Total Pressure Control Method for Blowdown Wind Tunnel Based on Dynamic Neural Network and Fuzzy Control

Abstract

The present invention discloses a total pressure control method for a blowdown wind tunnel based on a dynamic neural network and fuzzy control, which relates to the field of wind tunnel tests and includes: S1, constructing a preset opening table for a pressure regulating valve; S2, disconnecting the fuzzy controller and the pressure controller during the wind tunnel startup stage, and using a PI controller for the opening of the pressure regulating valve to control the opening of the pressure regulating valve; S3, determining whether the rough adjustment stage of the opening control of the pressure regulating valve is completed by collecting the total pressure in real time. If it is not completed, continue to control the valve opening, otherwise enter the fine adjustment stage of S4; S4, substituting the real-time collected gas source pressure II and the known target total pressure I and target Mach number I into the neural network preset opening controller to complete the fine adjustment of the opening of the pressure regulating valve by the PI controller; S5, judging whether the fine adjustment stage is over based on the total pressure difference. The present invention provides a total pressure control method for a blowdown wind tunnel based on a dynamic neural network and fuzzy control, which can effectively save the total pressure adjustment time and reduce the test energy consumption.

Description

Technical Field

[0001] The present invention relates to the field of wind tunnel tests. More specifically, the present invention relates to a total pressure control method for a blowdown wind tunnel based on a dynamic neural network and fuzzy control. Background Art

[0002] Total pressure is one of the most concerned aerodynamic parameters in a wind tunnel, and its control accuracy determines the quality of the wind tunnel flow field and the quality of test data. A blowdown wind tunnel releases high-pressure gas stored in a gas tank through a pressure regulating valve, and at the same time adjusts the air flow rate in the wind tunnel by controlling the opening of the pressure regulating valve, thereby controlling the total pressure.

[0003] To reduce the disturbance to the total pressure caused by the pressure regulating valve and the pipeline between the pressure regulating valve and the settling chamber, and improve the dynamic quality of the flow field, the method of increasing the distance between the pressure regulating valve and the settling chamber is often adopted in wind tunnel design. Although this method can reduce the disturbance of the pipeline to the total pressure and improve the dynamic quality of the flow field, for total pressure control, increasing the pipeline length not only causes the volume of the charging pipeline to become larger, prolongs the total pressure control lag time of the valve, but also frequent adjustment is likely to lead to total pressure control oscillation, which not only affects the quality of test data but also increases the energy consumption of wind tunnel tests.

[0004] In this regard, blowdown high-speed wind tunnels often adopt the method of presetting the opening of the pressure regulating valve before the test and using PID control during the test for total pressure control. However, in addition to the valve opening, wind tunnel control also involves comprehensive influences such as external air source pressure changes and Mach number in the wind tunnel test system. The above control method still has the following deficiencies in the practical process:

[0005] First, the method of determining the valve opening before the test based on the valve characteristic curve is affected by factors such as the number of debugging vehicle times and changes in air source conditions. It is difficult to give the accurate opening of the pressure regulating valve under actual test conditions through the valve characteristic curve, resulting in a long charging start-up process for the wind tunnel;

[0006] Second, the conventional PID control algorithm cannot well cope with the disturbance caused by pressure lag and air source pressure fluctuations, resulting in the stability of total pressure control being affected. Summary of the Invention

[0007] An object of the present invention is to solve at least the above problems and / or defects, and provide at least the advantages described later.

