An intelligent control method for space electromagnetic environment based on similarity machine learning

Abstract

This invention provides a method and system for intelligently controlling a spatial electromagnetic environment based on similarity machine learning, relating to the field of adaptive electromagnetic environment control technology. The method includes: analyzing target electromagnetic environment simulation requirements to obtain the target electromagnetic environment; establishing similarity meta-sample information; generating control rules; simulating and generating a simulated electromagnetic environment that meets the target; monitoring and learning similarity deviations in real time; modifying the control rules and converting them into similarity control variables; adjusting the operating states of the simulation units to generate a new electromagnetic environment, and achieving dynamic regulation. This invention implements intelligent planning and adaptive control of the electromagnetic environment, overcoming the limitations of traditional control methods.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic environment adaptive control, and in particular to an intelligent control method and system for a space electromagnetic environment based on similarity machine learning. Background Art

[0002] With the rapid development of global wireless communication technology, the wireless communication industry based on electromagnetic spectrum has continued to develop. It is not only widely used in important industries such as telecommunications, aviation, railways, and media, but also plays an important role in key areas such as public safety, major event security, and even national defense construction. This has led to an increasing dependence of the entire society on radio spectrum resources, and has also put forward higher requirements on the service level provided by radio management departments for social and economic development and national defense construction.

[0003] However, traditional planning and control methods for large-scale space electromagnetic environments are limited to remote control of electromagnetic simulation equipment, failing to fully realize pre-intelligent planning and real-time adaptive control of the electromagnetic environment, and unable to complete one-click intelligent construction of the electromagnetic environment. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide an intelligent control method for the spatial electromagnetic environment based on similarity machine learning, which solves the problem that traditional electromagnetic environment planning and control means are limited to remote control of electromagnetic simulation equipment, fail to fully realize pre-intelligent planning and real-time adaptive control of the electromagnetic environment, and cannot complete one-click intelligent construction of the electromagnetic environment.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] An intelligent control method for a space electromagnetic environment based on similarity machine learning, comprising:

[0007] Receiving a simulation requirement of a target electromagnetic environment, and analyzing the simulation requirement to obtain an electromagnetic environment target;

[0008] Generate control rules according to the electromagnetic environment target, and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment;

[0009] Generate a simulated electromagnetic environment that meets the electromagnetic environment target based on control rules simulation;

[0010] Real-time monitoring and collection of simulated electromagnetic environment monitoring data, performing similarity learning based on the monitoring data, and generating specific similarity deviations between the simulated electromagnetic environment and the target electromagnetic environment;

[0011] Modifying the control rule by using the similarity deviation and similarity element sample information, and converting the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment;

[0012] According to the correction amount of the similarity control quantity, the working status of various simulation units is readjusted to generate a new electromagnetic environment, thereby completing the dynamic regulation of the simulated electromagnetic environment.

[0013] Preferably, receiving a simulation requirement of a target electromagnetic environment and parsing the simulation requirement to obtain an electromagnetic environment target includes:

[0014] Format conversion and verification are performed on the received simulation requirements to obtain simulation data that conforms to a preset format and range;

[0015] Extracting key information of the simulation data; the key information includes: frequency range, target area, environmental characteristics and simulation time;

[0016] Normalizing the key information to obtain normalized data;

[0017] Mapping the normalized data with features in a preset electromagnetic environment model to obtain mapping parameters; the features in the preset electromagnetic environment model include: electromagnetic interference sources, propagation paths, and boundary conditions in the environment;

[0018] Generating an electromagnetic environment target data structure according to the mapping parameters; the electromagnetic environment target data structure includes key information for generating and controlling a simulated electromagnetic environment;

[0019] The electromagnetic environment target is determined according to the electromagnetic environment target data structure.

[0020] Preferably, generating control rules according to the electromagnetic environment target and establishing similar meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment include:

[0021] Analyze the electromagnetic environment targets to clarify the attributes of the parameters in each of the electromagnetic environment targets and the conditions for the parameters to meet;

[0022] Constructing control rules based on the attributes and parameters meeting the conditions; the control rules include: spatial domain control rules, time domain control rules, frequency domain control rules, energy control rules and modulation domain control rules;

[0023] Collect actual electromagnetic environment data to build an actual electromagnetic environment dataset;

[0024] Extract key feature information from the actual electromagnetic environment data set; the key feature information includes signal strength, path loss, reflection and attenuation characteristics

[0025] A meta-sample library with similar electromagnetic environments is established using the extracted key feature information; each sample in the meta-sample library includes key feature information and corresponding specific environmental condition labels; and each sample is the similar meta-sample information.

