Mach number feedforward control device and method based on long short-term memory neural network

Abstract

This specification provides a Mach number feedforward control device and method based on a Long Short-Term Memory (LSTM) neural network. The device includes: a feedforward controller for predicting a first valve opening increment at the current moment using an LSTM model; a PID controller for outputting a second valve opening increment; an arithmetic logic unit (ALU) for calculating the input to the PID controller at the current moment and weighting the valve opening increments output by the feedforward controller and the PID controller; a wind tunnel system for adjusting the valve opening value based on the final valve opening increment, thereby adjusting the Mach number of the test section; a data acquisition unit for acquiring the output data of the wind tunnel system and sending it to the feedforward controller, outputting the current Mach number of the test section to the ALU; and a host computer for acquiring the target Mach number value and sending it to the ALU and the feedforward controller, and for turning the entire device on and off.

Description

Technical Field

[0001] This document relates to the field of artificial intelligence technology, and in particular to a Mach number feedforward control device and method based on a long short-term memory neural network. Background Technology

[0002] Stable control of Mach number in subsonic and transonic wind tunnels has always been a challenging problem. Current Mach number control systems in subsonic and transonic wind tunnels rely on PID control strategies, with PID control parameters manually tuned empirically. Generally, the PID parameters remain constant from the start to the end of control. This control strategy struggles to respond quickly and effectively to external disturbances such as changes in model angle of attack and fluctuations in air source pressure. This results in significant fluctuations in the Mach number around the target value under certain operating conditions, severely impacting the accuracy and reliability of the test data.

[0003] Because Mach number control systems have characteristics such as multivariables, strong coupling, and dynamic lag, the relationship between input and output exhibits a highly complex nonlinear relationship. Traditional linear models or modeling methods based on simplification assumptions are difficult to accurately describe this nonlinear dynamic process. Summary of the Invention

[0004] The purpose of this invention is to provide a Mach number feedforward control device and method based on a long short-term memory neural network, aiming to solve the above-mentioned problems in the prior art.

[0005] This invention provides a Mach number feedforward control device based on a long short-term memory neural network, comprising: The feedforward controller is used to predict the first valve opening increment at the current moment based on historical and future disturbance data and the target Mach number value, using a long short-time memory neural network (LSTM) model. A PID controller is used to output the second valve opening increment based on the difference between the current Mach number and the target Mach number. The arithmetic unit, connected to the feedforward controller and the PID controller, is used to calculate the difference between the current Mach number and the target Mach number input of the PID controller at the current moment, and send it to the PID controller. It also weights the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and outputs the final valve opening increment to the wind tunnel system. A wind tunnel system, connected to the arithmetic unit, is used to adjust the valve opening value according to the final valve opening increment, and to adjust the Mach number of the test section according to the valve opening value. The data acquisition unit is connected to the wind tunnel system, the feedforward controller, and the arithmetic unit. It is used to acquire the output data of the wind tunnel system, send the historical and future N time steps of the output data to the feedforward controller, and output the Mach value of the test segment at the current moment to the arithmetic unit. The host computer, connected to the arithmetic unit and the feedforward controller, is used to obtain the target Mach number value set by the user, send the target Mach number value to the arithmetic unit and the feedforward controller, and turn the Mach number feedforward control device based on the long short-term memory neural network on and off.

[0006] This invention provides a Mach number feedforward control method based on a long short-term memory neural network, used in the aforementioned Mach number feedforward control device based on a long short-term memory neural network, comprising: The model training module is called to train the LSTM of the feedforward controller based on historical flow field test data to obtain an LSTM model. When multiple Mach number target values ​​are set, LSTM training is performed separately for each Mach number target value to obtain multiple LSTM models. The host computer is invoked to obtain the target Mach number value set by the user, and the target Mach number value is sent to the arithmetic unit and the feedforward controller. The Mach number feedforward control device based on long short-term memory neural network is turned on and off. The feedforward controller is invoked to predict the first valve opening increment at the current moment based on historical and future disturbance data and the target Mach number value using a Long Short Time Memory (LSTM) neural network model. The PID controller is invoked to output the second valve opening increment based on the difference between the current Mach number and the target Mach number. The arithmetic unit is invoked to calculate the difference between the current Mach number and the target Mach number input of the PID controller at the current moment, and sends it to the PID controller. The first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment are weighted, and the final valve opening increment is output to the wind tunnel system. The wind tunnel system is invoked to adjust the valve opening value based on the final valve opening increment, and the Mach number of the test section is adjusted based on the valve opening value. The data acquisition unit is invoked to collect the output data of the wind tunnel system, and the output data of the past and future N time steps are sent to the feedforward controller. The Mach value of the test segment at the current moment is output to the arithmetic unit.

