Intelligent inversion method of sonic boom signal based on conditional generative adversarial network

Abstract

The present invention relates to the field of aircraft design and acoustic technology, and specifically to a sonic boom signal intelligent inversion method based on a conditional generative adversarial network, comprising the following steps: step S1: selecting a sonic boom reference near-field signal, and calculating a sonic boom reference far-field signal; step S2: designing a sonic boom target far-field signal based on the obtained sonic boom reference far-field signal; step S3: adding disturbances on the basis of the sonic boom reference near-field signal selected in step S1, obtaining a sonic boom near-field signal after superimposing disturbances, using the sonic boom near-field signal after superimposing disturbances, calculating a corresponding sonic boom far-field signal, and establishing a sonic boom near-field and far-field signal sample set; step S4: using the obtained sonic boom near-field and far-field signal sample set, training a conditional generative adversarial network cGAN; step S5: inputting the sonic boom target far-field signal designed in step S2 into the trained conditional generative adversarial network cGAN, and inverting and outputting a sonic boom target near-field signal. The present invention can get rid of the constraints of the sonic boom inversion distance and improve the application scope of the sonic boom inversion.

Description

Technical Field

[0001] The present invention relates to the field of aircraft design and acoustic technology, and in particular to an intelligent inversion method for sonic boom signals based on a conditional generative adversarial network. Background Art

[0002] Supersonic passenger aircraft are one of the important ways to reduce travel time in the future. In the second half of the 20th century, the first generation of supersonic passenger aircraft was successfully developed, but it had problems with high sonic booms and high fuel consumption during actual operation, and eventually ended with the retirement of the Concorde. The new generation of supersonic passenger aircraft places more emphasis on sonic boom suppression to broaden its application scenarios and markets.

[0003] A sonic boom is an acoustic phenomenon produced when an aircraft flies at supersonic speed: the shock waves and expansion waves generated by the aircraft propagate from high altitude to the ground, causing the ears of people on the ground to feel the change in air pressure and produce the hearing of an explosion. The study of sonic booms can be divided into two categories: near field and far field. The near field generally refers to the position within several times the reference length around the aircraft, and the far field generally refers to the position dozens or hundreds of times the reference length away from the aircraft. The ground is a commonly used research position for the far field.

[0004] The modified linearization theory and computational fluid dynamics (CFD) methods are commonly used to calculate the near-field signal of sonic boom, and the waveform parameter method, augmented Burgers equation method, and KZK equation method are commonly used to calculate the far-field signal. The above methods are all for solving the forward propagation of sonic boom, that is, the sonic boom signal propagates from the aircraft to the near field first, and then propagates to the far field. In addition, the sonic boom signal can also be solved by reverse propagation, which is called sonic boom signal inversion. Sonic boom inversion can give a near-field signal that meets the far-field low sonic boom characteristics, providing a design goal for the optimization of supersonic passenger aircraft, and also has strong research significance and application value.

[0005] Existing sonic boom inversion mostly uses the method of inversely solving the augmented Burgers equation, which constructs the inverse augmented Burgers equation by making the time term of the augmented Burgers equation flow in reverse. However, due to the dissipation in the propagation process of the augmented Burgers equation, the near-field signal obtained by solving the inverse augmented Burgers equation has a flattened waveform. Therefore, in order to improve the inversion accuracy, the inversion distance can only be shortened and the inversion can only be started from the midfield position, which greatly restricts the application scenarios of this method. Summary of the invention

[0006] In order to solve the problems existing in the prior art, the present invention proposes a sonic boom signal intelligent inversion method based on a conditional generative adversarial network. The conditional generative adversarial network cGAN is trained by sonic boom near-field and far-field sample data, so that the near-field signal corresponding to a given far-field signal can be intelligently predicted, and the constraint of the sonic boom inversion distance can be broken. The sonic boom inversion calculation can be performed from any position (including the ground), which greatly improves the application scope of the sonic boom inversion. At the same time, the near-field signal can be designed based on the low sonic boom ground signal, providing a design goal for the low sonic boom optimization of supersonic passenger aircraft.

