Electromagnetic wave transmission loss prediction method and system based on data mechanism hybrid drive

Abstract

The invention provides an electromagnetic wave transmission loss prediction method and system based on data mechanism hybrid drive, and the method comprises the steps: completing the basic term weighting calculation of three attenuation models of free space loss, irregular ground loss and weather loss based on a semi-deterministic model, and obtaining the basic term of the electromagnetic wave transmission loss; inputting the position information of the receiving end and the transmitting end into a neural network, and using the neural network to complete difference term prediction of electromagnetic wave transmission loss; and according to the basic item of the electromagnetic wave transmission loss and the difference item of the electromagnetic wave transmission loss, calculating to obtain a path loss difference, and completing electromagnetic wave transmission loss prediction. According to the technical scheme, the technical problems that in the prior art, a statistical model lacks physical meanings, a deterministic model is too complex, and electromagnetic wave transmission loss modeling complexity and simulation time cost are increased exponentially are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic wave transmission environment in urban block scenes, and in particular to a method and system for predicting electromagnetic wave transmission loss based on data mechanism hybrid drive. Background Art

[0002] Electromagnetic channel transmission environment modeling is a key part of communication system design and analysis. The accuracy and generalization of the electromagnetic channel transmission environment model directly affect the communication quality and communication capability evaluation results. Since the 1950s, foreign research teams and scholars have conducted research on the modeling of large and small scale characteristics of electromagnetic channels in various typical scenarios, and established some classic electromagnetic channel models. Classic electromagnetic channel modeling can usually be divided into three modeling methods: statistical channel modeling, deterministic channel modeling, and semi-deterministic channel modeling.

[0003] Statistical channel modeling is an empirical modeling method. It obtains a large amount of measured data by performing channel measurements in some typical scenarios, and analyzes these data to obtain the statistical characteristics of the channel in the corresponding scenario, so as to establish an empirical channel model that can reflect the attenuation of electromagnetic wave signals. Common empirical channel models include log-normal shadow fading model, Egli, Longley-Rice, Okumura-Hata, etc. However, the establishment of empirical channel models is based on some assumptions (such as samples are independent of each other and samples follow a normal distribution, etc.). These assumptions ignore the specific propagation details. Therefore, the empirical channel model obtained by statistical modeling is not accurate in predicting the channel characteristics of specific scenarios.

[0004] Deterministic channel modeling uses the characteristics of the propagation environment to solve the approximate solution of electromagnetic waves to Maxwell's equations. This modeling method requires detailed environmental information. Common deterministic channel modeling methods include ray tracing and finite-difference time domain (FDTD). Among them, the ray tracing method is based on geometric optics and consistent scattering theory. It divides the electromagnetic communication environment into many tiny grids, regards the propagation of electromagnetic waves as a variety of ray path transmissions, and uses the reflection, refraction and diffraction of ray propagation to calculate the amplitude, frequency, phase, delay and other parameters of each ray path. The ray tracing method provides high-resolution channel modeling and is suitable for complex indoor and urban environments. The finite-difference time domain (FDTD) method is based on Maxwell's equations. It divides space and time into discrete grids and approximates the derivative terms in differential form, thereby converting partial differential equations into differential equations. However, limited by the storage of cell information, FDTD cannot calculate large amounts of data.

[0005] In order to solve the problem that the prediction error of statistical channel modeling is large and the computational complexity of deterministic channel modeling is large, researchers have proposed a modeling method between statistical modeling and deterministic modeling - semi-deterministic modeling. It considers that the channel has both certain deterministic characteristics and certain randomness or uncertainty components. This method can more comprehensively describe the behavior of the channel in some complex communication environments. The semi-deterministic channel modeling methods are mainly divided into random geometric modeling method and correlation matrix method. The random geometric modeling method is a simplification of the ray tracing method. It does not require a detailed signal propagation environment. It randomly distributes scatterer groups near the base station and mobile station according to certain statistical characteristics. Each scatterer group corresponds to a channel path, and the channel impulse response is obtained by adding all the channel paths. Some common channel models obtained by random geometric modeling methods include Cost259 model, IMT-Advanced model, SCM / SCME model, WINNER channel model, etc. The correlation matrix method reflects the correlation between spatial channels. It uses the actual measured data or channel statistical information to obtain parameters such as path delay and incident angle, and then derives the channel spatial correlation matrix from these parameters.