[0007] This invention also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the above-described Mach number feedforward control method based on a long short-term memory neural network.

[0008] This invention also provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements the steps of the above-described Mach number feedforward control method based on a long short-term memory neural network.

[0009] By employing the embodiments of the present invention, the Mach number during the testing phase can be controlled, thereby improving the testing accuracy and reliability of the test data. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of a Mach number feedforward control device based on a long short-term memory neural network according to an embodiment of the present invention; Figure 2 This is a detailed structural schematic diagram of the Mach number feedforward control device based on a long short-term memory neural network according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the network structure of the feedforward controller according to an embodiment of the present invention; Figure 4 This is a flowchart of the Mach number feedforward control method based on a long short-term memory neural network according to an embodiment of the present invention; Figure 5 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0012] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0013] Device Examples According to embodiments of the present invention, a Mach number feedforward control device based on a long short-term memory neural network is provided. Figure 1 This is a schematic diagram of a Mach number feedforward control device based on a long short-term memory neural network according to an embodiment of the present invention, as shown below. Figure 1 As shown, the Mach number feedforward control device based on a long short-term memory neural network according to an embodiment of the present invention specifically includes: Feedforward controller 10 is used to predict the first valve opening increment at the current moment using a Long Short-Term Memory (LSTM) neural network model based on historical and future disturbance data and the target Mach number; specifically, feedforward controller 10 is used for: Based on historical and future perturbation data and the target Mach number, the valve opening increment at the current moment is predicted using a Long Short-Term Memory (LSTM) neural network model. The LSTM model specifically includes an input layer, an LSTM layer, and an output layer. The input data for the input layer includes: the model angle of attack values ​​for N historical and future time steps, the gas source pressure values ​​for N historical and future time steps, the valve opening values ​​for N historical time steps, and the Mach number output values ​​for N historical and future time steps. The last two Mach number output values ​​in the future data clock are the target Mach number values, where N is a natural number. The LSTM layer has 256 layers. The output data for the output layer is the first valve opening increment at the current moment.

[0014] PID controller 11 is used to output the second valve opening increment based on the difference between the current Mach number and the target Mach number; The arithmetic unit 12, connected to the feedforward controller and the PID controller, is used to calculate the difference between the current Mach number and the target Mach number input to the PID controller at the current moment, and send it to the PID controller. It also weights the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and outputs the final valve opening increment to the wind tunnel system. Specifically, the arithmetic unit 12 includes: The first calculation module is used to calculate the input of the PID controller at the current moment: e(t) i ) = Ma(t i ) - Ma obj , where e(t) i Ma(t) represents the difference between the current Mach number and the target Mach number. i ) represents the current Mach number, Ma obj Indicates the target Mach number; The second calculation module is used to weight the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and output the final valve opening increment: ΔL(t i) = AΔL1(t i ) + BΔL2(t i ), where ΔL(t) i ) represents the final valve opening increment, ΔL1(t) i ) represents the first valve opening increment, ΔL2(t) i ) represents the second valve opening increment, and A and B are the weighting parameters of the first valve opening increment and the second valve opening increment, respectively.

[0015] The wind tunnel system 13 is connected to the arithmetic unit and is used to adjust the valve opening value according to the final valve opening increment and adjust the Mach number of the test section according to the valve opening value. The data acquisition unit 14 is connected to the wind tunnel system, the feedforward controller, and the arithmetic unit. It is used to acquire the output data of the wind tunnel system, send the output data of the past and the next N time steps to the feedforward controller, and output the Mach number of the test segment at the current moment to the arithmetic unit. The output data specifically includes: the valve opening, air source pressure, model angle of attack, and Mach number output by the wind tunnel system.

[0016] The host computer 15 is connected to the arithmetic unit and the feedforward controller, and is used to obtain the target Mach number value set by the user, send the target Mach number value to the arithmetic unit and the feedforward controller, and turn the Mach number feedforward control device based on the long short-term memory neural network on and off.

[0017] The device further includes: The model training module is used to train the LSTM of the feedforward controller based on historical flow field test data to obtain an LSTM model. When multiple Mach number target values ​​are set, LSTM training is performed separately for each Mach number target value to obtain multiple LSTM models.