[0007] The present invention is achieved through the following technical solutions:

[0008] A sonic boom signal intelligent inversion method based on a conditional generative adversarial network comprises the following steps:

[0009] Step S1: Select a sonic boom reference near-field signal and calculate a sonic boom reference far-field signal;

[0010] Step S2: Based on the sonic boom reference far-field signal calculated in step S1, a sonic boom target far-field signal is designed;

[0011] Step S3: adding disturbance to the sonic boom reference near-field signal selected in step S1 to obtain the sonic boom near-field signal after superimposing the disturbance, using the sonic boom near-field signal after superimposing the disturbance to calculate the corresponding sonic boom far-field signal, and establishing a sonic boom near-field and far-field signal sample set;

[0012] Step S4: Using the sonic boom near-field and far-field signal sample sets established in step S3, training a conditional generative adversarial network cGAN;

[0013] The conditional generative adversarial network cGAN includes a discriminator and a generator; wherein the discriminator takes the real samples, labels and generated samples output by the generator in the near and far field signal sample set of the sonic boom as input, and outputs the probability that the input data belongs to the real samples and the generated samples; the generator takes the labels in the near and far field signal sample set of the noise and the sonic boom as input, and outputs the generated samples;

[0014] Step S5: Input the far-field signal of the sonic boom target designed in step S2 into the trained conditional generative adversarial network cGAN, and invert and output the near-field signal of the sonic boom target.

[0015] Furthermore, in step S1, the sonic boom reference near-field signal extraction position is 3 to 4 times the fuselage length below the fuselage.

[0016] Furthermore, in step S1, the selected sonic boom reference near-field signal is used to calculate the sonic boom reference far-field signal by adopting the augmented Burgers equation method; in step S3, the sonic boom near-field signal after superimposing the disturbance is used to calculate the corresponding sonic boom far-field signal by adopting the augmented Burgers equation method.

[0017] Furthermore, in step S2, the process of designing the sonic boom far-field target signal is as follows:

[0018] The signal within the set range of the peak in the sonic boom benchmark far-field signal is truncated, and the sonic boom target far-field signal within the truncation range is supplemented by spline curve interpolation and smoothing, so that the peak is reduced and smoothed; the signal within the set range of the trough in the sonic boom benchmark far-field signal is truncated, and the spline curve interpolation and smoothing is supplemented by the sonic boom target far-field signal within the truncation range, so that the trough is reduced and smoothed.

[0019] Furthermore, in the step S3, the sonic boom near-field signal disturbance waveforms used are divided into three categories: left sawtooth wave, symmetrical triangle wave, and right sawtooth wave; the process of adding disturbance to the sonic boom reference near-field signal is: starting from the starting point of the sonic boom reference near-field signal, a disturbance waveform with a set amplitude and a set duration is added in sections.

[0020] Furthermore, in the step 4, the samples collected in the sonic boom near-field and far-field signal sample set are first standardized: the sonic boom near-field signal in the sonic boom near-field and far-field signal sample set is subtracted from the sonic boom reference near-field signal in step S1 to obtain a sonic boom near-field signal disturbance value; the sonic boom far-field signal in the sonic boom near-field and far-field signal sample set is subtracted from the sonic boom reference far-field signal in step S1 to obtain a sonic boom far-field signal disturbance value; the sonic boom near-field signal disturbance value is used as a true sample after standardized processing, and the sonic boom far-field signal disturbance value is used as a label after standardized processing.

[0021] The present invention also provides a computer device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores program instructions executable by the processor, and when the computer device is running, the processor communicates with the storage medium through the bus, and when the processor executes the program instructions, it is used to implement the intelligent inversion method of sonic boom signals based on a conditional generative adversarial network.

[0022] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a computer, the intelligent inversion method for sonic boom signals based on a conditional generative adversarial network is executed.

