A simulation method and platform for embedded control systems of adaptive cycle engines

Abstract

The present invention discloses a simulation method and platform for an adaptive cycle engine embedded control system. First, according to the adaptive cycle engine component-level model, an adaptive cycle engine neural network dynamic model is established to complete the transplantation of the code to the embedded processor; secondly, an adaptive cycle engine control plan is designed, a control algorithm is developed, and the automatic code generation is completed and transplanted to the embedded processor; then the communication method between the engine simulator and the controller is designed, and the communication interface function of the engine model and the control algorithm is developed; finally, a rapid prototype is developed to monitor the analog signal acquisition of the controller output, and the experimental verification is completed. This not only overcomes the shortcomings of the previous advanced control system design that can only be verified on the PC side and cannot simulate the noise in the actual signal transmission, but also improves the development efficiency of the control system, reduces the development cost, and has a positive role in verifying whether the advanced control system can be applied to aircraft engines.

Description

Technical Field

[0001] The invention belongs to the field of aero-engine control system simulation platform design, and in particular relates to an adaptive cycle engine embedded control system simulation method and platform. Background Art

[0002] The engine structure is complex, and the working state is bad and changeable. The engine control system is an effective way to ensure the safe and stable flight of the aircraft. Advanced aircraft engines have higher requirements for safety, stability and reliability. It must not only enable the engine to work stably and reliably under changing conditions, meet performance requirements and control functions, but also ensure that the engine has sufficient safety margin and give full play to its performance benefits. With the continuous improvement of aircraft engine performance requirements, aircraft engine control systems have become more and more complex, but the current domestic aircraft engine control system development process has always been carried out according to the traditional serial process, resulting in low system development efficiency, long cycle, high cost and lack of flexible verification methods. With the increase in the complexity of the control system, if it is developed according to the traditional method, the accuracy and operating efficiency of manually written code cannot be well guaranteed. On the other hand, the traditional simulation platform used for experimental verification uses too many traditional instruments, and the platform's scalability, maintainability, development time, instrument cost and humanization are not satisfactory.

[0003] In recent years, as chip technology matures, embedded processors have become an effective tool for solving complex system development and simulation. By selecting a suitable embedded processor chip to simulate the entire control system multiple times and conduct physical in-loop tests, the feasibility of the system's hardware and software solutions can be verified. The Raspberry Pi 4B hardware development board is a small, low-cost single-board computer based on the Linux system. By installing the official operating system, it will come with more than 35,000 software packages and pre-compiled software. This not only provides users with a full-featured LXDE desktop environment, but also provides commonly used program packages and tools. At the same time, the Raspberry Pi 4B uses a 64-bit quad-core processor with a main frequency of up to 1.5GHz, which can meet the computing power requirements of the engine model and control algorithm. In the early stage of control system design, the controller platform is quickly built, and the control algorithm developed by Matlab / Simulink software is converted into C language form and downloaded to the hardware platform through automatic code generation, and the controlled object is connected for real-time simulation, which facilitates and verifies each design step in a timely manner, discovers problems in the control algorithm in advance, and enables the control algorithm to be quickly adjusted and optimized, minimizing the introduction of errors from the source. The use of embedded processors can integrate various stages such as control algorithm design, software development, and hardware platform construction. This not only overcomes the deficiency that advanced control systems could only be verified on PCs in the early stages of design, but also improves the development efficiency of the control system and reduces research and development costs. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings that the previous advanced control system can only be verified on the PC side in the early stage of design and cannot simulate the noise occurring in the actual signal transmission. A simulation method and platform for an adaptive cycle engine embedded control system are provided, which are used for rapid verification in the design stage of an adaptive cycle engine control system, verify the effectiveness of the control algorithm, improve the development efficiency of the control system, reduce the development cost, and solve the problem of efficient design and rapid verification of the control system.

[0005] Technical solution: The present invention adopts the following technical solution to solve the above technical problems:

[0006] Step 1) According to a certain type of adaptive cycle engine component level model, a neural network dynamic model of the adaptive cycle engine is established, and the code is transplanted to the embedded processor;

[0007] Step 2) Design an adaptive cycle start control plan, develop a control algorithm, complete automatic code generation and port it to an embedded processor;

[0008] Step 3) Design the communication method between the engine simulator and the controller, and develop the communication interface function of the engine model and control algorithm;

[0009] Step 4) Develop a rapid prototype machine to collect and monitor the analog signal output by the controller to complete experimental verification.