[0018] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0019] This invention provides a Mach number feedforward control device based on a Long Short-Term Memory (LSTM) neural network for Mach number control in a transient subspan wind tunnel. The flow field power source of the transient subspan wind tunnel is an upstream high-pressure gas source. A valve is installed between the wind tunnel and the high-pressure gas source. By adjusting the valve opening, the airflow velocity in the wind tunnel test section is adjusted to obtain the target Mach number. Figure 2As shown, the Mach number feedforward control device based on a long short-term memory neural network according to an embodiment of the present invention consists of the following parts: 1. Feedforward controller (1-1, input layer; 1-2, LSTM layer; 1-3, output layer), 2. PID controller, 3. Wind tunnel system, 4. Arithmetic unit 1, 5. Arithmetic unit 2, 6. Data acquisition unit, 7. Host computer, Ma is the Mach number, L is the valve opening, ps is the air source pressure, α is the angle of attack of the test model, and e is the difference between the current Mach number and the target Mach number. 1) Feedforward controller Based on historical and future disturbance data, the required valve opening increment ΔL2 is predicted in advance to stabilize the Mach number at the target value. The feedforward controller is modeled using a Long Short-Time Memory (LSTM) neural network. LSTM has unique advantages in handling the long-term dependencies of time-series data and is suitable for solving time delay and nonlinear problems in Mach number control modeling. The LSTM network model includes an input layer, an LSTM layer, and an output layer (such as...). Figure 3 (As shown). The input data for the input layer includes: a) the model angle of attack values ​​α(t) for the historical and future 20 time steps. i-18 ), α(t i-17 ), ..., α( ti+1 b) Gas source pressure values ​​ps(t) for 20 historical and future time steps i-18 ), ps(t i-17 ), ..., ps(t i+1 c) Valve opening values ​​L(t) over 20 historical time steps i-20 ), L(t i-19 ), ..., L(t i-1 ); d) Mach number output values ​​Ma(t) for 20 time steps in the past and future i-18 ), Ma(t i-17 ), ..., Ma(t i+1 ), where the future data Ma(t) i ) and Ma(t i+1 The value of ) is the target Mach number Ma. obj The LSTM layer has 256 layers. The output data of the output layer is the valve opening increment ΔL2(t) at the current moment. i ).

[0020] 2) PID controller Based on the current Mach number Ma(t) i ) and target Mach number Ma obj The difference e between them, the output valve opening increment ΔL1(t) i ).

[0021] 3) Wind tunnel system The test wind tunnel control system adjusts the Mach number of the test section by adjusting the valve opening value L.

[0022] 4) Arithmetic Unit It includes Arithmetic Logic Unit 1 and Arithmetic Logic Unit 2. Arithmetic Logic Unit 1 is used to calculate the input of the PID controller at the current moment: e(t) i ) = Ma(t i ) - Ma obj Arithmetic unit 2 is used to weight the valve opening increments output by the feedforward controller and the PID controller at the current moment: ΔL(t i ) = AΔL1(t i ) + BΔL2(t i Typically, A = 0.4 and B = 0.6 are chosen, and the final ΔL is output to the wind tunnel system.

[0023] 5) Data collector The system collects data on valve opening, air source pressure, model angle of attack, and Mach number from the wind tunnel system, and outputs the historical and future data for 20 time steps to the feedforward controller, while outputting the current Mach number of the test segment to the arithmetic unit 1.

[0024] 6) Host computer Set the target Mach number Ma obj The output is sent to the arithmetic unit 1 and the feedforward controller, and controls the opening / closing of the entire device.

[0025] During control, the neural network of the feedforward controller is first trained based on historical flow field test data to obtain a feedforward controller model. If the experiment has multiple target Mach numbers, training should be performed separately at each target Mach number to obtain multiple feedforward controller models. Generally, the number of training data points should be no less than 2000 to obtain a sufficiently accurate model. The input, output, and parameters of the training data are... Figure 2 Consistency is key; the input and output data ranges should be sufficiently large to cover all possible operating conditions during the experiment, avoiding the impact of extrapolation on the model's prediction accuracy. Set the target Mach number in the host computer and select the corresponding feedforward controller model. Start the experiment, activate the Mach number control device, and complete the test items after the Mach number stabilizes.

[0026] Method Implementation Examples According to embodiments of the present invention, a Mach number feedforward control method based on a long short-term memory neural network is provided, which is used in a Mach number feedforward control device based on a long short-term memory neural network. Figure 4 This is a flowchart of the Mach number feedforward control method based on a long short-term memory neural network according to an embodiment of the present invention, as shown below. Figure 4 As shown, the Mach number feedforward control method based on a long short-term memory neural network according to an embodiment of the present invention specifically includes: Step S401: Call the model training module to train the LSTM of the feedforward controller based on historical flow field test data to obtain an LSTM model. When multiple Mach number target values ​​are set, LSTM training is performed separately for each Mach number target value to obtain multiple LSTM models. The LSTM model specifically includes: an input layer, an LSTM layer, and an output layer. The input data of the input layer includes: the model angle of attack values ​​of the past and the next N time steps, the air source pressure values ​​of the past and the next N time steps, the valve opening values ​​of the past N time steps, and the Mach number output values ​​of the past and the next N time steps. The last two Mach number output values ​​of the future data clock are the Mach number target values, and N is a natural number. The number of LSTM layers is 256. The output data of the output layer is the first valve opening increment at the current moment.