[0026] Preferably, generating a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rule simulation includes:

[0027] Initializing parameters of an electromagnetic environment simulation model according to the control rules; the parameters of the simulation model include electromagnetic source parameters, spatial distribution parameters, time parameters and boundary conditions;

[0028] Running the electromagnetic environment simulation model and generating a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rules;

[0029] During the simulation process, the simulated electromagnetic environment is corrected using the similar meta-sample information, so that the simulated electromagnetic environment has realism and accuracy.

[0030] Preferably, real-time monitoring and collection of simulated electromagnetic environment monitoring data, performing similarity learning based on the monitoring data, and generating a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment include:

[0031] Get the characteristic vector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i ;

[0032] Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Euclidean distance between them is calculated as follows: Among them, T i =[T i1 ,T i2 ,…,T in ] and S i =[S i1 ,S i2 ,…,S in ] are the feature vectors of the i-th monitoring point, and n is the dimension of the feature vector;

[0033] Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The cosine similarity between them is calculated as follows:

[0034] Calculate the eigenvector S of the simulated electromagnetic environment monitoring datai and the characteristic vector T of the target electromagnetic environment i The Manhattan distance between them is:

[0035] The specific similarity deviation is determined according to the comprehensive similarity deviation formula; the comprehensive similarity deviation formula is: d overall (T i ,S i )=α·d euclid (T i ,S i )+β·(1-cos(T i ,S i ))+γ·d manhattan (T i ,S i ); where α, β, and γ are the weight coefficients of the corresponding items, d overal (T i ,S i ) is the specific similarity deviation.

[0036] Preferably, the control rule is modified by using the similarity deviation and similarity sample information, and the modified control rule is converted into a correction amount for the similarity control amount of the simulated electromagnetic environment, including:

[0037] Constructing a similar feature vector and a target deviation vector based on the similar meta-sample information;

[0038] Inputting the similarity feature vector and the target deviation vector into a parameter regression model to fit the parameters of the control rule with the similarity deviation to obtain the modified control rule;

[0039] Based on the constructed correction model, the correction amount of each parameter of the control rule is calculated; the calculation formula of the correction amount is: C i =W·ΔT i Among them, C i is the correction amount of the i-th monitoring point, W is the weight of the correction model, ΔT i It is the deviation between the target electromagnetic environment and the simulated electromagnetic environment.

[0040] Preferably, according to the correction amount of the similarity control amount, the working state of each simulation unit is readjusted to generate a new electromagnetic environment, thereby completing the dynamic regulation of the simulated electromagnetic environment, including:

[0041] adjusting each simulation unit based on the correction amount to determine a new operating state of each simulation unit according to the correction amount;

[0042] A new simulation experiment is run based on the new working state to generate new simulated electromagnetic environment characteristic data, thereby completing the dynamic regulation of the simulated electromagnetic environment.

[0043] An intelligent control method for a space electromagnetic environment based on similarity machine learning, comprising:

[0044] The parsing unit is configured to receive a simulation requirement of a target electromagnetic environment and parse the simulation requirement to obtain an electromagnetic environment target;

[0045] A rule generating unit, configured to generate control rules according to the electromagnetic environment target and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment;

[0046] An environment generation unit, configured to generate a simulated electromagnetic environment that meets the electromagnetic environment target according to control rules;

[0047] a deviation generating unit, configured to monitor and collect simulated electromagnetic environment monitoring data in real time, perform similarity learning based on the monitoring data, and generate a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment;

[0048] a correction amount determination unit, configured to modify the control rule by using the similarity deviation and similarity element sample information, and convert the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment;

[0049] The control unit is used to readjust the working status of various simulation units according to the correction amount of the similarity control amount, generate a new electromagnetic environment, and complete the dynamic control of the simulated electromagnetic environment.

[0050] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0051] The present invention provides an intelligent control method and system for a space electromagnetic environment based on similarity machine learning. The method comprises: receiving a simulation requirement for a target electromagnetic environment, parsing the simulation requirement to obtain an electromagnetic environment target; generating control rules based on the electromagnetic environment target and establishing similarity meta-sample information between the simulated electromagnetic environment and the actual electromagnetic environment; simulating and generating a simulated electromagnetic environment that meets the electromagnetic environment target based on the control rules; monitoring and collecting simulated electromagnetic environment monitoring data in real time, performing similarity learning based on the monitoring data, and generating a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment; modifying the control rules based on the similarity deviation and utilizing the similarity meta-sample information, and converting the modified control rules into correction values ​​for the similarity control values ​​of the simulated electromagnetic environment; and readjusting the operating states of various simulation units based on the correction values ​​to generate a new electromagnetic environment, thereby achieving dynamic control of the simulated electromagnetic environment. Compared with traditional electromagnetic environment control methods, the present invention solves the problem that traditional large-scale space electromagnetic environment planning and control methods are limited to remote control of electromagnetic simulation equipment, fail to fully realize pre-intelligent planning and real-time adaptive control of the electromagnetic environment, and cannot complete one-click intelligent construction of the electromagnetic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 A flow chart of a method provided by an embodiment of the present invention;