[0008] To achieve these objects and other advantages of the present invention, there is provided a total pressure control method for a blowdown wind tunnel based on a dynamic neural network and fuzzy control, including:

[0009] S1. Constructing including the target total pressure 、the target Mach number 、the air source pressure 、the historical total pressure volatility The preset opening table of the pressure regulating valve related to the valve dynamic response delay coefficient For the maximum total pressure and the minimum total pressure difference within multiple control cycles after reaching 95% of the target total pressure, For the duration of multiple control cycles;

[0010] S2. During the wind tunnel startup phase, obtain the target total pressure I , the target Mach number I , the air source pressure I , disconnect the fuzzy controller and the pressure controller, and use the PI controller for the opening of the pressure regulating valve to control the opening of the pressure regulating valve. The coarse adjustment target value of the PI controller for the opening of the pressure regulating valve is obtained through the multi-modal time-delay neural network preset opening controller of the following formula:

[0011]

[0012] In the above formula, is the coarse adjustment target value of the opening controller obtained by using the neural network, f net (.) is the multi-modal time-delay neural network preset opening controller, , are obtained by querying the preset opening table of the pressure regulating valve respectively;

[0013] S3. Determine whether the coarse adjustment stage of the pressure regulating valve opening control is completed by real-time collecting the total pressure. If not completed, continue to control the valve opening. Otherwise, enter the fine adjustment stage of S4;

[0014] S4. Connect the fuzzy controller and the pressure controller for pressure fine adjustment control. Substitute the real-time collected air source pressure II and the known target total pressure I , the target Mach number I into the neural network preset opening controller to obtain the fine adjustment preset opening of the pressure regulating valve ;

[0015] is calculated by the pressure control PI controller ;

[0016] After the control parameters and of the PI controller for the opening of the pressure regulating valve are online optimized by the fuzzy controller, based on , and the actual opening of the pressure regulating valve calculate the opening control amount of the pressure regulating valve to complete the fine adjustment of the opening of the pressure regulating valve by the PI controller for the opening of the pressure regulating valve;

[0017] ​​S5. Make a judgment based on the following formula. If it is not satisfied, return to S4; otherwise, the fine-tuning stage ends:

[0018]

[0019] In the above formula, is the total pressure difference.

[0020] Preferably, in S2, the opening degree PI controller of the pressure regulating valve controls the opening degree of the pressure regulating valve through the following formula:

[0021]

[0022] In the above formula, is the opening degree output value, is the proportional coefficient of the opening degree PI control of the pressure regulating valve, is the integral coefficient of the opening degree PI control of the pressure regulating valve, is the position error between and the actual opening degree

[0023] of the pressure regulating valve. Preferably, in S3, whether the rough adjustment stage is completed is determined by collecting the total pressure in real time and judging whether reaches x of the target total pressure I. If it does not reach, continue the rough adjustment of the valve opening control, where

[0024] Preferably, in S4, the output of the pressure control PI controller is characterized by the following formula:

[0025]

[0026] In the above formula, is the proportional coefficient of the pressure controller, is the integral coefficient of the pressure controller, is the total pressure error, and .

[0027] Preferably, in S4, the fuzzy controller takes the total pressure error and the total pressure error change rate as inputs, and online optimizes the control parameters and of the opening degree PI controller of the pressure regulating valve according to the fuzzy control principle to obtain the optimized control parameters and ;

[0028] The opening degree PI controller of the pressure regulating valve is based on the optimized control parameters and Output the final opening control quantity of the pressure regulating valve through the following formula :

[0029]

[0030] In the above formula, is the corrected position error of the pressure regulating valve, and , is the correction coefficient, is the position correction quantity of the pressure regulating valve, and , is the actual opening of the pressure regulating valve, is the fine-tuning preset opening of the pressure regulating valve.

[0031] The present invention has at least the following beneficial effects: The present invention designs a multi-modal time-delay neural network preset opening controller. Compared with the prior art, it integrates static parameters such as target total pressure, Mach number, and air source pressure, as well as dynamic time-series characteristics such as total pressure historical pressure volatility and valve dynamic response delay coefficient. The valve time delay compensation is realized through the valve dynamic response delay coefficient, and it has the advantages of significantly improving the rapidity and accuracy of total pressure control.

[0032] The present invention introduces a fuzzy controller into the pressure regulating valve opening PI controller, and at the same time realizes the online optimization of the control parameters of the pressure regulating valve opening PI controller through the fuzzy controller, so that when adjusting the opening of the pressure regulating valve, the fluctuation amount will not exceed the target range, which can effectively save the total pressure adjustment time and reduce the test energy consumption.