[0006] With the rapid development of artificial intelligence technology, machine learning and deep learning have become the most active research fields. Domestic and foreign research teams and scholars have begun to pay attention to electromagnetic channel transmission environment modeling based on intelligent algorithms. Some neural networks (such as multi-layer perceptron neural network, generalized radial basis neural network, wavelet neural network, dynamic neural network and fuzzy neural network, etc.) have been applied to electromagnetic channel transmission environment modeling. The modeling methods have been expanded to three modeling methods: mechanism-driven modeling, data-driven modeling, and data-mechanism hybrid-driven modeling. Among them, data-mechanism hybrid modeling combines data-driven modeling with mechanism-driven modeling, integrating the advantages of data-driven and mechanism-driven modeling, aiming to make full use of the in-depth understanding of the mechanism and the rich information of the data to build a more accurate, comprehensive and flexible digital twin model.

[0007] At present, there are the following problems in modeling the electromagnetic channel transmission environment:

[0008] (1) Approximate models of electromagnetic wave transmission loss in ground-to-ground links, such as the free space propagation model and the logarithmic distance path loss model, are approximate calculations of electromagnetic wave transmission loss and suffer from inaccurate mechanism modeling.

[0009] (2) The large-scale parameters of the electromagnetic channel of the ground-to-ground link are affected by many factors, such as the distance between the transmitter and the receiver, the carrier frequency, the topography, and the climate and weather (such as rain, snow, and fog). There is an error term between the calculation method of superimposing multiple attenuations and the actual electromagnetic wave transmission loss. Summary of the invention

[0010] The present invention provides a method and system for predicting electromagnetic wave transmission loss based on a hybrid drive of data mechanisms, which can solve the technical problems in the prior art that the statistical model lacks physical meaning, the deterministic model is too complex, and the electromagnetic wave transmission loss modeling complexity and simulation time cost show an exponential growth.

[0011] According to one aspect of the present invention, a method for predicting electromagnetic wave transmission loss based on a hybrid drive of data mechanisms is provided. The method for predicting electromagnetic wave transmission loss based on a hybrid drive of data mechanisms comprises: based on a semi-deterministic model, completing a weighted calculation of the basic terms of the three attenuation models of free space loss, irregular ground loss and weather loss to obtain the basic terms of electromagnetic wave transmission loss; inputting the position information of the receiving end and the transmitting end into a neural network, and using the neural network to complete the prediction of the difference terms of the electromagnetic wave transmission loss; calculating the path loss difference based on the basic terms of the electromagnetic wave transmission loss and the difference terms of the electromagnetic wave transmission loss to complete the prediction of the electromagnetic wave transmission loss.

[0012] Furthermore, before using the neural network to complete the prediction of the difference term of the electromagnetic wave transmission loss, the electromagnetic wave transmission loss prediction method based on the data mechanism hybrid drive also includes: training the neural network, the input of the neural network is the receiving and transmitting position information, and the output is the difference between the basic item solution value based on the semi-deterministic model and the predicted value, and the mean square error between the basic item solution value based on the semi-deterministic model and the true value is used as the loss function. When the error between the neural network output and its expectation is minimized, the training process ends.

[0013] Furthermore, the basic term of electromagnetic wave transmission loss can be calculated according to S 基损 =αL bf +βA h +γS 天损 Calculate and obtain, where L bf is the free space loss, A h is the irregular ground loss, S 天损 is the weather loss, α is the free space loss weight coefficient, β is the irregular ground loss weight coefficient, and γ is the weather loss weight coefficient.

[0014] Furthermore, the free space loss can be calculated according to L bf =92.44+20lgd(km)+20lgf(GHz) is calculated, where d is the distance between the transmitting and receiving ends, and f is the frequency.

[0015] Furthermore, considering the urban lake and rural terrain types h<10m, the irregular ground loss A h According to A h =-[21.8+6.2log(f)]log(h / R) calculated; considering the ground type in the suburbs of the city, h<10m, irregular ground loss Ah According to A h = J(v)-6.03 calculated; considering the urban area ground object type, h <15m, irregular ground loss A h According to A h = J(v)-6.03 calculated; considering the type of landforms in densely populated urban areas, h<20m, irregular ground loss A h According to A h = J(v)-6.03 calculated, where h dif =Rh, v is the height gain terminal correction loss, h is the antenna height, R is the representative ground object height, J(v) is the single knife-edge diffraction loss approximation ... dif is the height difference between the ground object and the antenna, θ clut is the angle between the ground object, the antenna position and the ground plane, w s is the street width.

