal of the communication node; Perform cluster analysis on all node vectors to obtain the type information of node vectors and the node vectors contained in each type; Using the communication node data information corresponding to all node vectors of the same type, a communication data set of the corresponding communication type is constructed; For all node vectors of the same type, center point calculation and boundary value calculation processing are performed respectively to obtain the midpoint value and boundary value of the communication type corresponding to the type; Using the midpoint values and boundary values of all communication types, a data classification model is constructed; The expression for calculating the center point is: ， in, is the jth element of the i-th node vector of the same type, N is the number of node vectors contained in the type, and are respectively a preset first weighting factor and a preset second weighting factor, The jth element of the center point of the communication type corresponding to the type; The boundary value calculation expression is: ， in, is the range value of the jth element of all node vectors of the same type, 、 and are the mean, variance and median values of the jth element of all node vectors of the same type, is the boundary value of the j-th element of the node vector of the communication type corresponding to the type.

2. The intelligent receiving method for communication signals according to claim 1, wherein: The discriminant expression of the data classification model is: ， in, is the jth element of the node vector to be classified, M is the total number of elements contained in the node vector, is the discriminant expression corresponding to the j-th element of the node vector; When each element of the node vector to be classified satisfies the corresponding discriminant expression, it is determined that the node vector to be classified belongs to the communication type corresponding to the midpoint value and the boundary value.

3. The intelligent receiving method for communication signals according to claim 2, wherein: The signal demodulation detection model includes: a signal feature extraction network, a signal feature detection network and a signal feature fusion network; The signal feature extraction network includes a first input module, a first two-dimensional convolution module, a first activation module, a first residual module, and a first normalization module; the input end of the first input module is the input end of the signal feature extraction network, and is used to receive input data; in the signal feature extraction network, the first input module, the first two-dimensional convolution module, the first activation module, the first residual module, and the first normalization module are connected in sequence; the output end of the first normalization module is the output end of the signal feature extraction network, and is connected to the input end of the second input module; The signal feature detection network includes a second input module, a random permutation module, a second two-dimensional convolution module, a second activation module, and a second residual module; the input end of the signal feature detection network is the input end of the second input module; the output end of the signal feature detection network is the output end of the second residual module, which is connected to the input end of the third input module; in the signal feature detection network, the second input module, the random permutation module, the second two-dimensional convolution module, the second activation module, and the second residual module are connected in sequence; The signal feature fusion network includes: a third input module, a first convolution module, a second convolution module, a third convolution module, a first pooling module, a fourth convolution module, a first fully connected module, and a second fully connected module; The input end of the third input module of the signal feature fusion network is connected to the output end of the second residual module; the output end of the third input module of the signal feature fusion network is connected to the input end of the first convolution module of the signal feature fusion network; the output end of the first convolution module of the signal feature fusion network is connected to the input end of the second convolution module of the signal feature fusion network; the output end of the second convolution module of the signal feature fusion network is connected to the input end of the third convolution module of the signal feature fusion network; the output end of the third convolution module of the signal feature fusion network is connected to the input end of the first pooling module of the signal feature fusion network; the output end of the first pooling module of the signal feature fusion network is connected to the input end of the fourth convolution module of the signal feature fusion network; the output end of the fourth convolution module of the signal feature fusion network is connected to the input end of the first fully connected module of the signal feature fusion network; the output end of the first fully connected module of the signal feature fusion network is connected to the input end of the second fully connected module of the signal feature fusion network; the output end of the second fully connected module is used to output a detection signal for the input signal.

4. The intelligent receiving method for communication signals according to claim 3, wherein: The method of training a preset signal demodulation detection model using a communication data set of each communication type to obtain a signal demodulation detection model corresponding to each communication type includes: Initializing a training iteration count for a communication data set of a communication type; during the training process, using the location information, received signal, and environmental electromagnetic signal of the communication data set corresponding to the communication type as training data; and using the standard signal corresponding to the training data as label information; Using the training data in the communication data set as input data, inputting it into a signal demodulation detection model corresponding to the communication type; Processing the input data using the signal demodulation detection model to obtain a detection signal; Performing difference calculation processing on the obtained detection signal and the label information corresponding to the input data to obtain a difference value; Determine whether the difference value meets the convergence condition, and obtain a first determination result; When the first judgment result is no, determining whether the number of training iterations is equal to a training number threshold, and obtaining a second judgment result; When the second judgment result is no, determining that the model training state does not meet the training termination condition; When the second judgment result is yes, determining that the model training state satisfies the training termination condition; When the first judgment result is yes, determining that the model training state satisfies the training termination condition; When the model training state does not meet the training termination condition, the signal demodulation detection model is updated with the parameter update model, the number of training iterations is increased by 1, and the training data in the communication data set is triggered to be input into the signal demodulation detection model as input data; When the model training state satisfies the training termination condition, the training process of the signal demodulation detection model is completed to obtain a trained signal demodulation detection model; The communication data set of each communication type is used to train the preset signal demodulation detection model to obtain the signal demodulation detection model corresponding to each communication type.

5. The intelligent receiving method for communication signals according to claim 4, wherein: The parameter update model is: ， ； Where, is the difference value calculated for the i-th training data in the communication data set, Update the value for the parameter, are the parameters of the signal demodulation detection model, is the initial parameter learning rate, is the momentum angle parameter, , Indicates that for variables Find the partial derivative.

6. An intelligent receiving device for communication signals, characterized in that: The device comprises: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the intelligent receiving method for communication signals according to any one of claims 1 to 5.

7. A computer storable medium, characterized in that The computer storable medium stores computer instructions, and when the computer instructions are called by a computer, they are used to execute the intelligent receiving method for communication signals according to any one of claims 1 to 5.

Images