[0033] The pressure regulating valve opening PI controller of the present invention introduces a control parameter correction amount when calculating the opening of the pressure regulating valve, and has the advantages of being able to correct and finely adjust the total pressure control disturbance in real time and improving the total pressure control accuracy.

[0034] Other advantages, objectives and features of the present invention will be partially reflected by the following description, and partially will be understood by those skilled in the art through the research and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a schematic diagram of the total pressure control process of a blowdown wind tunnel based on a dynamic neural network and fuzzy control according to the present invention;

[0036] Figure 2 is a schematic diagram of the process in the rough adjustment stage in the total pressure control method of a blowdown wind tunnel based on a dynamic neural network and fuzzy control according to the present invention;

[0037] Figure 3 is a schematic diagram of the process in the fine adjustment stage in the total pressure control method of a blowdown wind tunnel based on a dynamic neural network and fuzzy control according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0038] The present invention will be further described in detail below in conjunction with the accompanying drawings, so that those skilled in the art can implement it with reference to the text of the specification.

[0039] The present invention provides a total pressure control method for a blowdown wind tunnel based on a dynamic neural network and fuzzy control. As Figure 1 shown, it is a total pressure control method based on the fusion of preset opening of a dynamic neural network and fuzzy control. Its operation process is as Figures 2 - 3 shown, including:

[0040] S1. Through the analysis of existing test control data, on the basis of the target total pressure , target Mach number and air source pressure , the historical total pressure volatility and valve dynamic response delay coefficient are obtained, and a data query table of a multi-dimensional input space and these five quantities is constructed respectively. is the difference between the maximum total pressure and the minimum total pressure within 5 control cycles after the total pressure reaches 95% of the target total pressure, is the duration of multiple control cycles, generally taking 5 control cycles of 100 ms.

[0041] Design a multi-modal time-delay neural network preset opening controller, and the formula is:

[0042] (1)

[0043] In the above formula, is the rough adjustment target value of the opening controller obtained by using the neural network, f net (.) is a multi-modal time-delay neural network preset opening controller, which integrates static parameters such as target total pressure I , Mach number I and air source pressure I , as well as dynamic time series features such as total pressure historical pressure volatility and valve dynamic response delay coefficient . Valve time-delay compensation is realized through the valve dynamic response delay coefficient , , are obtained by querying the preset opening table of the pressure regulating valve respectively;

[0044] Through the learning of existing test data, the neural network weights of formula (1) are obtained and built-in and fixed.

[0045] S2. Obtain test parameters before the test: Given target total pressure I , target Mach number I , Gas source pressure Ⅰ ;

[0046] S3. Disconnect the fuzzy controller and the pressure controller, enter the wind tunnel startup stage, and use the PI controller for the regulating valve opening to control the regulating valve opening;

[0047] S4. According to the obtained test parameters, query the corresponding total pressure volatility and the valve dynamic response delay coefficient from the data query table in S1, and call formula (1) to calculate the preset opening of the regulating valve , which is used as the target value of the PI controller for the regulating valve opening;

[0048] S5. Call the PI controller for the regulating valve opening to control the valve opening.

[0049] The PI controller for the regulating valve opening uses the PI algorithm, and the formula is as follows:

[0050] (2)

[0051] In the above formula, is the opening output value, is the proportional coefficient of the PI control for the regulating valve opening, and remains unchanged at the initial value, is the integral coefficient of the PI control for the regulating valve opening, and remains unchanged at the initial value, is the position error, which is minus the actual opening of the regulating valve ;

[0052] S6. Real-time collect the total pressure , and judge whether it reaches of the target total pressure Ⅰ (i.e., ), x The value range of is 90 - 100. If it does not reach, continue to control the valve opening, otherwise enter the fine-tuning stage.