[0010] As an implementation scheme of the adaptive cycle engine embedded control system simulation method of the present invention, step 1) includes:

[0011] Step 1.1) The input of the neural network dynamic model is selected as the flight altitude H, flight Mach number Ma, fuel flow rate W f , tail nozzle throat area A8, tail nozzle outlet area A9 and guide vane angles of each component; the output is selected from the low-pressure rotor speed n L , high pressure rotor speed n H , engine pressure ratio EPR, and engine thrust F;

[0012] Step 1.2) generating steady-state point data using an adaptive cycle engine component-level model;

[0013] Step 1.3) Generate steady-state point data as training samples according to the component-level model, and obtain a neural network steady-state model after neural network training;

[0014] Step 1.4) Add inertia link to the steady-state model of the engine neural network to simulate dynamic characteristics;

[0015] Step 1.5) The established engine neural network model is transplanted to the embedded processor, which serves as an engine simulator, and the hardware environment is configured.

[0016] Preferably, the embedded processor may be a Raspberry Pi 4B embedded processor.

[0017] As an implementation scheme of the adaptive cycle engine embedded control system simulation method of the present invention, step 1.2) specifically includes: selecting a working point within the flight envelope, selecting state points within the engine flight envelope at intervals of 1 km in altitude and 0.1 in Mach number; for each state point, f,max ,W f,min ] range, 0.5%W f,d Select the operating point for the interval; where the subscript min represents the lower limit of the parameter, max represents the upper limit of the parameter, and d represents the design point parameter, where W f , A8, A9 take physical values, n L 、n H , F performs similar normalization processing on the design point and at the same time limits the rotor speed for protection.

[0018] As an implementation scheme of the adaptive cycle engine embedded control system simulation method of the present invention, step 2) includes:

[0019] Step 2.1) Select the control object as the adaptive cycle engine neural network dynamic model, W f Adopt closed-loop control, the controlled quantity is n H , other control parameters (flight altitude H, flight Mach number Ma, tail nozzle throat area A8, tail nozzle outlet area A9 and guide vane angles of each component, low-pressure rotor speed n L , engine pressure ratio EPR, and engine thrust F) are controlled by open loop;

[0020] Step 2.2) Design a controller, wherein the closed loop adopts a PI controller and adds a limiting protection function;

[0021] Step 2.3) Convert the built adaptive cycle engine controller into C language form;

[0022] Step 2.4) Port the control algorithm in C language to the embedded processor, which serves as the controller, and configure the hardware environment.

[0023] Preferably, step 2.2) designs the controller based on Matlab / Simulink software.

[0024] As an implementation scheme of the adaptive cycle engine embedded control system simulation method of the present invention, the step 3) includes:

[0025] Step 3.1) reserve an analog signal channel, use software to develop an analog-to-digital conversion program as the signal input of the engine simulator, and use software to develop a digital-to-analog conversion program as the signal output of the controller;

[0026] Step 3.2) Reserve a digital signal channel, use software to develop a sending end program as the signal output of the engine simulator, and use software to develop a receiving end program as the signal input of the controller.

[0027] Preferably, in step 3.1), an analog signal channel is reserved using the PCF8591 chip, an analog-to-digital conversion program is developed using Python3.7 software as the signal input of the engine simulator, and a digital-to-analog conversion program is developed using Geany software as the signal output of the controller; in step 3.2), a digital signal channel is reserved using the TCP communication protocol, a sending end program is developed using Python3.7 software as the signal output of the engine simulator, and a receiving end program is developed using Geany software as the signal input of the controller.

[0028] As an implementation scheme of the adaptive cycle engine embedded control system simulation method of the present invention, step 4) includes:

[0029] Step 4.1) Develop a program to collect the analog signal output by the controller.

[0030] Step 4.2) Perform hardware connection between the rapid prototype machine and the embedded processor platform according to the acquisition program and complete experimental verification.

[0031] Preferably, in step 4.1), a DAQ module is used in Labview 2018 software to develop an acquisition program for the analog signal output by the controller.

[0032] The present invention provides an adaptive cycle engine embedded control system simulation platform, comprising two embedded processors, serving as an engine simulator and a controller respectively; a plurality of digital-to-analog and analog-to-digital conversion chips, realizing the conversion of the controller output digital signal into the analog signal, and the engine simulator input analog signal into the digital signal; and a PXIe monitoring host computer, realizing the acquisition and monitoring function of the analog signal in the closed-loop circuit.