[0016] Furthermore, the weather loss S 天损 According to S 天损 =γ R +L 雾 +L 雪 Calculate and obtain, where γ R is the rain layer loss, L 雾 is the fog layer loss, L 雪 For snow layer loss.

[0017] Furthermore, the rain layer loss γ R According to γ R =K·R 雨 a Calculate and obtain, where R 雨 is the rainfall rate in mm / h, K = [K H +K V +(K H -K V )cos 2 θ·cos2τ] / 2, a=[K H a H +K V a V +(K H a H -K V a V )cos 2 θ·cos2τ] / (2K), θ is the path elevation angle, τ is the polarization inclination angle relative to the horizontal direction, K is the rain layer loss coefficient, a is the rain layer loss power, K H is the horizontal polarization value of the rain layer loss coefficient, KV is the vertical polarization value of the rain layer loss coefficient, a H is the horizontal polarization value of the rain layer loss power, a V is the vertical polarization value of the rain layer loss power.

[0018] Furthermore, the fog layer loss L 雾 According to L 雾 =K l (f,T)η(f) is calculated and obtained, K l is the specific attenuation coefficient, η is the density of liquid water in the fog,

[0019]

[0020]

[0021] f p =20.09-142(θ-1)+294(θ-1) 2 , f s =590-1500(θ-1), where f is the frequency, T is the temperature, θ is the temperature coefficient, and f p is the main relaxation frequency, f s is the secondary relaxation frequency, ε'(f) is the first derivative of the dielectric constant, and ε"(f) is the second derivative of the dielectric constant.

[0022] Furthermore, the snow layer loss L 雪 According to Calculate and obtain, where r is the equivalent humidity, c is the speed of light, R 距 is the distance between the transmitter and the receiver, and f is the frequency.

[0023] According to another aspect of the present invention, there is provided an electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive, the electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive uses the electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive as described above to predict electromagnetic wave transmission loss, the electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive comprises: a basic item calculation module for electromagnetic wave transmission loss, the basic item calculation module for electromagnetic wave transmission loss is used to complete the basic item weighted calculation of the three attenuation models of free space loss, irregular ground loss and weather loss based on a semi-deterministic model, and obtain the basic item of electromagnetic wave transmission loss; a difference item prediction module, the difference item prediction module is used to input the position information of the receiving end and the transmitting end into a neural network, and use the neural network to complete the difference item prediction of the electromagnetic wave transmission loss; an electromagnetic wave transmission loss prediction module, the electromagnetic wave transmission loss prediction module is used to obtain the path loss difference according to the basic item of the electromagnetic wave transmission loss and the difference item of the electromagnetic wave transmission loss, and complete the electromagnetic wave transmission loss prediction.

[0024] By applying the technical solution of the present invention, a method for predicting electromagnetic wave transmission loss based on data-mechanism hybrid drive is provided. The method realizes the prediction of electromagnetic wave transmission loss of ground-to-ground link by integrating pure mechanism modeling and intelligent modeling technology. The calculation results of the electromagnetic wave transmission loss attenuation model are used as basic data, and the neural network intelligent model is used as a correction term. By utilizing the learning ability of the neural network, a prediction method combining the mechanism attenuation model and the intelligent model is explored. The bottlenecks of the traditional modeling method, such as strong dependence on principles, high modeling difficulty, low calculation efficiency, and insufficient calculation result accuracy, are broken through, and the real-time and accuracy of the electromagnetic wave transmission loss prediction model are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The included drawings are used to provide a further understanding of the embodiments of the present invention, which constitute a part of the specification, are used to illustrate the embodiments of the present invention, and together with the text description, explain the principles of the present invention. Obviously, the drawings in the following description are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 A schematic diagram of an electromagnetic wave transmission loss calculation architecture based on data mechanism hybrid drive provided according to a specific embodiment of the present invention is shown;

[0027] Figure 2 A schematic diagram of difference term prediction based on a neural network provided according to a specific embodiment of the present invention is shown;