[0054] Figure 2 A schematic diagram of the technical route provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0057] like Figure 1 and Figure 2As shown, the present invention provides an intelligent control method for a space electromagnetic environment based on similarity machine learning, comprising:

[0058] Step 100: receiving a simulation requirement of a target electromagnetic environment, and analyzing the simulation requirement to obtain an electromagnetic environment target;

[0059] Step 200: Generate control rules according to the electromagnetic environment target, and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment;

[0060] Step 300: Generate a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rule simulation;

[0061] Step 400: monitoring and collecting simulated electromagnetic environment monitoring data in real time, performing similarity learning based on the monitoring data, and generating a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment;

[0062] Step 500: modifying the control rule by using the similarity deviation and similarity element sample information, and converting the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment;

[0063] Step 600: According to the correction amount of the similarity control amount, the working state of each type of simulation unit is readjusted to generate a new electromagnetic environment, thereby completing the dynamic regulation of the simulated electromagnetic environment.

[0064] Preferably, receiving a simulation requirement of a target electromagnetic environment and parsing the simulation requirement to obtain an electromagnetic environment target includes:

[0065] Format conversion and verification are performed on the received simulation requirements to obtain simulation data that conforms to a preset format and range;

[0066] Extracting key information of the simulation data; the key information includes: frequency range, target area, environmental characteristics and simulation time;

[0067] Normalizing the key information to obtain normalized data;

[0068] Mapping the normalized data with features in a preset electromagnetic environment model to obtain mapping parameters; the features in the preset electromagnetic environment model include: electromagnetic interference sources, propagation paths, and boundary conditions in the environment;

[0069] Generating an electromagnetic environment target data structure according to the mapping parameters; the electromagnetic environment target data structure includes key information for generating and controlling a simulated electromagnetic environment;

[0070] The electromagnetic environment target is determined according to the electromagnetic environment target data structure.

[0071] Specifically, the target environment determination method of this embodiment includes:

[0072] Step 101: Design a requirements input interface. In this embodiment, a user interface or API is designed to enable users or other systems to submit electromagnetic environment simulation requirements. Exemplarily, the user interface or API in this embodiment is a web form, command line tool, API, or other suitable interaction method.

[0073] Step 102: Requirements Formatting and Verification. This embodiment formats the input electromagnetic environment simulation requirements to ensure consistency and integrity. It also performs preliminary verification to check whether the input conforms to the expected format and range. For example, this embodiment can check parameters such as spectrum range, spatial coordinates, and time period.

[0074] Step 103: This embodiment is provided with a requirement parsing module, which is responsible for parsing the formatted simulation requirements into specific electromagnetic environment target parameters. The parsing process involves the following sub-steps:

[0075] Step 1031: Extract core parameters from the requirements, including frequency range, target area, environmental characteristics (such as antenna layout, obstacle type, etc.), simulation time, etc.

[0076] Step 1032: normalize the extracted parameters to adapt them to the standard format required for internal processing.

[0077] Step 1033: Map the user's needs with the features in the internal electromagnetic environment model, where the features include electromagnetic interference sources, propagation paths, boundary conditions, etc. in the environment.

[0078] Step 1034: Generate an electromagnetic environment target data structure based on the parsed parameters, which includes all key information required to generate and control the simulated electromagnetic environment.

[0079] Step 104: Feedback the analyzed electromagnetic environment target to the user or initiating system for confirmation or further instructions. Ensure that the analysis results are consistent with user expectations and make adjustments if necessary.

[0080] Step 105: This embodiment also handles abnormal situations that may occur during the parsing process, such as missing parameters, unreasonable range values, etc., and provides error prompts and suggested repair solutions.

[0081] Step 106: This embodiment logs the entire demand analysis process, including input data, processing steps, intermediate results, and the final analyzed electromagnetic environment target, to facilitate subsequent tracking and problem troubleshooting.

[0082] This embodiment generates a detailed electromagnetic environment target through a series of processes such as interface design, data verification, information extraction and mapping, providing a solid foundation for subsequent simulation generation and regulation.