[0027] Step S402: Call the host computer to obtain the target Mach number value set by the user, send the target Mach number value to the arithmetic unit and the feedforward controller, and turn the Mach number feedforward control device based on long short-term memory neural network on and off. Step S403: Invoke the feedforward controller and predict the first valve opening increment at the current moment using a Long Short Time Memory (LSTM) neural network model based on historical and future disturbance data and the target Mach number. Step S404: Call the PID controller and output the second valve opening increment based on the difference between the current Mach number and the target Mach number; Step S405: Call the arithmetic unit to calculate the difference between the current Mach number and the target Mach number input to the PID controller, and send it to the PID controller. Weight the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and output the final valve opening increment to the wind tunnel system; specifically including: The first arithmetic module is invoked to calculate the input of the PID controller at the current moment: e(t) i ) = Ma(t i ) - Ma obj , where e(t) i Ma(t) represents the difference between the current Mach number and the target Mach number. i ) represents the current Mach number, Ma obj Indicates the target Mach number; The second calculation module is invoked to weight the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and outputs the final valve opening increment: ΔL(t i ) = AΔL1(t i ) + BΔL2(t i), where ΔL(t) i ) represents the final valve opening increment, ΔL1(t) i ) represents the first valve opening increment, ΔL2(t) i ) represents the second valve opening increment, and A and B are the weighting parameters of the first valve opening increment and the second valve opening increment, respectively.

[0028] Step S406: Call the wind tunnel system, adjust the valve opening value according to the final valve opening increment, and adjust the Mach number of the test section according to the valve opening value; Step S407: Call the data acquisition unit to collect the output data of the wind tunnel system, and send the output data of the past and future N time steps to the feedforward controller, and output the Mach value of the test segment at the current moment to the arithmetic unit.

[0029] Device Example 2 This invention provides an electronic device, such as... Figure 5 As shown, it includes: a memory 50, a processor 52, and a computer program stored in the memory 50 and executable on the processor 52, wherein the computer program, when executed by the processor 52, performs the steps as described in the method embodiment.

[0030] Device Example 3 This invention provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor 52, performs the steps described in the method embodiment.

[0031] The computer-readable storage media described in this embodiment include, but are not limited to, ROM, RAM, disk, or optical disk.

[0032] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A Mach number feedforward control device based on a long short-term memory neural network, characterized in that, Mach number control for transient subspan wind tunnels includes: The feedforward controller is used to predict the first valve opening increment at the current moment based on historical and future disturbance data and the target Mach number value, using a long short-time memory neural network (LSTM) model. A PID controller is used to output the second valve opening increment based on the difference between the current Mach number and the target Mach number. The arithmetic unit, connected to the feedforward controller and the PID controller, is used to calculate the difference between the current Mach number and the target Mach number input of the PID controller at the current moment, and send it to the PID controller. It also weights the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and outputs the final valve opening increment to the wind tunnel system. A wind tunnel system, connected to the arithmetic unit, is used to adjust the valve opening value according to the final valve opening increment, and to adjust the Mach number of the test section according to the valve opening value. The data acquisition unit is connected to the wind tunnel system, the feedforward controller, and the arithmetic unit. It is used to acquire the output data of the wind tunnel system, send the historical and future N time steps of the output data to the feedforward controller, and output the Mach value of the test segment at the current moment to the arithmetic unit. The host computer, connected to the arithmetic unit and the feedforward controller, is used to obtain the target Mach number value set by the user, send the target Mach number value to the arithmetic unit and the feedforward controller, and turn the Mach number feedforward control device based on the long short-term memory neural network on and off.

2. The apparatus according to claim 1, characterized in that, The feedforward controller is specifically used for: Based on historical and future perturbation data and the target Mach number, the valve opening increment at the current moment is predicted using a Long Short-Term Memory (LSTM) neural network model. The LSTM model specifically includes an input layer, an LSTM layer, and an output layer. The input data for the input layer includes: the model angle of attack values ​​for N historical and future time steps, the gas source pressure values ​​for N historical and future time steps, the valve opening values ​​for N historical time steps, and the Mach number output values ​​for N historical and future time steps. The last two Mach number output values ​​in the future data clock are the target Mach number values, where N is a natural number. The LSTM layer has 256 layers. The output data for the output layer is the first valve opening increment at the current moment.