[0033] The beneficial effects of the present invention are: 1. The embedded processor selected by the present invention is Raspberry Pi 4B, which overcomes the problems of high hardware cost, high computing power requirements, long development cycle, etc. in the development process of aircraft engine control systems, integrates various stages such as control algorithm design, software development, and hardware platform construction, and can quickly deploy and verify engine models and controllers, realize online observation of the entire design and simulation process, and greatly improve the development efficiency of engine control systems.

[0034] 2. The present invention designs a signal transmission module, which realizes the conversion of the control quantity from a digital signal to a voltage signal, and then converts it into a digital signal and substitutes it into the engine model for calculation. Compared with pure digital simulation, it realizes high-precision simulation and real-time transmission characteristics of the sensor, and effectively improves the closed-loop simulation accuracy of the controller. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is the architecture diagram of the embedded control system simulation platform;

[0036] Figure 2 It is a structural diagram of the dynamic model of the neural network of the adaptive cycle engine;

[0037] Figure 3 It is the structure diagram of the control system built by Matlab / Simulink software;

[0038] Figure 4 It is the design structure diagram of the communication method between the engine simulator and the controller;

[0039] Figure 5 is the working point I in digital simulation W f Control loop simulation results;

[0040] Figure 6 is the working point II in digital simulation Wf Control loop simulation results;

[0041] Figure 7(a) shows the working point I in the embedded hardware platform simulation W f Control loop simulation results;

[0042] Figure 7(b) shows the neural network model W of the working point I when simulating on the embedded hardware platform f Input curve;

[0043] Figure 8(a) shows the working point II when simulating on the embedded hardware platform. f Control loop simulation results;

[0044] Figure 8(b) shows the neural network model W of working point II when simulating on the embedded hardware platform f Input curve. DETAILED DESCRIPTION

[0045] In order to help technicians better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods.

[0046] The idea of ​​the present invention is that the Raspberry Pi 4B embedded processor has powerful computing power and can meet the calculation requirements of the aircraft engine model and controller, and the modular and miniaturized embedded processor is also similar to the actual application on the engine; in view of the strong nonlinear characteristics of the aircraft engine, an engine neural network model is established to describe the mapping relationship between input and output; further, in view of the characteristic of noise in the sensor measurement on the engine, the corresponding A / D and D / A conversion chips are selected to simulate the actual noise, and the TCP communication protocol is used for digital signal transmission to meet the real-time transmission characteristics, which effectively improves the closed-loop simulation accuracy of the controller.

[0047] The specific implementation of the present invention takes the control system simulation of an adaptive cycle engine as an example to introduce the content of the present invention in detail, including the following steps:

[0048] Step 1) According to a certain type of adaptive cycle engine component level model, a neural network dynamic model of the adaptive cycle engine is established, and the code is transplanted to the embedded processor;

[0049] Step 1.1) The input of the neural network dynamic model is selected as the flight altitude H, flight Mach number Ma, fuel flow rate W f , tail nozzle throat area A8, tail nozzle outlet area A9 and guide vane angles of each component. Output quantity is selected as low pressure rotor speed n L , high pressure rotor speed n H , engine pressure ratio EPR, and engine thrust F.

[0050] Step 1.2) Generate steady-state point data using the adaptive cycle engine component-level model: Select the working point within the flight envelope, and select the state point within the engine flight envelope at an interval of 1 km altitude and 0.1 Mach number. For each state point, in [W f,max ,W f,min ] range, 0.5%W f,d Select the operating point for the interval. Where the subscript min represents the lower limit of the parameter, max represents the upper limit of the parameter, and d represents the design point parameter. f , A8, A9 physical value, n L 、n H , F performs similar normalization processing on the design point and at the same time limits the rotor speed for protection.

[0051] Step 1.3) Generate steady-state point data as training samples based on the component-level model. The sample dimension is 18 dimensions, including 14-dimensional input and 4-dimensional output. The BP neural network structure has 14 nodes in the input layer, 128 nodes in the hidden layer, and 4 nodes in the output layer. After neural network training, the neural network steady-state model is obtained.

[0052] Step 1.4) Add the inertia link to simulate the dynamic characteristics based on the steady-state model of the engine neural network, so that the dynamic model of the engine neural network in the state above slow speed can be obtained. The structure diagram of the neural network dynamic model is shown in the attached figure. Figure 2 shown.

