A transient external heat flux loading method for infrared heating cage based on neural network

Abstract

This invention provides a neural network-based method for transient external heat flux loading in an infrared heating cage, comprising the following steps: S1: preliminary testing; S2: establishing a system thermal model; S3: predicting power application to the infrared heating cage; S4: transient external heat flux loading; S5: correcting the thermal model; and S6: repeating steps S3-S5. This method addresses the key issues of existing infrared heating cages in thermal testing of large manned spacecraft, such as large thermal capacity, slow response, and difficulty in transient testing. The method offers advantages such as automation, high efficiency, and low cost, and can be used in thermal testing of a variety of large manned spacecraft.

Description

Technical Field

[0001] The present invention relates to the technical field of spacecraft ground testing, and in particular to a transient external heat flux loading method of an infrared heating cage based on a neural network. Background Art

[0002] During spacecraft vacuum thermal testing, an infrared heating cage is typically used as an external heat flux simulator. This infrared simulator uses a resistor as a radiation source. The resistor is typically made of nickel-chromium steel and stainless steel strips. During design, the infrared heating cage is divided into several zones corresponding to the spacecraft. By varying the applied current, the temperature of the resistor changes, thereby varying the heat flux absorbed by the spacecraft. In current testing, the infrared heating cage is controlled using a traditional PID algorithm. However, numerous factors influence the controlled object during testing, and these factors vary from test to test, leading to unstable thermal model parameters. To improve external heat flux control and reduce human intervention during testing, the design and invention of a neural network-based transient external heat flux loading method for the infrared heating cage has positive practical significance. Summary of the Invention

[0003] In response to the problem of improving the external heat flow control level mentioned in the background technology, the present invention proposes a transient external heat flow loading method for an infrared heating cage based on a neural network, which specifically includes the following steps:

[0004] S1: Preliminary test: By applying stepped low power to the heating circuit in sequence, the system temperature and heat flow responses are obtained, and the data required for establishing the thermal model are collected;

[0005] S2: Establishing a system thermal model. Based on the data in step S1, a multi-input and multi-output thermal model of the system is established through a neural network with dynamic characteristics.

[0006] S3: Predict the power applied by the infrared heating cage, and predict the next moment t according to the system thermal model established in step S2. n+1 The infrared heating cage applies power;

[0007] S4: transient external heat flux loading, controlling the infrared heating cage according to the applied power value predicted in step S3;

[0008] S5: Thermal model correction, t n+1 The actual external heat flow at the moment is compared with the target value, and the thermal model is corrected to obtain the corrected system thermal model, making it closer to the real state.

[0009] S6: Repeat steps S3-S5.

[0010] Furthermore, the data in step S1 include the current and voltage values ​​applied by the external heat flow simulation device, as well as the temperature and heat flow value of the corresponding measuring point.

[0011] Furthermore, step S2 is specifically as follows: based on the data from the preliminary test, a multi-input-multi-output thermal model of the system is established through a neural network with dynamic characteristics, the current temperature or heat flow value, the current applied power value and the temperature or heat flow value to be reached at the next moment are selected as input variables, and the output variable of the neural network is the power value applied in the next cycle.

[0012] Furthermore, step S3 is specifically as follows: in the test t n Time point, according to t n+1 Time required heat flow, for t n ~t n+1 The time the infrared heating cage applies power is predicted.

[0013] Furthermore, step S4 specifically includes driving the heater to apply the calculated heat flow; controlling the corresponding external heat flow simulation control device according to the predicted applied power, and controlling the infrared heating cage according to the predicted applied power.

[0014] Furthermore, step S5 is specifically, n+1 The established thermal model is corrected by the difference between the target value of the point and the actually measured temperature or heat flow value.

[0015] Furthermore, in step S5, the correction method is to back-calculate the actual heat flow according to the temperature value measured by the heat flow meter and apply it in a linear correction manner, that is, to apply the heat flux density ,in is the heat flux density, is the emissivity, is the Stefan-Boltzmann constant, The temperature value measured by the heat flow meter.

[0016] Furthermore, in step S5, if a partition corresponds to the measurement values ​​of n heat flow meters, the mean value is calculated according to the fourth power average method, that is, Where n is the number of heat flow meters corresponding to a partition, is the average heat flux density of n heat flux meters.

[0017] Furthermore, the heat flow meter is an adiabatic heat flow meter.

[0018] Beneficial effects: The present invention proposes a transient external heat flux loading method for an infrared heating cage based on a neural network, that is, a thermal model is established based on a neural network, and according to the input conditions, the power to be applied in the next cycle is predicted, thereby improving the level of external heat flux control, improving the degree of system automation, and stabilizing the thermal model parameters. It has the advantages of high efficiency and low cost, and solves the key problems of existing effective infrared heating cages such as large thermal capacity, slow response, and difficulty in transient testing. It can be used in thermal tests of various large manned spacecraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the steps of the neural network-based transient external heat flux loading method for an infrared heating cage in the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] See also Figure 1 , the present invention provides a technical solution:

[0022] A method for transient external heat flux loading of an infrared heating cage based on a neural network specifically includes the following steps: preliminary testing, establishing a system thermal model, predicting power applied to the infrared heating cage, transient external heat flux loading, and thermal model correction.