[0028] Figure 3 A neural network architecture diagram provided according to a specific embodiment of the present invention is shown. DETAILED DESCRIPTION

[0029] It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is by no means intended to limit the present invention and its application or use. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0031] Unless otherwise specifically stated, the relative arrangement of the parts and steps described in these embodiments, numerical expressions and numerical values ​​do not limit the scope of the present invention. At the same time, it should be understood that, for ease of description, the sizes of the various parts shown in the accompanying drawings are not drawn according to the actual proportional relationship. The technology, method and equipment known to ordinary technicians in the relevant field may not be discussed in detail, but in appropriate cases, the technology, method and equipment should be regarded as a part of the authorization specification. In all examples shown and discussed here, any specific value should be interpreted as being merely exemplary, rather than as a limitation. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once a certain item is defined in an accompanying drawing, it does not need to be further discussed in subsequent drawings.

[0032] like Figures 1 to 3 As shown, according to a specific embodiment of the present invention, a method for predicting electromagnetic wave transmission loss based on a hybrid drive of data mechanisms is provided. The method for predicting electromagnetic wave transmission loss based on a hybrid drive of data mechanisms includes: based on a semi-deterministic model, completing the weighted calculation of the basic terms of the three attenuation models of free space loss, irregular ground loss and weather loss, and obtaining the basic terms of electromagnetic wave transmission loss; inputting the position information of the receiving end and the transmitting end into a neural network, and using the neural network to complete the prediction of the difference terms of the electromagnetic wave transmission loss; calculating the path loss difference based on the basic terms of the electromagnetic wave transmission loss and the difference terms of the electromagnetic wave transmission loss, and completing the prediction of the electromagnetic wave transmission loss.

[0033] By applying this configuration, a method for predicting electromagnetic wave transmission loss based on data-mechanism hybrid drive is provided. This method realizes the prediction of electromagnetic wave transmission loss of ground-to-ground links by integrating pure mechanism modeling and intelligent modeling technology. The calculation results of the electromagnetic wave transmission loss attenuation model are used as basic data, and the neural network intelligent model is used as a correction term. Utilizing the learning ability of neural networks, a prediction method combining the mechanism attenuation model and the intelligent model is explored. This breaks through the bottlenecks of traditional modeling methods such as strong dependence on principles, high modeling difficulty, low calculation efficiency, and insufficient calculation result accuracy, and improves the real-time and accuracy of the electromagnetic wave transmission loss prediction model. Therefore, compared with the prior art, the electromagnetic wave transmission loss prediction method based on data-mechanism hybrid drive provided by the present invention has the following advantages:

[0034] 1) Fusion of mechanism model and data-driven model. This patent innovatively uses semi-deterministic models such as free space loss, irregular terrain attenuation, rain, snow and fog weather attenuation as the basis of the mechanism model. These models are based on physical principles and experimental data and can accurately capture the basic transmission characteristics of electromagnetic waves in different environments. At the same time, a neural network is introduced as an error model to predict the loss difference terms that are difficult to accurately describe by the mechanism model in a data-driven way. This fusion method not only ensures the physical significance of the prediction results, but also improves the efficiency and accuracy of the prediction.

[0035] 2) Intelligent correction mechanism for the difference term of electromagnetic wave transmission loss. As a component of the electromagnetic wave loss model, the neural network learns historical data and provides real-time feedback. It utilizes the powerful ability of neural networks in data processing and prediction, and continuously corrects the difference term, making the prediction results of the entire model closer to reality, thereby achieving high-precision prediction.

[0036] Specifically, in the present invention, the electromagnetic wave transmission loss model based on data mechanism hybrid drive uses the semi-deterministic model as the basic term of electromagnetic wave transmission loss, and realizes the difference term prediction of electromagnetic wave transmission loss through neural network. The final output result of the model is the difference between the basic term and the difference term (see Figure 1 ).

[0037] Step 1: Solve the basic terms based on the semi-deterministic model

[0038] The electromagnetic wave transmission loss model is mainly calculated by weighted calculation of three attenuation models: free space loss, irregular ground loss, and weather loss. The basic term of electromagnetic wave transmission loss can be calculated based on S 基损 =αL bf +βA h +γS 天损 Calculate and obtain, where L bf is the free space loss, A h is the irregular ground loss, S 天损is the weather loss, α is the free space loss weight coefficient, β is the irregular ground loss weight coefficient, and γ is the weather loss weight coefficient.