[0053] S7. Enter the fine-tuning stage, and real-time collect the total pressure , the actual opening of the regulating valve and the gas source pressure Ⅱ ;

[0054] S8. Perform pressure control, including two parallel control sub-processes. The working steps of one sub-process are S801 - S803, and the other is S810 - S812;

[0055] S801. Calculate the total pressure error ;

[0056] S802. Call the fuzzy controller with the total pressure error and the total pressure error change rate as the input, and online optimize the control parameters and of the pressure regulating valve opening PI controller according to the fuzzy control principle to generate new control parameters and .

[0057] In actual applications, for the test condition of the target total pressure Ⅰ of 150 kPa, the total pressure error is usually required to be no more than 3 kPa. When using the traditional pressure regulating valve opening control, the control parameters and of the pressure regulating valve opening PI controller are kept fixed, which are 25.5 and 0.89 respectively. During the test process of the traditional method, it takes 5 s for the actual total pressure value to reach within the range of 150 kPa ± 3 kPa, and then there are at least 3 more cases where the total pressure fluctuates outside the range of 150 kPa ± 3 kPa. Finally, the actual total pressure value can be stabilized within the range of 150 kPa ± 3 kPa. After adopting the control method in this step, new control parameters and can be generated in real time according to the total pressure error and its change rate in each control cycle. For the test condition of the target total pressure Ⅰ of 150 kPa, it only takes 1 s for the actual total pressure value to reach within the range of 150 kPa ± 3 kPa, and the total pressure is stable without any fluctuation exceeding the target range. It can be seen from this that the present invention saves the total pressure regulation time and effectively reduces the test energy consumption compared with the existing methods;

[0058] S803. Output the optimized control parameters and of the pressure regulating valve opening PI controller;

[0059] S810. Invoke the neural network preset opening controller. As shown in formula (1), input the real-time collected gas source pressure Ⅱ obtained from S7 and the target total pressure Ⅰ obtained from S2 , the target Mach number Ⅰ physical quantities, as well as the corresponding total pressure volatility queried from the S1 data query table and the valve dynamic response delay coefficient , and output the fine-tuning preset opening of the pressure regulating valve;

[0060] S811. Calculate the pressure regulating valve position correction amount :

[0061] (3)

[0062] S812. Calculate the correction amount of the control parameters of the pressure regulating valve opening PI controller: , is a correction coefficient, with a value range of 0.0 to 1.0;

[0063] S9. The pressure controller adopts a PI control algorithm. The algorithm is as follows, and the input is the total pressure error :

[0064] (4)

[0065] In the above formula, is the proportional coefficient of the pressure controller, and its initial value remains unchanged, is the integral coefficient of the pressure controller, and its initial value remains unchanged, is the intermediate opening control output of the pressure regulating valve calculated by the PI control algorithm of the pressure controller;

[0066] S10. Calculate , and substitute it as the corrected position error of the pressure regulating valve into the PI controller for the opening of the pressure regulating valve;

[0067] S11. Call the PI controller for the opening of the pressure regulating valve, and use as the input of the PI control algorithm, and the output is the final opening control amount of the pressure regulating valve:

[0068] (5)

[0069] In the above formula, is the proportional coefficient optimized by S803, is the integral coefficient optimized by S803, is the final opening control amount of the pressure regulating valve;

[0070] S12. Make a judgment based on the following formula:

[0071]

[0072] In the above formula, is the total pressure difference. If it is not satisfied, return to S7; otherwise, execute the next step;

[0073] S13. Perform operations such as data acquisition and changing the angle of attack of the model;

[0074] S14. Judge whether the test end instruction is received. If not, return to S7; otherwise, execute the next step;

[0075] S15. Return the model to zero and shut down the engine, and the test ends.

[0076] The above solution is only an illustration of a preferred example, but it is not limited to this. When implementing the present invention, appropriate substitutions and / or modifications can be made according to the needs of users.

[0077] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those skilled in the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the examples shown and described herein.