[0083] Preferably, generating control rules according to the electromagnetic environment target and establishing similar meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment include:

[0084] Analyze the electromagnetic environment targets to clarify the attributes of the parameters in each of the electromagnetic environment targets and the conditions for the parameters to meet;

[0085] Constructing control rules based on the attributes and parameters meeting the conditions; the control rules include: spatial domain control rules, time domain control rules, frequency domain control rules, energy control rules and modulation domain control rules;

[0086] Collect actual electromagnetic environment data to build an actual electromagnetic environment dataset;

[0087] Extracting key feature information from the actual electromagnetic environment data set; the key feature information includes signal strength, path loss, reflection and attenuation characteristics;

[0088] A meta-sample library with similar electromagnetic environments is established using the extracted key feature information; each sample in the meta-sample library includes key feature information and corresponding specific environmental condition labels; and each sample is the similar meta-sample information.

[0089] Specifically, step 200 of this embodiment includes:

[0090] Step 201: Generate control rules. This embodiment reads and interprets the generated electromagnetic environment target parameters, identifies the attributes of each target parameter and the conditions that must be met. Then, based on the attributes of the electromagnetic environment target, preliminary control rules are formulated. These rules cover the following key points:

[0091] Spectrum Control: Determines the frequency range and bandwidth of the desired simulation.

[0092] Spatial distribution: Determine the spatial arrangement of electromagnetic sources and sensors, boundary conditions of the simulation area, etc.

[0093] Time control: Determine dynamic control parameters such as simulation time period and time step.

[0094] Environmental characteristics: Determine the main electromagnetic interference sources, attenuation patterns, reflection / refraction conditions, etc. in the environment.

[0095] Rule optimization: Use machine learning models based on historical data or expert knowledge base to optimize preliminary control rules to ensure their high adaptability and efficiency.

[0096] Step 202: Establishing similar meta-sample information. This embodiment first collects a large amount of actual electromagnetic environment data. This data should include various parameters such as different environmental conditions, frequency ranges, and spatial distribution. Data collection can be carried out through field measurements, existing databases, and publicly available data. Next, this embodiment extracts key features from the actual electromagnetic environment dataset, such as signal strength, path loss, reflection, and attenuation characteristics. These features serve as the information basis for similar meta-samples. Next, this embodiment uses the extracted feature information to establish a library of similar electromagnetic environment meta-samples. Each sample in the meta-sample library should include a feature value and the corresponding specific environmental condition label.

[0097] Furthermore, the control rules of this embodiment include: spectrum control, spatial distribution, time control, environmental characteristics, energy control rules and modulation domain control rules. The details are as follows:

[0098] Spectrum Control:

[0099] ① Determine the frequency range and bandwidth required for simulation, and determine the frequency range required for simulation based on the electromagnetic environment target. The target frequency range is f min , f max ;

[0100] ② Bandwidth calculation: Bandwidth B is calculated using the following formula:

[0101] B=f max -f min

[0102] ③ In order to perform numerical simulation, the frequency range is discretized into multiple frequency points. Assuming that N frequency points are required, the interval Δf between each frequency point is:

[0103]

[0104] The frequency point set is:

[0105] f i =f min +i·Δf(i=0,1,…,N-1)

[0106] Spatial distribution:

[0107] Spatial distribution, determine the spatial arrangement of electromagnetic sources and sensors, and the boundary conditions of the simulation area; first divide the spatial grid: divide the simulation area into three-dimensional grids, assuming that the simulation area is [x min ,x max ]×[y min ,y max ]×[z min ,z max], the grid size is Δx, Δy, Δz. According to the electromagnetic environment target, determine the location of the electromagnetic source. Assume that the location of the electromagnetic source is (x s ,y s ,z s ). According to the monitoring requirements, determine the location of the sensor. Assume that the location of the sensor is (x t ,y t ,z t Finally, determine the boundary conditions of the simulation area. You can choose absorbing boundary conditions (ABC) or perfectly matched layers.

[0108] Time distribution: Determine dynamic control parameters such as simulation time period and time step, as follows:

[0109] ①Simulation time period: Determine the simulation time period T according to the electromagnetic environment target sim .

[0110] ② Time step: Assuming that the simulation requires M time steps, the time step Δt of each time step is:

[0111]

[0112] Environmental characteristics: Determine the main electromagnetic interference sources, attenuation models, reflection / refraction conditions, etc. in the environment, as follows:

[0113] ① Electromagnetic interference source: Determine the location and characteristics of the main electromagnetic interference source in the environment. Assume that the location of the interference source is (x d ,y d ,z d ), the frequency of the interference source is f d .