[0039] 1) Free space loss

[0040] The free space loss model is based on the basic physical principle of electromagnetic wave propagation in free space, and uses mathematical formulas to quantify the energy attenuation caused by diffusion during the propagation of electromagnetic waves. When electromagnetic waves propagate in free space, their energy will gradually diffuse as the propagation distance increases, resulting in a decrease in signal strength at the receiving point. The free space loss model calculates the degree of this energy attenuation by considering factors such as the propagation distance and frequency of the electromagnetic waves. Recommendation ITU-RP.525-3 aims to provide a method for calculating free space loss to ensure link quality in radio communication systems. The basic loss of free space can be calculated using the following formula:

[0041] L bf =92.44+20lgd(km)+20lgf(GHz)

[0042] Where d is the distance between the transmitter and receiver, and f is the frequency.

[0043] 2) Irregular ground loss

[0044] The irregular ground loss model is statistical in nature and is used as a terminal correction to the propagation model for long-distance paths. The clutter loss is defined as the difference between the propagation loss and the basic transmission loss with or without terminal clutter at either end of the path, with all other path details being the same.

[0045] The height gain terminal correction model takes into account the blocking loss and reflection caused by objects at the representative altitude, as well as the scattering and reflection caused by the ground and smaller objects. An additional loss A is calculated by the height gain terminal correction model. h , which is a correction for the end loss added to the basic transmission loss calculated based on the representative clutter height used.

[0046]

[0047] Note: J(-0.78)≈0, and this defines a lower limit on the approximation that should be used. For ν≤-0.78, J(ν) is set to 0.

[0048] Table 1 Input parameters of the height gain terminal correction model

[0049] enter symbol unit frequency f GHz Antenna height h m Street Width <![CDATA[w s ]]> m Representative feature height R m Feature Type - -

[0050] When the transmitter or receiver antenna is located below the ground coverage height R representing the transmitter or receiver environment, the additional loss Ah The calculation is as follows. Where available, representative clutter heights based on accurate clutter height information should be used. The loss modeling approach is the same for both transmitter and receiver.

[0051] Considering the urban lake and rural landform types h<10m, A h =-[21.8+6.2log(f)]log(h / R)dB;

[0052] Considering the landform type in the suburbs of the city, h<10m, A h = J(v)-6.03dB;

[0053] Considering the urban terrain type, h<15m, A h = J(v)-6.03dB;

[0054] Considering the landform type in densely populated urban areas, h<20m, A h = J(v)-6.03dB;

[0055] The v term and K are calculated by the following formula h2 item:

[0056]

[0057] h dif =R-hm

[0058]

[0059] K h2 =21.8+6.2log(f)

[0060]

[0061] Where v is a dimensionless parameter, h is the antenna height, R is the representative object height, v is the height gain terminal correction loss, h is the antenna height, R is the representative object height, J(v) is the single knife-edge diffraction loss approximation, and h dif is the height difference between the ground object and the antenna, θ clut is the angle between the ground object, the antenna position and the ground plane, w s is the street width (unit: meter), w s Unless specific local information is available, this value should be set to 27, where f is the frequency in GHz.

[0062] 3) Weather loss

[0063] For the ground-to-ground link model, water droplets will scatter and absorb electromagnetic waves when propagating in rain, snow, and fog. Weather loss S 天损 According to S 天损=γ R +L 雾 +L 雪 Calculate and obtain, where γ R is the rain layer loss, L 雾 is the fog layer loss, L 雪 For snow layer loss.