[0023] Beneficial effects:

[0024] The present invention proposes a sonic boom signal intelligent inversion method based on a conditional generative adversarial network. By training a conditional generative adversarial network cGAN with sonic boom near-field and far-field sample data, the near-field signal corresponding to a given far-field signal can be intelligently predicted, freeing the constraint of the sonic boom inversion distance. The sonic boom inversion calculation can be performed from any position (including the ground), greatly improving the application scope of the sonic boom inversion. At the same time, the near-field signal can be designed based on the low sonic boom ground signal, providing a design goal for the low sonic boom optimization of supersonic passenger aircraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flow chart of a cGAN-based intelligent inversion method for sonic boom signals proposed in an embodiment of the present invention;

[0026] Figure 2 is a sonic boom reference near-field signal selected in an embodiment of the present invention;

[0027] Figure 3 The sonic boom reference far-field signal calculated by an embodiment of the present invention;

[0028] Figure 4 It is a far-field signal of a sonic explosion target designed in an embodiment of the present invention;

[0029] Figure 5 The sonic explosion near-field signal disturbance used in the embodiment of the present invention;

[0030] Figure 6 A schematic diagram of a conditional generative adversarial network cGAN used in an embodiment of the present invention;

[0031] Figure 7 Schematic diagram of the fully connected layer network structure used by the discriminator in an embodiment of the present invention;

[0032] Figure 8 Schematic diagram of the fully connected layer network structure used by the generator in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention clearer and more understandable, and to enable those skilled in the art to better understand the solutions of the present invention, the present invention is further described and fully described in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] This embodiment takes the LM1021 aircraft as an example, and adopts an intelligent inversion method for sonic boom signals based on a conditional generative adversarial network proposed in the present invention to invert its sonic boom signals to obtain a near-field signal that meets the far-field low sonic boom characteristics.

[0035] like Figure 1As shown, an intelligent inversion method for sonic boom signals based on a conditional generative adversarial network proposed in an embodiment of the present invention comprises the following steps:

[0036] Step S1: Select a sonic boom reference near-field signal and calculate a sonic boom reference far-field signal;

[0037] In the embodiment of the present invention, the flight conditions of LM1021 are shown in Table 1, the sonic boom reference near-field signal extraction position is 3 to 4 times the fuselage length below the fuselage of the LM1021 aircraft, and the sonic boom near-field signal extraction position is shown in Table 2.

[0038] Table 1 LM1021 flight conditions

[0039]

[0040] Table 2 LM1021 sonic boom near-field signal extraction position

[0041]

[0042] Figure 2 is the sonic boom reference near-field signal selected in the embodiment of the present invention, Figure 3 It is a sonic boom reference far-field signal calculated by an embodiment of the present invention.

[0043] Using the selected sonic boom reference near-field signal, the augmented Burgers equation method is used to calculate the sonic boom reference far-field signal, as follows:

[0044] The augmented Burgers equation describes the propagation of sound waves in dissipative media such as the atmosphere. The derivation process is as follows:

[0045] The Navier-Stokes equations are approximated to obtain the Westervelt equations:

[0046]

[0047] In the formula, is the atmospheric pressure; is the speed of sound; is the propagation time starting from the moment when the sound boom occurs; is the atmospheric density; is the classical absorption coefficient; is the nonlinear coefficient;

[0048] The one-dimensional Westervelt equation is:

[0049]

[0050] In the formula, is the distance; introduced to spread to The time at which the delay is the time starting point , then there is the following conversion relationship:

[0051]

[0052] By transforming the coordinates of the one-dimensional Westervelt equation and ignoring the high-order small quantities, we get the classic Burgers equation:

[0053]

[0054] The classic Burgers equation is dimensionless, and dimensionless pressure is introduced , dimensionless distance , reference length , dimensionless time , dimensionless gas dissipation parameter , is the reference pressure, As the reference frequency, the dimensionless classic Burgers equation is obtained:

[0055]

[0056] Considering atmospheric stratification and geometric diffusion on the basis of the classic Burgers equation, the generalized Burgers equation is obtained:

[0057]

[0058] In the formula is the area of ​​the sound tube.

[0059] By adding the molecular relaxation effect term to the generalized Burgers equation, we get the augmented Burgers equation:

[0060]

[0061] In the formula For some gas molecules; is the dimensionless molecular relaxation time; is the dimensionless molecular relaxation coefficient.