[0114] ② Attenuation model: Using the free space path loss model, the attenuation formula is:

[0115]

[0116] Among them, d is the propagation distance, f is the frequency, and c is the speed of light.

[0117] ③Reflection / refraction conditions: Determine the reflection and refraction conditions based on the environmental characteristics, and use the Fresnel formula to calculate the reflection and refraction coefficients.

[0118] Energy control rules:

[0119] ① Transmit power control, transmit power P t The electromagnetic coverage requirements of the target area should be met, and the formula is:

[0120] P t =P r +L(d)+G t +Gr

[0121] Among them, P r is the received power, L(d) is the path loss, G t and G r are the transmit and receive antenna gains, respectively.

[0122] Modulation domain control rules:

[0123] ① Modulation mode selection: Select the appropriate modulation mode (such as AM, FM, QAM, etc.) according to the electromagnetic environment target.

[0124] ② Modulation parameter setting: determine the modulation parameters, such as carrier frequency f c , modulation index β, etc.

[0125] ③ Modulation signal generation: Generate the modulation signal according to the selected modulation method and parameters. Taking AM modulation as an example, the modulation signal s(t) is:

[0126] s(t)=A c [1+βm(t)]cos(2πf c t)

[0127] Among them, A c is the carrier amplitude, and m(t) is the baseband signal.

[0128] Preferably, generating a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rule simulation includes:

[0129] Initializing parameters of an electromagnetic environment simulation model according to the control rules; the parameters of the simulation model include electromagnetic source parameters, spatial distribution parameters, time parameters and boundary conditions;

[0130] Running the electromagnetic environment simulation model and generating a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rules;

[0131] During the simulation process, the simulated electromagnetic environment is corrected using the similar meta-sample information, so that the simulated electromagnetic environment has realism and accuracy.

[0132] Furthermore, this embodiment realizes the simulation generation of a simulated electromagnetic environment, and the specific steps are as follows:

[0133] Step 301: Select an appropriate simulation model (such as a ray tracing model, a finite difference time domain method, a statistical model, etc.) according to the electromagnetic environment target.

[0134] Step 302: Initialize the parameters of the simulation model according to the generated control rules, including electromagnetic source parameters, spatial distribution parameters, time parameters, boundary conditions, etc.

[0135] Step 303: Run the electromagnetic environment simulation model to generate a simulated electromagnetic environment that meets the electromagnetic environment target based on the control rules. The information of the similar meta-sample library should be considered during the simulation process to ensure that the simulation results are realistic and accurate.

[0136] Step 304: Monitor the simulation results in real time, compare the simulation data with the target data, and if any deviation is found, record and analyze the cause.

[0137] Step 305: Based on the monitoring results and deviation analysis, similarity learning is performed using a similar element sample library, control rules are dynamically adjusted, simulation parameters are optimized, and simulation is re-executed until the results meet the predetermined electromagnetic environment target.

[0138] Step 306: During the simulation process, the working state of the simulation unit is adjusted in real time according to the monitoring data to perform dynamic regulation.

[0139] Step 307: After the simulation is completed, the results are fed back to the user or related systems, and log information of each process is recorded to facilitate subsequent analysis and improvement.

[0140] Step 308: Update the optimized control rules, monitoring data and simulation results in this simulation into the similar element sample library to further improve and enhance the intelligence level of the entire system.

[0141] Through the above technical solution, this embodiment can achieve the entire process from electromagnetic environment target generation to control rule generation, establishment of similar meta-sample information, and ultimately generation of a simulated electromagnetic environment that meets the target through simulation. This process requires the comprehensive consideration of multiple technologies and methods such as machine learning, data analysis, and simulation modeling.

[0142] Preferably, real-time monitoring and collection of simulated electromagnetic environment monitoring data, performing similarity learning based on the monitoring data, and generating a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment include:

[0143] Get the characteristic vector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i ;

[0144] Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Euclidean distance between them is calculated as follows: Among them, T i =[T i1 ,[T i2 ,…,T in ] and S i =[Si1 ,S i2 ,…,S in ] are the feature vectors of the i-th monitoring point, and n is the dimension of the feature vector;

[0145] Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The cosine similarity between them is calculated as follows:

[0146] Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Manhattan distance between them is:

[0147] The specific similarity deviation is determined according to the comprehensive similarity deviation formula; the comprehensive similarity deviation formula is: d overall (T i ,S i )=α·d euclid (T i ,S i )+β·(1-cos(T i ,S i ))+γ·d manhattan (T i ,S i ); where α, β, and γ are the weight coefficients of the corresponding items, d overall (T i ,S i ) is the specific similarity deviation.