[0064] ①Rain decline

[0065] The ITU-RP.838-3 standard gives the relationship between rain attenuation rate and rainfall rate as an exponential model as follows

[0066] γ R =K·R 雨 a

[0067] Where R is the rainfall rate in mm / h, and K and a are determined by the following formula:

[0068] K=[K H +K V +(K H -K V )cos 2 θ·cos2τ] / 2

[0069] a=[K H a H +K V a V +(K H a H -K V a V )cos 2 θ·cos2τ] / (2K)

[0070] Where θ is the path elevation angle; τ is the polarization inclination angle relative to the horizontal direction (τ = 0° for horizontal polarization, τ = 90° for vertical polarization, and τ = 45° for circular polarization). K is the rain layer loss coefficient, a is the rain layer loss power, and K H is the horizontal polarization value of the rain layer loss coefficient, K V is the vertical polarization value of the rain layer loss coefficient, a H is the horizontal polarization value of the rain layer loss power, a V is the vertical polarization value of the rain layer loss power. The horizontal polarization value of the rain layer loss coefficient, the vertical polarization value of the rain layer loss coefficient, the horizontal polarization value of the rain layer loss power, and the vertical polarization value of the rain layer loss power can be obtained by looking up the table in Recommendation ITU-RP.838-3.

[0071] ②Fog decay

[0072] ITU-RP.840-6 standard provides an empirical model of fog propagation loss

[0073] L 雾 =K l (f,T)η(f)

[0074] Among them, K l is the specific attenuation coefficient ((dB / km) / (g / m 3 )), η is the density of liquid water in the fog.

[0075]

[0076]

[0077] And there is

[0078]

[0079]

[0080]

[0081] Where f is the frequency, T is the temperature (K), θ is the temperature coefficient, ε'(f) is the first derivative of the dielectric constant, and ε"(f) is the second derivative of the dielectric constant. The primary and secondary relaxation frequencies are:

[0082] f p =20.09-142(θ-1)+294(θ-1) 2

[0083] f s =590-1500(θ-1)

[0084] Among them, f p is the main relaxation frequency, f s is the secondary relaxation frequency.

[0085] ③Snow decline

[0086] For navigation radar signals, snowfall also causes propagation loss. The attenuation caused by propagation in snow is independent of the polarization mode and varies only with the signal frequency. The Gunn-East attenuation model stipulates that the propagation loss is

[0087]

[0088] The coefficient r represents the equivalent humidity. The greater the snowfall, the greater r is. c is the speed of light, R 距 is the distance between the transmitter and the receiver, and f is the frequency.

[0089] Step 2: Prediction of difference terms based on neural network

[0090] The input of the neural network is the sending and receiving position information, and the output is the difference between the solution value and the predicted value of the basic term based on the semi-deterministic model (see Figure 2 ). When training the neural network, the mean square error between the solution of the basic term based on the semi-deterministic model and the true value is used as the loss function. The training process ends when the error between the neural network output and its expectation is minimized. After the neural network is trained, it has sufficient ability to estimate the value of the error term. During the test, the predicted value of the electromagnetic wave transmission loss is the solution of the basic term based on the semi-deterministic model minus the difference term.

[0091] Using a multi-layer perceptron neural network (see Figure 3 ) To make difference item predictions, it is necessary to do sufficient training on the training set to prevent overfitting. At the same time, the difference item prediction model based on the neural network must have accuracy and generalization. Therefore, the data is divided into three groups. The first group is used to train the network for weight adjustment. The second group is used as a validation set to check the prediction accuracy and learning efficiency of the neural network, and is also used as a stopping criterion. The third group is defined as a test set to estimate the performance of the neural network on new data and check the generalization of the neural network prediction.

[0092] According to another aspect of the present invention, there is provided an electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive, the electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive uses the electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive as described above to predict electromagnetic wave transmission loss, the electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive includes a basic item calculation module of electromagnetic wave transmission loss, a difference item prediction module and an electromagnetic wave transmission loss prediction module, the basic item calculation module of electromagnetic wave transmission loss is used to complete the basic item weighted calculation of the three attenuation models of free space loss, irregular ground loss and weather loss based on a semi-deterministic model, and obtain the basic item of electromagnetic wave transmission loss, the difference item prediction module is used to input the position information of the receiving end and the transmitting end into a neural network, and use the neural network to complete the difference item prediction of the electromagnetic wave transmission loss, the electromagnetic wave transmission loss prediction module is used to obtain the path loss difference according to the basic item of electromagnetic wave transmission loss and the difference item of electromagnetic wave transmission loss, and complete the electromagnetic wave transmission loss prediction.