[0062] The operator splitting method is used to solve the above augmented Burgers equation, splitting the equation into the following five small equations:

[0063]

[0064] The above equations correspond to nonlinear effects, classical dissipation, inhomogeneous media, geometric diffusion and molecular relaxation effects respectively; ultimately they can be discretely solved using numerical formats.

[0065] Step S2: Based on the sonic boom reference far-field signal calculated in step S1, a sonic boom target far-field signal is designed;

[0066] In step S2, the sonic boom far-field target signal is designed to adopt a low sonic boom waveform modification, that is, the ground waveform is adjusted to a quasi-sine waveform to suppress the high-frequency energy of the waveform. The specific process is as follows:

[0067] The ground far-field signal presents as a sinusoidal waveform, which helps to weaken the high-frequency energy of the sound wave waveform, thereby reducing the intensity of the sonic boom. Therefore, based on the sonic boom baseline far-field signal calculated in step S1, the sonic boom target far-field signal is designed, and the following measures are taken specifically: the signal within the set range of the peak in the sonic boom baseline far-field signal is cut off, and the sonic boom target far-field signal within the cut-off range is supplemented by spline curve interpolation and smoothing, so that the peak is reduced and smoothed; the trough processing method is similar, the signal within the set range of the trough in the sonic boom baseline far-field signal is cut off, and the sonic boom target far-field signal within the cut-off range is supplemented by spline curve interpolation and smoothing, so that the trough is reduced and smoothed. Table 3 shows the peak and trough processing positions of the sonic boom target far-field waveform in an embodiment of the present invention. Figure 4 It is the far-field signal of the sonic explosion target designed in the embodiment of the present invention.

[0068] Table 3 Peak and trough processing positions of far-field waveform of sonic boom target

[0069]

[0070] Step S3: Add disturbance to the sonic boom reference near-field signal selected in step S1 to obtain the sonic boom near-field signal after superimposing the disturbance. Use the sonic boom near-field signal after superimposing the disturbance to calculate the corresponding sonic boom far-field signal by using the augmented Burgers equation method to establish a sonic boom near-field and far-field signal sample set.

[0071] The sonic boom near-field signal disturbance waveforms used in the embodiments of the present invention are divided into three categories: left sawtooth wave, symmetrical triangle wave, and right sawtooth wave. In the embodiments of the present invention, starting from the starting point of the sonic boom reference near-field signal, disturbance waveforms with set amplitude and set duration are added in sections. Table 4 shows the amplitude and duration parameters of the sonic boom near-field signal disturbance waveforms used in the embodiments of the present invention. Figure 5 This is a diagram of the near-field disturbance signal of a sonic explosion used in an embodiment of the present invention.

[0072] Table 4 Waveform parameters of sonic explosion near-field signal disturbance

[0073]

[0074] The corresponding sonic boom far-field signal is calculated using the augmented Burgers equation. The specific calculation method is the same as step S1, and the augmented Burgers equation method is also used. The final sonic boom near-field and far-field signal sample set has a total of 5000 pairs of near-field and far-field signals.

[0075] Step S4: Using the sonic boom near-field and far-field signal sample sets established in step S3, training a conditional generative adversarial network cGAN;

[0076] The conditional generative adversarial network cGAN includes: a discriminator and a generator , each of which contains a network structure; the discriminator is used to judge the authenticity of the input data, and the generator is used to generate fake samples to deceive the discriminator; in the generative adversarial network, the discriminator should judge as accurately as possible whether the input data comes from real samples or samples generated by the generator, and the generator should try to generate samples that the discriminator cannot distinguish between true and false. The discriminator and the generator are trained alternately, and finally the trained generator can generate samples that conform to the distribution of real data.

[0077] Among them, the generator takes noise and labels as input, and outputs generated samples with the same data structure as real samples; the discriminator takes real samples, corresponding labels and generated samples as input, and determines whether the input is a real sample or a generated sample.