[0148] Optionally, in this embodiment, a global deviation assessment is also performed, and the specific steps are as follows:

[0149] ①Average the similarity deviations of all monitoring points to obtain the global similarity deviation:

[0150]

[0151] Where N is the number of monitoring points.

[0152] ② Dynamic adjustment and optimization, first conduct deviation analysis:

[0153] According to the global similarity deviation D global Analyze the differences between the simulated electromagnetic environment and the target electromagnetic environment. Identify the main sources of deviation (such as specific frequency points, spatial locations, etc.).

[0154] ③ Adjust the control rules based on the deviation analysis results. This can be done by fine-tuning the transmission power, adjusting the electromagnetic source position, optimizing environmental characteristic parameters, etc.

[0155] ④ Iterate the simulation, re-simulate, and recalculate the similarity deviation, iteratively optimize until the global similarity deviation D global Meet expected goals.

[0156] Preferably, the control rule is modified by using the similarity deviation and similarity sample information, and the modified control rule is converted into a correction amount for the similarity control amount of the simulated electromagnetic environment, including:

[0157] Constructing a similar feature vector and a target deviation vector based on the similar meta-sample information;

[0158] Inputting the similarity feature vector and the target deviation vector into a parameter regression model to fit the parameters of the control rule with the similarity deviation to obtain the modified control rule;

[0159] Based on the constructed correction model, the correction amount of each parameter of the control rule is calculated; the calculation formula of the correction amount is: C i =W·ΔT i Among them, C i is the correction amount of the i-th monitoring point, W is the weight of the correction model, ΔT i It is the deviation between the target electromagnetic environment and the simulated electromagnetic environment.

[0160] Furthermore, this embodiment uses a parameter regression model or other appropriate machine learning model to fit the control rule parameters and the similarity deviation. Commonly used models include linear regression, support vector machine (SVM), neural network, etc. This embodiment then uses the trained model to predict the similarity deviation corresponding to the new control rule. Exemplarily, this embodiment can determine the adjustment direction of the control rule parameters, analyze the relationship between the similarity deviation and the control rule parameters, find out which parameters have a significant impact on the similarity deviation, and adjust these parameters so that the simulation results are closer to the target electromagnetic environment. If the deviation is large, it is necessary to increase the value of certain parameters (for example, increase the transmission power, adjust the position of the electromagnetic source). Similarly, if the deviation is small, it is necessary to reduce the value.

[0161] Preferably, according to the correction amount of the similarity control amount, the working state of each simulation unit is readjusted to generate a new electromagnetic environment, thereby completing the dynamic regulation of the simulated electromagnetic environment, including:

[0162] adjusting each simulation unit based on the correction amount to determine a new operating state of each simulation unit according to the correction amount;

[0163] A new simulation experiment is run based on the new working state to generate new simulated electromagnetic environment characteristic data, thereby completing the dynamic regulation of the simulated electromagnetic environment.

[0164] Specifically, this embodiment first prepares data to obtain characteristic data for the target electromagnetic environment and the initial simulated electromagnetic environment. This data includes signal strength, transmit power, location, frequency, path loss, and more. The similarity deviation between the target and simulated electromagnetic environments is then calculated using the aforementioned steps. This deviation should encompass multiple dimensions, including spatial, spectral, and temporal.

[0165] Furthermore, a similar element sample training model is used to predict the correction amount of the control parameter. And based on the current electromagnetic environment characteristics, the correction amount of the control parameter is predicted. Secondly, this embodiment adjusts the working state of the simulation unit according to the correction amount, adjusts each simulation unit (such as the position of the transmitter, the transmission power, etc.), and applies the calculated correction amount. This embodiment is programmed with a dynamic adjustment algorithm to determine the new working state of each simulation unit based on the correction amount. This embodiment reconfigures the simulation unit and reconfigures the working state of the simulation unit according to the adjusted parameters. Among them, the parameters such as the transmission power and frequency of the transmitter should be adjusted according to the correction amount. Finally, this embodiment runs a new simulation experiment to generate new simulated electromagnetic environment characteristic data.

[0166] Furthermore, this embodiment recalculates the similarity deviation, evaluates the effect, and continuously iterates and adjusts until the desired goal is achieved.

[0167] Corresponding to the above method, this embodiment provides an intelligent control method for a spatial electromagnetic environment based on similarity machine learning, including:

[0168] The parsing unit is configured to receive a simulation requirement of a target electromagnetic environment and parse the simulation requirement to obtain an electromagnetic environment target;

[0169] A rule generating unit, configured to generate control rules according to the electromagnetic environment target and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment;

[0170] An environment generation unit, configured to generate a simulated electromagnetic environment that meets the electromagnetic environment target according to control rules;

[0171] a deviation generating unit, configured to monitor and collect simulated electromagnetic environment monitoring data in real time, perform similarity learning based on the monitoring data, and generate a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment;

[0172] a correction amount determination unit, configured to modify the control rule by using the similarity deviation and similarity element sample information, and convert the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment;

[0173] The control unit is used to readjust the working status of various simulation units according to the correction amount of the similarity control amount, generate a new electromagnetic environment, and complete the dynamic control of the simulated electromagnetic environment.