[0053] Step 1.5) Download the established engine neural network model to the Raspberry Pi 4B embedded processor, which serves as an engine simulator and configures the hardware environment.

[0054] Step 2) Design an adaptive cycle start control plan, develop a control algorithm, complete automatic code generation and port it to an embedded processor;

[0055] Step 2.1) Select the control object as the adaptive cycle engine neural network dynamic model, W f Adopt closed-loop control, the controlled quantity is n H , other control parameters adopt open-loop control.

[0056] Step 2.2) Design a controller based on Matlab / Simulink software, in which the closed-loop circuit adopts a PI controller and adds a limiting protection function.

[0057] Step 2.3) Use Simulink automatic code generation to convert the built adaptive cycle engine controller into C language.

[0058] Step 2.4) Port the control algorithm in C language to the Raspberry Pi 4B embedded processor, which serves as the controller, and configure the hardware environment.

[0059] Step 3) Design the communication method between the engine simulator and the controller, develop the communication interface function of the engine model and control algorithm, and design the communication method structure as shown in the attached figure. Figure 4 As shown;

[0060] Step 3.1) Use the PCF8591 chip to reserve an analog signal channel, use Python3.7 software to develop an analog-to-digital conversion program as the signal input of the engine simulator, and use Geany software to develop a digital-to-analog conversion program as the signal output of the controller.

[0061] Step 3.2) Use the TCP communication protocol to reserve a digital signal channel, use Python3.7 software to develop a sending program as the signal output of the engine simulator, and use Geany software to develop a receiving program as the signal input of the controller.

[0062] Step 4) Develop a rapid prototype machine to collect and monitor the analog signal output by the controller to complete experimental verification.

[0063] Step 4.1) Use the DAQ module in Labview 2018 software to develop an acquisition program for the analog signal output by the controller.

[0064] Step 4.2) Perform hardware connection between the rapid prototype machine and the embedded processor platform according to the acquisition program and complete experimental verification.

[0065] In order to verify the effectiveness of the adaptive cycle engine embedded control system simulation method and platform proposed in the present invention, the adaptive cycle engine control system loop simulation verification is carried out using the hardware platform constructed above. The specific scheme is as follows:

[0066] (1) Two working points of the adaptive engine (IH = 0km, Ma = 0; II.H = 11km, Ma = 1.2) are selected for simulation verification. The control command for working point I is selected according to the output value of the component-level model of this model: The control command for 0 to 3 seconds after the neural network model works stably is n H =0.970, 3 to 11 seconds control command is n H =0.990, the control command from 11 to 16 seconds is n H =0.980; the control instruction of working point II is: the control instruction is n from 0 to 3 seconds after the neural network model runs steadily H =0.937, the control command from 3 to 11 seconds is n H =0.977, the control command from 11 to 16 seconds is n H=0.958. The digital simulation results are shown in the attached Figure 5 , 6 As shown, the simulation results of the embedded control system simulation platform are shown in Figures 7 and 8.

[0067] By the attached Figure 5 As shown in Figure 8, the control effect obtained by the closed-loop verification of the simulation system built by the present invention is consistent with the result obtained by the digital simulation, the adjustment time is less than 4 seconds, the steady-state error is less than 0.1%, and the control effect is good. When the embedded control system simulation platform is used for verification, the controller output curve has a slight fluctuation because the signal is transmitted using an analog signal, and the generated voltage signal has noise, causing the control quantity W f There is noise, which causes small fluctuations in the adaptive cycle engine output, which is consistent with the actual situation.

[0068] The present invention provides an adaptive cycle engine embedded control system simulation method and platform that can realize the seamless integration of the adaptive cycle engine control system from modeling, controller design to automatic code generation and downloading, and physical in-loop simulation process, conveniently and timely testing and verifying each design step, and discovering problems existing in the control algorithm in advance, so that the control algorithm can be quickly adjusted and optimized, and the introduction of errors can be minimized from the source. The use of embedded processors can integrate various stages such as control algorithm design, software development, and hardware platform construction, which not only overcomes the deficiency that the previous advanced control system design can only be verified on the PC side in the early stage, but also improves the development efficiency of the control system and reduces the development cost.