[0078] In the embodiment of the present invention, the 5000 sonic boom near-field signals established in step S3 are used as real samples and the 5000 sonic boom far-field signals are used as labels, so that the network can eventually generate a sonic boom near-field signal corresponding to the sonic boom target far-field signal. Figure 6 Schematic diagram of the conditional generative adversarial network cGAN used in the embodiment of the present invention; the step S4 specifically includes the following sub-steps:

[0079] Step S41, collect training samples in the training set and train the discriminator ;

[0080] Before training the network, the samples are standardized: the sonic boom near-field signal in the sonic boom near-field and far-field signal sample set is subtracted from the sonic boom reference near-field signal in step S1 to obtain the sonic boom near-field signal disturbance value; the sonic boom far-field signal in the sonic boom near-field and far-field signal sample set is subtracted from the sonic boom reference far-field signal in step S1 to obtain the sonic boom far-field signal disturbance value; the sonic boom near-field signal disturbance value is used as the real sample after standardization, and the sonic boom far-field signal disturbance value is used as the label after standardization:

[0081]

[0082]

[0083] In the formula, represents the disturbance value of the near-field signal of the sonic boom, Represents the sonic boom near-field signal, Represents the sonic boom reference near field signal; represents the disturbance value of the far-field signal of the sonic explosion, represents the far-field signal of the sonic boom, represents the sonic boom reference far-field signal;

[0084] For the discriminator, the input is the near-field signal disturbance value of the sonic boom as a real sample, the far-field signal disturbance value of the sonic boom as a label, and the generated sample output by the generator. The output is the judgment of the true or false of the input data, that is, whether the input data is a real sample or a generated sample; the probability that the discriminator outputs the input data as a real sample and the generator generated sample is

[0085]

[0086]

[0087] In the formula, represents probability; Represents input data, Indicates the judgment result; Indicates that the input data comes from real samples, Indicates that the input data comes from the generator to generate samples; represents the discriminator, are the discriminator parameters.

[0088] The goal of the discriminator is to improve the accuracy of determining the source of input data, which can be expressed mathematically as minimizing the cross entropy:

[0089] minutes ϕ − (  x [ y log p ( y = 1 | x ) + ( 1 − y ) log p ( y = 0 | x ) ] )

[0090] In the formula, represents mathematical expectation.

[0091] Assuming a probability distribution is the true sample probability distribution , the generator generates sample probability distribution If the mixture is obtained in equal proportions, the above formula can be written as:

[0092] max ϕ (  x ~ p r ( x ) [ l o g D ( x ; ϕ ) ] +  x ′ ~ p i ( x ′ ) [ l o g ( 1 − D ( x ′ ; ϕ ) ) ] ) = max ϕ (  x ~ p r ( x ) [ l o g D ( x ; ϕ ) ] +  z ~ p i ( z ) [ l o g ( 1 − D ( G ( z ; i ) ; ϕ ) ) ] )

[0093] In the formula, Indicates that the discriminator input data comes from real samples; Indicates that the discriminator input data comes from the samples generated by the generator; Indicates that the generator input data comes from noise, that is, a random vector; represents the discriminator, is the discriminator parameter; represents a generator, The generator parameters.

[0094] Rewrite the mathematical expectation in the objective function into integral form:

[0095]  x ~ p r ( x ) [ l o g D ( x ; ϕ ) ] +  z ~ p ( z ) [ l o g ( 1 − D ( G ( z ; i ) ; ϕ ) ) ] = ∫ x p r ( x ; ϕ ) log D ( x ; ϕ ) dx + ∫ x p i ( x ; ϕ ) log( 1 − D ( G ( z ; i ); ϕ )) dx = ∫ x [ p r ( x ; ϕ ) log D ( x ; ϕ ) + p i ( x ; ϕ ) log( 1 − D ( G ( z ; i ); ϕ )) ] dx

[0096] make , , , the above integrand can be written as:

[0097]

[0098] Find extrema via derivatives:

[0099]

[0100] The extreme point is:

[0101]

[0102] In the formula, represents an extreme point, that is, if and only if When Get the best solution.