Claims

1. A total pressure control method for a temporary wind tunnel based on dynamic neural network and fuzzy control, characterized in that: include: S1. Construct a system that includes the target total pressure , Target Mach number , Gas source pressure , Historical total pressure fluctuation rate and valve dynamic response delay coefficient The preset opening table of the pressure regulating valve, To achieve 95% of the target total pressure, the maximum and minimum total pressure differences within multiple control cycles are: is the duration of multiple control cycles; S2. During the wind tunnel startup phase, obtain the target total pressure I 、Target Mach number I 、Gas source pressure Ⅰ , disconnect the fuzzy controller and the pressure controller, and use the pressure regulating valve opening PI controller to control the pressure regulating valve opening. The coarse adjustment target value of the pressure regulating valve opening PI controller is obtained by the multi-modal time-delay neural network preset opening controller of the following formula: In the above formula, The coarse adjustment target value of the opening controller obtained by using the neural network is: f net (.) is a multi-modal time-delay neural network preset opening controller, , Obtained by querying the preset opening table of the pressure regulating valve respectively; S3, collect total pressure in real time To determine whether the coarse adjustment stage of the pressure regulating valve opening control is completed, if not, continue the valve opening control, otherwise enter the fine adjustment stage of S4; S4, connect the fuzzy controller and pressure controller to fine-tune the pressure, and use the real-time collected gas source pressure II and the known target total pressure I 、Target Mach number I Substitute into the neural network preset opening controller to obtain the fine adjustment preset opening of the pressure regulating valve ; Calculated by the pressure control PI controller , The intermediate opening control output of the pressure regulating valve calculated by the PI control algorithm of the pressure controller; Control parameters of PI controller of pressure regulating valve opening by fuzzy controller and After online optimization, based on , and the actual opening of the pressure regulating valve Calculate the opening control amount of the pressure regulating valve , complete the fine adjustment of the pressure regulating valve opening by the pressure regulating valve opening PI controller; S5: Make a judgment based on the following formula. If it is not satisfied, return to S4. Otherwise, the fine-tuning stage ends: In the above formula, is the total pressure difference.

2. The method for controlling total pressure of a temporary wind tunnel based on dynamic neural network and fuzzy control according to claim 1, characterized in that: In S2, the pressure regulating valve opening PI controller controls the opening of the pressure regulating valve through the following formula: In the above formula, is the opening output value, is the proportional coefficient of the PI control of the pressure regulating valve opening, is the integral coefficient of the PI control of the pressure regulating valve opening, for The actual opening of the pressure regulating valve The position error between .

3. The method for controlling total pressure of a temporary wind tunnel based on dynamic neural network and fuzzy control according to claim 1, characterized in that: In S3, whether the rough adjustment stage is completed is determined by real-time acquisition of the total pressure. ,judge Whether the target total pressure I is reached If the value is not reached, the coarse adjustment of the valve opening control is continued, wherein: x The value range is 90 to 100.

4. The method for controlling total pressure of a temporary wind tunnel based on dynamic neural network and fuzzy control according to claim 1, characterized in that: In S4, the output of the pressure control PI controller Characterized by the following formula: In the above formula, is the proportional coefficient of the pressure controller, is the integral coefficient of the pressure controller, is the total pressure error, and .

5. The method for controlling total pressure of a temporary wind tunnel based on dynamic neural network and fuzzy control according to claim 1, characterized in that: In S4, the fuzzy controller uses the total pressure error And the total pressure error change rate As input, according to the fuzzy control principle, the control parameters of the PI controller of the pressure regulating valve opening are and Perform online optimization to obtain optimized control parameters , ; The PI controller of the pressure regulating valve opening is based on the optimized control parameters , , the final opening control amount of the pressure regulating valve is output through the following formula : In the above formula, is the corrected pressure regulating valve position error, and , is the correction factor, is the position correction of the pressure regulating valve, and , is the actual opening of the pressure regulating valve, Fine-tune the preset opening of the pressure regulating valve.

Images