[0069] It should be pointed out that the above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes and substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A simulation method for an adaptive cycle engine embedded control system, characterized in that: The following steps are involved: Step 1) establishing an adaptive cycle engine neural network dynamic model based on the adaptive cycle engine component level model, and completing the code migration to the embedded processor; The step 1) specifically includes: Step 1.1) The input of the neural network dynamic model is selected as the flight altitude H, flight Mach number Ma, fuel flow rate W f , tail nozzle throat area A8, tail nozzle outlet area A9 and guide vane angles of each component; the output is selected from the low-pressure rotor speed n L , high pressure rotor speed n H , engine pressure ratio EPR, and engine thrust F; Step 1.2) generating steady-state point data using an adaptive cycle engine component-level model; Step 1.3) Generate steady-state point data as training samples according to the component-level model. The sample dimension is 18 dimensions, including 14-dimensional input and 4-dimensional output. The BP neural network structure has 14 nodes in the input layer, 128 nodes in the hidden layer, and 4 nodes in the output layer. After neural network training, a neural network steady-state model is obtained; Step 1.4) Add inertia link to the steady-state model of the engine neural network to simulate dynamic characteristics; Step 1.5) transplanting the established engine neural network model to an embedded processor, which serves as an engine simulator, and configuring the hardware environment; Step 2) Design an adaptive cycle start control plan, develop a control algorithm, complete automatic code generation and port it to an embedded processor; The step 2) specifically includes: Step 2.1) Select the control object as the adaptive cycle engine neural network dynamic model, W f Adopt closed-loop control, the controlled quantity is n H , other control parameters adopt open-loop control; Step 2.2) Design a controller, wherein the closed loop adopts a PI controller and adds a limiting protection function; Step 2.3) Convert the built adaptive cycle engine controller into C language form; Step 2.4) transplant the control algorithm in C language to the embedded processor, which serves as the controller, and configure the hardware environment; Step 3) Design the communication method between the engine simulator and the controller, and develop the communication interface function of the engine model and control algorithm; Step 4) Develop a rapid prototype machine to collect and monitor the analog signal output by the controller to complete experimental verification.

2. The method for simulating an adaptive cycle engine embedded control system according to claim 1, characterized in that: The embedded processor is a Raspberry Pi 4B embedded processor.

3. The method for simulating an adaptive cycle engine embedded control system according to claim 1, characterized in that: Step 1.2) includes: selecting a working point within the flight envelope, selecting a state point within the engine flight envelope at an interval of 1 km in altitude and 0.1 in Mach number; for each state point, f,max ,W f,min ] range, 0.5%W f,d Select the operating point for the interval; where the subscript min represents the lower limit of the parameter, max represents the upper limit of the parameter, and d represents the design point parameter, where W f , A8, A9 take physical values, n L 、n H , F performs similar normalization processing on the design point and at the same time limits the rotor speed for protection.

4. The method for simulating an adaptive cycle engine embedded control system according to claim 1, characterized in that: Step 2.2) Design the controller based on Matlab / Simulink software.

5. The method for simulating an adaptive cycle engine embedded control system according to claim 1, characterized in that: The step 3) comprises: Step 3.1) reserve an analog signal channel, use software to develop an analog-to-digital conversion program as the signal input of the engine simulator, and use software to develop a digital-to-analog conversion program as the signal output of the controller; Step 3.2) Reserve a digital signal channel, use software to develop a sending end program as the signal output of the engine simulator, and use software to develop a receiving end program as the signal input of the controller.

6. The method for simulating an adaptive cycle engine embedded control system according to claim 5, characterized in that: Step 3.1) Use the PCF8591 chip to reserve an analog signal channel, use Python3.7 software to develop an analog-to-digital conversion program as the signal input of the engine simulator, and use Geany software to develop a digital-to-analog conversion program as the signal output of the controller; Step 3.2) Use the TCP communication protocol to reserve a digital signal channel, use Python3.7 software to develop a sending end program as the signal output of the engine simulator, and use Geany software to develop a receiving end program as the signal input of the controller.

7. The method for simulating an adaptive cycle engine embedded control system according to claim 1, characterized in that: The step 4) comprises: Step 4.1) Develop a program to collect the analog signal output by the controller; Step 4.2) Perform hardware connection between the rapid prototype machine and the embedded processor platform according to the acquisition program and complete experimental verification.

8. An adaptive cycle engine embedded control system simulation platform for executing the method described in any one of claims 1 to 7, characterized in that: It includes two embedded processors, one as the engine simulator and the other as the controller; several digital-to-analog and analog-to-digital conversion chips, which realize the conversion of the controller output digital signal to analog signal and the engine simulator input analog signal to digital signal. 1 PXIe monitoring host computer to realize the acquisition and monitoring function of analog signals in the closed loop.

Images