[0174] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0175] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. An intelligent control method for space electromagnetic environment based on similarity machine learning, characterized in that: include: Receiving a simulation requirement of a target electromagnetic environment, and analyzing the simulation requirement to obtain an electromagnetic environment target; Generate control rules according to the electromagnetic environment target, and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment; Generate a simulated electromagnetic environment that meets the electromagnetic environment target based on control rules simulation; Real-time monitoring and collection of simulated electromagnetic environment monitoring data, performing similarity learning based on the monitoring data, and generating specific similarity deviations between the simulated electromagnetic environment and the target electromagnetic environment; Modifying the control rule by using the similarity deviation and similarity element sample information, and converting the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment; According to the correction amount of the similarity control amount, the working state of various simulation units is readjusted to generate a new electromagnetic environment, thus completing the dynamic regulation of the simulated electromagnetic environment; Real-time monitoring and collection of simulated electromagnetic environment monitoring data, similarity learning based on the monitoring data, and generation of specific similarity deviations between the simulated electromagnetic environment and the target electromagnetic environment, including: Get the characteristic vector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i ; Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Euclidean distance between them is calculated as follows: Among them, T i =[T i1 ,T i2 ,…,T in ] and S i =[S i1 ,S i2 ,…,S in ] are the feature vectors of the i-th monitoring point, and n is the dimension of the feature vector; Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The cosine similarity between them is calculated as follows: Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Manhattan distance between them is: The specific similarity deviation is determined according to the comprehensive similarity deviation formula; the comprehensive similarity deviation formula is: d overall (T i ,S i )=α·d euclid (T i ,S i )+β·(1-cos(T i ,S i ))+γ·d manhattan (T i ,S i ); where α, β, and γ are the weight coefficients of the corresponding items, d overal l(T i ,S i ) is the specific similarity deviation; The control rule is modified by using the similarity deviation and similarity element sample information, and the modified control rule is converted into a correction amount for the similarity control amount of the simulated electromagnetic environment, including: Constructing a similar feature vector and a target deviation vector based on the similar meta-sample information; Inputting the similarity feature vector and the target deviation vector into a parameter regression model to fit the parameters of the control rule with the similarity deviation to obtain the modified control rule; Based on the constructed correction model, the correction amount of each parameter of the control rule is calculated; the calculation formula of the correction amount is: C i =W·ΔT i Among them, C i is the correction amount of the i-th monitoring point, W is the weight of the correction model, ΔT i is the deviation between the target electromagnetic environment and the simulated electromagnetic environment; According to the correction value of the similarity control quantity, the working state of each simulation unit is readjusted to generate a new electromagnetic environment, completing the dynamic regulation of the simulated electromagnetic environment, including: adjusting each simulation unit based on the correction amount to determine a new operating state of each simulation unit according to the correction amount; A new simulation experiment is run based on the new working state to generate new simulated electromagnetic environment characteristic data, thereby completing the dynamic regulation of the simulated electromagnetic environment.

2. The intelligent control method of the space electromagnetic environment based on similarity machine learning according to claim 1 is characterized in that: Receiving a simulation requirement of a target electromagnetic environment and parsing the simulation requirement to obtain an electromagnetic environment target, including: Format conversion and verification are performed on the received simulation requirements to obtain simulation data that conforms to a preset format and range; Extracting key information of the simulation data; the key information includes: frequency range, target area, environmental characteristics and simulation time; Normalizing the key information to obtain normalized data; Mapping the normalized data with features in a preset electromagnetic environment model to obtain mapping parameters; the features in the preset electromagnetic environment model include: electromagnetic interference sources, propagation paths, and boundary conditions in the environment; Generating an electromagnetic environment target data structure according to the mapping parameters; the electromagnetic environment target data structure includes key information for generating and controlling a simulated electromagnetic environment; The electromagnetic environment target is determined according to the electromagnetic environment target data structure.