[0093] By applying this configuration, a data-mechanism hybrid-driven electromagnetic wave transmission loss prediction system is provided. The system realizes the prediction of electromagnetic wave transmission loss of ground-to-ground links by integrating pure mechanism modeling and intelligent modeling technology. The calculation results of the electromagnetic wave transmission loss attenuation model are used as basic data, and the neural network intelligent model is used as a correction item. By utilizing the learning ability of the neural network, the prediction method combining the mechanism attenuation model and the intelligent model is explored. The bottlenecks of traditional modeling methods, such as strong dependence on principles, high modeling difficulty, low calculation efficiency, and insufficient calculation result accuracy, are broken through, and the real-time and accuracy of the electromagnetic wave transmission loss prediction model are improved.

[0094] The present invention builds a complex electromagnetic environment channel transmission characteristic model based on the method of continuous calibration of mechanism parameters and intelligent correction of error models, and continuously corrects the difference terms through neural networks to realize the prediction of electromagnetic wave transmission loss based on the hybrid drive of mechanism and data. The research results lay the foundation for the construction of electromagnetic digital parallel space, and are of great significance for promoting the development and application of electromagnetic space integration technology, improving the intelligent level of detection and communication technology, and promoting multidisciplinary cross-integration.

[0095] The method proposed in this patent can be used to train a data mechanism model of electromagnetic wave transmission loss in ground-to-ground links. When testing the trained model, the following operations are included:

[0096] 1) The electromagnetic wave frequency, the location information (longitude, latitude, and altitude) of the transmitter and receiver, the irregular terrain loss parameters, the weather loss parameters, etc. are used as the input of the uncertainty model, and the value of the basic item of electromagnetic wave transmission loss is obtained through the corresponding loss calculation;

[0097] 2) The location information of the transmitter and the receiver is used as the input of the trained neural network intelligent model, and the difference term of the electromagnetic wave transmission loss is calculated by the neural network;

[0098] 3) The final predicted output value of the data mechanism model of the electromagnetic wave transmission loss of the ground-to-ground link is the difference between the basic term and the difference term.

[0099] In summary, the present invention provides a method for predicting electromagnetic wave transmission loss based on data-mechanism hybrid drive. This method realizes the prediction of electromagnetic wave transmission loss of ground-to-ground links by integrating pure mechanism modeling and intelligent modeling technology. The calculation results of the electromagnetic wave transmission loss attenuation model are used as basic data, and the neural network intelligent model is used as a correction term. By utilizing the learning ability of the neural network, a prediction method combining the mechanism attenuation model and the intelligent model is explored. This method breaks through the bottlenecks of traditional modeling methods, such as strong dependence on principles, high modeling difficulty, low calculation efficiency, and insufficient calculation result accuracy, and improves the real-time and accuracy of the electromagnetic wave transmission loss prediction model.

[0100] For ease of description, spatially relative terms such as "above", "above", "on the upper surface of", "above", etc. may be used here to describe the spatial positional relationship between a device or feature and other devices or features as shown in the figure. It should be understood that spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation described in the figure. For example, if the device in the accompanying drawings is inverted, the device described as "above other devices or structures" or "above other devices or structures" will be positioned as "below other devices or structures" or "below other devices or structures". Thus, the exemplary term "above" can include both "above" and "below". The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatially relative descriptions used here are interpreted accordingly.

[0101] In addition, it should be noted that the use of terms such as "first" and "second" to limit components is only for the convenience of distinguishing the corresponding components. If not otherwise stated, the above terms have no special meaning and therefore cannot be understood as limiting the scope of protection of the present invention.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting electromagnetic wave transmission loss based on data mechanism hybrid drive, characterized in that: The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive includes: Based on the semi-deterministic model, the weighted calculation of the basic items of the three attenuation models of free space loss, irregular ground loss and weather loss is completed to obtain the basic items of electromagnetic wave transmission loss; The location information of the receiving end and the transmitting end is input into the neural network, and the difference term prediction of the electromagnetic wave transmission loss is completed by using the neural network; The path loss difference is calculated based on the basic term of the electromagnetic wave transmission loss and the difference term of the electromagnetic wave transmission loss to complete the electromagnetic wave transmission loss prediction.

2. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 1 is characterized in that: Before using a neural network to complete the prediction of the difference term of electromagnetic wave transmission loss, the electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive also includes: training the neural network, the input of the neural network is the receiving and transmitting position information, and the output is the difference between the basic term solution value based on the semi-deterministic model and the predicted value, and the mean square error between the basic term solution value based on the semi-deterministic model and the true value is used as the loss function. When the error between the neural network output and its expectation is minimized, the training process ends.

3. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 2 is characterized in that: The basic term of electromagnetic wave transmission loss can be calculated based on S 基损 =αL bf +βA h +γS 天损 Calculate and obtain, where L bf is the free space loss, A h is the irregular ground loss, S 天损 is the weather loss, α is the free space loss weight coefficient, β is the irregular ground loss weight coefficient, and γ is the weather loss weight coefficient.

4. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 3 is characterized in that: The free space loss can be calculated according to L bf =92.44+20lgd(km)+20lgf(GHz) is calculated, where d is the distance between the transmitting and receiving ends, and f is the frequency.

5. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 4 is characterized in that: Considering the urban lake and rural terrain types h<10m, the irregular ground loss A h According to A h =-[21.8+6.2log(f)]log(h / R) calculated; considering the ground type in the suburbs of the city, h<10m, irregular ground loss A h According to A h = J(v)-6.03 calculated; considering the urban area ground object type, h <15m, irregular ground loss A h According to A h = J(v)-6.03 calculated; considering the type of landforms in densely populated urban areas, h<20m, irregular ground loss A h According to A h = J(v)-6.03 calculated, where h dif =Rh, v is a dimensionless parameter, h is the antenna height, R is the height of the representative object, J(v) is the approximate value of the single knife-edge diffraction loss, and h dif is the height difference between the ground object and the antenna, θ clut is the angle between the ground object, the antenna position and the ground plane, w s is the street width.

6. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 5 is characterized in that: The weather loss S 天损 According to S 天损 =γ R +L 雾 +L 雪 Calculate and obtain, where γ R is the rain layer loss, L 雾 is the fog layer loss, L 雪 For snow layer loss.

7. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 6 is characterized in that: The rain layer loss γ R According to γ R =K·R 雨 a Calculate and obtain, where R 雨 is the rainfall rate in mm / h, K = [K H +K V +(K H -K V )cos 2 θ·cos2τ] / 2, a=[K H a H +K V a V +(K H a H -K V a V )cos 2 θ·cos2τ] / (2K), θ is the path elevation angle, τ is the polarization inclination angle relative to the horizontal direction, K is the rain layer loss coefficient, a is the rain layer loss power, K H is the horizontal polarization value of the rain layer loss coefficient, K V is the vertical polarization value of the rain layer loss coefficient, a H is the horizontal polarization value of the rain layer loss power, a V is the vertical polarization value of the rain layer loss power.

8. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 7 is characterized in that: The fog layer loss L 雾 According to L 雾 =K l (f,T)η(f) is calculated and obtained, K l is the specific attenuation coefficient, η is the density of liquid water in the fog, f p =20.09-142(θ-1)+294(θ-1) 2 , f s =590-1500(θ-1), where f is the frequency, T is the temperature, θ is the temperature coefficient, and f p is the main relaxation frequency, f s is the secondary relaxation frequency, ε'(f) is the first derivative of the dielectric constant, and ε"(f) is the second derivative of the dielectric constant.

9. The electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive according to claim 8 is characterized in that: The snow layer loss L 雪 According to Calculate and obtain, where r is the equivalent humidity, c is the speed of light, R 距 is the distance between the transmitter and the receiver, and f is the frequency.

10. An electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive, characterized in that: The electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive uses the electromagnetic wave transmission loss prediction method based on data mechanism hybrid drive as described in claims 1 to 9 to predict the electromagnetic wave transmission loss, and the electromagnetic wave transmission loss prediction system based on data mechanism hybrid drive includes: A basic item calculation module for electromagnetic wave transmission loss, which is used to complete the weighted calculation of the basic items of the three attenuation models of free space loss, irregular ground loss and weather loss based on a semi-deterministic model to obtain the basic items of electromagnetic wave transmission loss; A difference item prediction module, the difference item prediction module is used to input the position information of the receiving end and the transmitting end into a neural network, and use the neural network to complete the difference item prediction of the electromagnetic wave transmission loss; The electromagnetic wave transmission loss prediction module is used to calculate the path loss difference based on the basic term of the electromagnetic wave transmission loss and the difference term of the electromagnetic wave transmission loss, so as to complete the electromagnetic wave transmission loss prediction.

Images