3. The apparatus according to claim 1, characterized in that, The arithmetic unit specifically includes: The first calculation module is used to calculate the input of the PID controller at the current moment: e(t) i ) = Ma(t i ) - Ma obj , where e(t) i Ma(t) represents the difference between the current Mach number and the target Mach number. i ) represents the current Mach number, Ma obj Indicates the target Mach number; The second calculation module is used to weight the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and output the final valve opening increment: ΔL(t i ) = AΔL1(t i ) + BΔL2(t i ), where ΔL(t) i ) represents the final valve opening increment, ΔL1(t) i ) represents the first valve opening increment, ΔL2(t) i ) represents the second valve opening increment, and A and B are the weighting parameters of the first valve opening increment and the second valve opening increment, respectively.

4. The method according to claim 1, characterized in that, The output data specifically includes: valve opening, air source pressure, model angle of attack, and Mach number output by the wind tunnel system.

5. The method according to claim 1, characterized in that, The device further includes: The model training module is used to train the LSTM of the feedforward controller based on historical flow field test data to obtain an LSTM model. When multiple Mach number target values ​​are set, LSTM training is performed separately for each Mach number target value to obtain multiple LSTM models.

6. A Mach number feedforward control method based on a long short-term memory neural network, characterized in that, The Mach number feedforward control device based on a long short-term memory neural network according to any one of claims 1 to 5, wherein the method specifically includes: The model training module is called to train the LSTM of the feedforward controller based on historical flow field test data to obtain an LSTM model. When multiple Mach number target values ​​are set, LSTM training is performed separately for each Mach number target value to obtain multiple LSTM models. The host computer is invoked to obtain the target Mach number value set by the user, and the target Mach number value is sent to the arithmetic unit and the feedforward controller. The Mach number feedforward control device based on long short-term memory neural network is turned on and off. The feedforward controller is invoked to predict the first valve opening increment at the current moment using a Long Short Time Memory (LSTM) neural network model based on historical and future disturbance data and the target Mach number. The PID controller is invoked to output the second valve opening increment based on the difference between the current Mach number and the target Mach number. The arithmetic unit is invoked to calculate the difference between the current Mach number and the target Mach number input of the PID controller at the current moment, and sends it to the PID controller. The first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment are weighted, and the final valve opening increment is output to the wind tunnel system. The wind tunnel system is invoked to adjust the valve opening value based on the final valve opening increment, and the Mach number of the test section is adjusted based on the valve opening value. The data acquisition unit is invoked to collect the output data of the wind tunnel system, and the output data of the past and future N time steps are sent to the feedforward controller. The Mach value of the test segment at the current moment is output to the arithmetic unit.

7. The method according to claim 6, characterized in that, The LSTM model specifically includes an input layer, an LSTM layer, and an output layer. The input data of the input layer includes: the model angle of attack values ​​of the past and future N time steps, the gas source pressure values ​​of the past and future N time steps, the valve opening values ​​of the past N time steps, and the Mach number output values ​​of the past and future N time steps. The last two Mach number output values ​​of the future data clock are the target Mach number values, and N is a natural number. The number of LSTM layers is 256. The output data of the output layer is the first valve opening increment at the current moment.

8. The method according to claim 6, characterized in that, The arithmetic logic unit (ALU) is invoked to calculate the difference between the current Mach number and the target Mach number input to the PID controller, and this difference is sent to the PID controller. The first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment are weighted, and the final valve opening increment is output to the wind tunnel system. Specifically, this includes: The first arithmetic module is invoked to calculate the input of the PID controller at the current moment: e(t) i ) = Ma(t i ) - Ma obj , where e(t) i Ma(t) represents the difference between the current Mach number and the target Mach number. i ) represents the current Mach number, Ma obj Indicates the target Mach number; The second calculation module is invoked to weight the first valve opening increment and the second valve opening increment output by the feedforward controller and the PID controller at the current moment, and outputs the final valve opening increment: ΔL(t i ) = AΔL1(t i ) + BΔL2(t i ), where ΔL(t) i ) represents the final valve opening increment, ΔL1(t) i ) represents the first valve opening increment, ΔL2(t) i ) represents the second valve opening increment, and A and B are the weighting parameters of the first valve opening increment and the second valve opening increment, respectively.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the Mach number feedforward control method based on a long short-term memory neural network as described in any one of claims 6 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an implementation program for information transmission, which, when executed by a processor, implements the steps of the Mach number feedforward control method based on a long short-term memory neural network as described in any one of claims 6 to 8.

Images