[0103] Discriminator A fully connected network structure is adopted, which includes 1 input layer, 6 hidden layers and 1 output layer. Figure 7 A schematic diagram of the fully connected layer network structure used by the discriminator in the embodiment of the present invention is given. Table 5 shows the number of neurons in each layer of the discriminator and the type of activation function.

[0104] Table 5 Network structure parameters of the fully connected layer of the discriminator

[0105]

[0106] The network is trained using the Adam optimizer, and the model weights are updated using gradient descent:

[0107]

[0108] In the formula, is the weight; is the learning rate; is the loss function; Derivative of the loss function with respect to the weights.

[0109] Step S42, input noise and train the generator;

[0110] For the generator, the noise and the sonic boom far-field signal disturbance value as the label are input, and the generated fake sonic boom near-field signal disturbance value is output;

[0111] The goal of the generator is to maximize the authenticity of the generated samples. The mathematical expression is:

[0112] max i (  z ~ p i ( z ) [ l o g D ( G ( z ; i ) ; ϕ ) ] ) = minutes i (  z ~ p i ( z ) [ l o g ( 1 − D ( G ( z ; i ) ; ϕ ) ) ] )

[0113] Comprehensive Discriminator Targets and Generators Objective, the objective function of the conditional generative adversarial network cGAN is written as:

[0114] minutes i max ϕ  x ~ p r ( x ) [ l o g D ( x ; ϕ ) ] +  z ~ p i ( z ) [ l o g ( 1 − D ( G ( z ; i ) ; ϕ ) ) ]

[0115] Figure 8 Schematic diagram of the fully connected layer network structure used by the generator in the embodiment of the present invention. The network structure of the fully connected neural network used is: 1 input layer, 6 hidden layers, and 1 output layer. The parameters of each layer are shown in Table 6. The Adam optimizer is used to train the network and the model weights are updated by gradient descent.

[0116] Table 6 Network structure parameters of the generator fully connected layer

[0117]

[0118] Step S43: Repeat the training process of steps S41 and S42, and output the generator after completion. .

[0119] In this embodiment, the mean absolute error (MAE), mean absolute percentage error (MAPE), and correlation coefficient (R 2 ), root mean square error (RMSE) and four other indicators are used to evaluate the accuracy of the conditional generative adversarial network cGAN trained in step S4:

[0120] The mean absolute error (MAE) is calculated as:

[0121]

[0122] The closer the mean absolute error (MAE) is to 0, the better the prediction performance of the model. is the number of data points; is the true value; is the predicted value; the same below; in this embodiment, the average absolute error of the conditional generative adversarial network cGAN after training is 0.06.

[0123] The mean absolute percentage error (MAPE) is calculated as:

[0124]

[0125] The closer the mean absolute percentage error (MAPE) is to 0, the better the prediction performance of the model; in this embodiment, the mean absolute percentage error of the conditional generative adversarial network cGAN after training is 3.2%.

[0126] Correlation coefficient (R 2 ) is calculated as:

[0127]

[0128] Correlation coefficient (R 2 ) is closer to 1, the better the prediction performance of the model, where is the regression sum of squares; is the total sum of squares; in this embodiment, the correlation coefficient of the conditional generative adversarial network cGAN after training is 0.98.

[0129] The root mean square error (RMSE) is calculated as:

[0130]

[0131] The closer the root mean square error (RMSE) is to 0, the better the prediction performance of the model. In this embodiment, the root mean square error of the conditional generative adversarial network cGAN after training is 0.09.

[0132] Step S5: Input the far-field signal of the sonic boom target designed in step S2 into the trained conditional generative adversarial network cGAN, and invert and output the near-field signal of the sonic boom target; the specific process is:

[0133] The difference between the sonic boom target far-field signal and the sonic boom reference far-field signal is input to the generator, and the generator outputs the difference between the sonic boom target near-field signal and the sonic boom reference near-field signal. The output result plus the sonic boom reference near-field signal can obtain the inverted sonic boom target near-field signal. :

[0134]

[0135] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and intent of the present invention.