3. The intelligent control method of the space electromagnetic environment based on similarity machine learning according to claim 1 is characterized in that: Generate control rules based on the electromagnetic environment target and establish similar meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment, including: Analyze the electromagnetic environment targets to clarify the attributes of the parameters in each of the electromagnetic environment targets and the conditions for the parameters to meet; Constructing control rules based on the attributes and parameters meeting the conditions; the control rules include: spatial domain control rules, time domain control rules, frequency domain control rules, energy control rules and modulation domain control rules; Collect actual electromagnetic environment data to build an actual electromagnetic environment dataset; Extract key feature information from the actual electromagnetic environment data set; the key feature information includes signal strength, path loss, reflection and attenuation characteristics A meta-sample library with similar electromagnetic environments is established using the extracted key feature information; each sample in the meta-sample library includes key feature information and corresponding specific environmental condition labels; and each sample is the similar meta-sample information.

4. The intelligent control method of the space electromagnetic environment based on similarity machine learning according to claim 1 is characterized in that: Generate a simulated electromagnetic environment that meets the electromagnetic environment objectives based on control rules, including: Initializing parameters of an electromagnetic environment simulation model according to the control rules; the parameters of the simulation model include electromagnetic source parameters, spatial distribution parameters, time parameters and boundary conditions; Running the electromagnetic environment simulation model and generating a simulated electromagnetic environment that meets the electromagnetic environment target according to the control rules; During the simulation process, the simulated electromagnetic environment is corrected using the similar meta-sample information, so that the simulated electromagnetic environment has realism and accuracy.

5. An intelligent control method for space electromagnetic environment based on similarity machine learning, characterized in that: include: The parsing unit is configured to receive a simulation requirement of a target electromagnetic environment and parse the simulation requirement to obtain an electromagnetic environment target; A rule generating unit, configured to generate control rules according to the electromagnetic environment target and establish similarity meta-sample information of the simulated electromagnetic environment and the actual electromagnetic environment; An environment generation unit, configured to generate a simulated electromagnetic environment that meets the electromagnetic environment target according to control rules; a deviation generating unit, configured to monitor and collect simulated electromagnetic environment monitoring data in real time, perform similarity learning based on the monitoring data, and generate a specific similarity deviation between the simulated electromagnetic environment and the target electromagnetic environment; a correction amount determination unit, configured to modify the control rule by using the similarity deviation and similarity element sample information, and convert the modified control rule into a correction amount for the similarity control amount of the simulated electromagnetic environment; The control unit is used to readjust the working status of various simulation units according to the correction amount of the similarity control amount, generate a new electromagnetic environment, and complete the dynamic control of the simulated electromagnetic environment; Real-time monitoring and collection of simulated electromagnetic environment monitoring data, similarity learning based on the monitoring data, and generation of specific similarity deviations between the simulated electromagnetic environment and the target electromagnetic environment, including: Get the characteristic vector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i ; Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Euclidean distance between them is calculated as follows: Among them, T i =[T i1 ,T i2 ,…,T in ] and S i =[S i1 ,S i2 ,…,S in ] are the feature vectors of the i-th monitoring point, and n is the dimension of the feature vector; Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The cosine similarity between them is calculated as follows: Calculate the eigenvector S of the simulated electromagnetic environment monitoring data i and the characteristic vector T of the target electromagnetic environment i The Manhattan distance between them is: The specific similarity deviation is determined according to the comprehensive similarity deviation formula; the comprehensive similarity deviation formula is: d overall (T i ,S i )=α·d euclid (T i ,S i )+β·(1-cos(T i ,S i ))+γ·d manhattan (T i ,S i ); where α, β, and γ are the weight coefficients of the corresponding items, d overall (T i ,S i ) is the specific similarity deviation; The control rule is modified by using the similarity deviation and similarity element sample information, and the modified control rule is converted into a correction amount for the similarity control amount of the simulated electromagnetic environment, including: Constructing a similar feature vector and a target deviation vector based on the similar meta-sample information; Inputting the similarity feature vector and the target deviation vector into a parameter regression model to fit the parameters of the control rule with the similarity deviation to obtain the modified control rule; Based on the constructed correction model, the correction amount of each parameter of the control rule is calculated; the calculation formula of the correction amount is: C i =W·ΔT i Among them, C i is the correction amount of the i-th monitoring point, W is the weight of the correction model, ΔT i is the deviation between the target electromagnetic environment and the simulated electromagnetic environment; According to the correction value of the similarity control quantity, the working state of each simulation unit is readjusted to generate a new electromagnetic environment, completing the dynamic regulation of the simulated electromagnetic environment, including: adjusting each simulation unit based on the correction amount to determine a new operating state of each simulation unit according to the correction amount; A new simulation experiment is run based on the new working state to generate new simulated electromagnetic environment characteristic data, thereby completing the dynamic regulation of the simulated electromagnetic environment.

Images