Claims

1. An intelligent inversion method for sonic boom signals based on conditional generative adversarial networks, characterized by: The following steps are involved: Step S1: Select a sonic boom reference near-field signal and calculate a sonic boom reference far-field signal; Step S2: Based on the sonic boom reference far-field signal calculated in step S1, a sonic boom target far-field signal is designed; Step S3: adding disturbance to the sonic boom reference near-field signal selected in step S1 to obtain the sonic boom near-field signal after superimposing the disturbance, using the sonic boom near-field signal after superimposing the disturbance to calculate the corresponding sonic boom far-field signal, and establishing a sonic boom near-field and far-field signal sample set; Step S4: Using the sonic boom near-field and far-field signal sample sets established in step S3, training a conditional generative adversarial network cGAN; The conditional generative adversarial network cGAN includes a discriminator and a generator; wherein the discriminator takes the real samples, labels and generated samples output by the generator in the near and far field signal sample set of the sonic boom as input, and outputs the probability that the input data belongs to the real samples and the generated samples; the generator takes the labels in the near and far field signal sample set of the noise and the sonic boom as input, and outputs the generated samples; Step S5: Input the far-field signal of the sonic boom target designed in step S2 into the trained conditional generative adversarial network cGAN, and invert and output the near-field signal of the sonic boom target.

2. The method for intelligent inversion of sonic boom signals based on conditional generative adversarial networks according to claim 1 is characterized in that: In the step S1, the sonic boom reference near-field signal extraction position is 3 to 4 times the fuselage length below the fuselage.

3. The intelligent inversion method of sonic boom signal based on conditional generative adversarial network according to claim 1 is characterized by: In the step S1, the selected sonic boom reference near-field signal is used to calculate the sonic boom reference far-field signal by adopting the augmented Burgers equation method; in the step S3, the sonic boom near-field signal after superimposing the disturbance is used to calculate the corresponding sonic boom far-field signal by adopting the augmented Burgers equation method.

4. The intelligent inversion method of sonic boom signal based on conditional generative adversarial network according to claim 1 is characterized by: In step S2, the process of designing the sonic boom far-field target signal is as follows: The signal within the set range of the peak in the sonic boom benchmark far-field signal is truncated, and the sonic boom target far-field signal within the truncation range is supplemented by spline curve interpolation and smoothing, so that the peak is reduced and smoothed; the signal within the set range of the trough in the sonic boom benchmark far-field signal is truncated, and the spline curve interpolation and smoothing is supplemented by the sonic boom target far-field signal within the truncation range, so that the trough is reduced and smoothed.

5. The intelligent inversion method of sonic boom signal based on conditional generative adversarial network according to claim 1 is characterized by: In the step S3, the sonic boom near-field signal disturbance waveforms used are divided into three categories: left sawtooth wave, symmetrical triangle wave, and right sawtooth wave; the process of adding disturbance to the sonic boom reference near-field signal is: starting from the starting point of the sonic boom reference near-field signal, adding disturbance waveforms with set amplitude and set duration in sections.

6. The intelligent inversion method of sonic boom signal based on conditional generative adversarial network according to claim 1 is characterized by: In the step 4, the samples collected in the sonic boom near-field and far-field signal sample set are first standardized: the sonic boom near-field signal in the sonic boom near-field and far-field signal sample set is subtracted from the sonic boom reference near-field signal in step S1 to obtain a sonic boom near-field signal disturbance value; Subtract the sonic boom far-field signal in the sonic boom near-field and far-field signal sample set from the sonic boom reference far-field signal in step S1 to obtain a sonic boom far-field signal disturbance value; The disturbance value of the sonic boom near-field signal is used as the normalized true sample, and the disturbance value of the sonic boom far-field signal is used as the normalized label.

7. A computer device comprising: A processor, a storage medium and a bus, wherein the storage medium stores program instructions executable by the processor, characterized in that: when the computer device is running, the processor and the storage medium communicate through the bus, and the processor executes the program instructions to implement any one of the methods described in claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a computer, the method according to any one of claims 1 to 6 is executed.

Images