[0023] The preliminary experiment step is used to obtain the data required to build the model;

[0024] The step of establishing a system thermal model is used to establish a thermal multi-input-multi-output thermal model of the system;

[0025] The power applied to the infrared heating cage is indicated for the test t n Time point, according to t n+1 Time required heat flow, for t n ~t n+1 The infrared heating cage applied power during the time period to predict;

[0026] The thermal model correction step is used to correct the model in time when there is an error between the actual value and the theoretical value.

[0027] Specific examples Figure 1As shown: Conduct a preliminary test. First, connect all powered circuits. Using a programmable power supply, apply a 0.1A current to each circuit in turn. The program records the temperature and heat flow of all measurement circuits. After the test piece's temperature and heat flow stabilize, apply a 0.2A current to each circuit in turn. The program also records the temperature and heat flow of all measurement circuits. This current is applied in steps until the test piece's temperature approaches the predicted temperature during the test. This generates the matrix I (current) and matrix q (heat flow) data.

[0028] Specific examples Figure 1 As shown: Based on the data matrix of the pre-test, automatic modeling is performed through the BP neural network.

[0029] Specific examples Figure 1 As shown, the current temperature or heat flux value, the current applied power value, and the desired temperature or heat flux value at the next moment are selected as input variables. The output variable of the neural network is the power value to be applied in the next cycle. Based on the predicted applied power, the corresponding external heat flux simulation control device is controlled, and the infrared heating cage is controlled according to the predicted applied power.

[0030] Specific examples Figure 1 As shown in the figure: The weights and biases of the model are updated according to the difference between the target value and the actual measured temperature or heat flow value, and then the neural network model is adjusted. This process continues until the error of the network output is reduced to an allowable range.

[0031] Specific examples Figure 1 Shown: The applied power predicted for the next cycle using the revised model.

[0032] A neural network-based transient external heat flux loading method for an infrared heating cage includes the following specific steps:

[0033] A01: Start the automatic self-setting process, that is, start the program;

[0034] A02: Conduct preliminary tests to obtain current, voltage values, and corresponding measuring point temperature and heat flow values;

[0035] A03: Establish thermal model based on neural network;

[0036] A04: predicts the power to be applied in the next cycle based on the current input variables;

[0037] A05: Compare target value with actual value;

[0038] A06: Does not meet the requirements, and the thermal model is modified according to the difference;

[0039] A07: Repeat this process until the end of the experiment;

[0040] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for transient external heat flux loading of an infrared heating cage based on a neural network, characterized in that: The following steps are involved: S1: Preliminary test: By applying stepped low power to the heating circuit in sequence, the system temperature and heat flow responses are obtained, and the data required for establishing the thermal model are collected; S2: Establishing a system thermal model. Based on the data in step S1, a multi-input and multi-output thermal model of the system is established through a neural network with dynamic characteristics. S3: Predict the power applied by the infrared heating cage, and predict the next moment t according to the system thermal model established in step S2. n+1 The infrared heating cage applies power; S4: transient external heat flux loading, controlling the infrared heating cage according to the applied power value predicted in step S3; S5: Thermal model correction, t n+1 The actual external heat flow at the moment is compared with the target value, and the thermal model is corrected to obtain the corrected system thermal model, making it closer to the real state. S6: Repeat steps S3-S5.

2. The method according to claim 1, characterized in that The data in step S1 include the current and voltage values ​​applied by the external heat flow simulation device, as well as the temperature and heat flow value of the corresponding measuring point.

3. The method according to claim 1, characterized in that Specifically, step S2 is to establish a multi-input and multi-output thermal model of the system through a neural network with dynamic characteristics based on the data of the preliminary test, select the current temperature or heat flow value, the current applied power value and the temperature or heat flow value to be reached at the next moment as input variables, and the output variable of the neural network is the power value applied in the next cycle.

4. The method according to claim 1, wherein Step S3 is specifically, in the test t n Time point, according to t n+1 Time required heat flow, for t n ~t n+1 The time the infrared heating cage applies power is predicted.

5. The method according to claim 1, characterized in that Step S4 specifically includes driving the heater to apply the calculated heat flow; controlling the corresponding external heat flow simulation control device according to the predicted applied power, and controlling the infrared heating cage according to the predicted applied power.

6. The method according to claim 1, characterized in that Step S5 specifically includes: n+1 The established thermal model is corrected by the difference between the target value of the point and the actually measured temperature or heat flow value.

7. The method according to claim 1, characterized in that In step S5, the correction method is to back-calculate the actual heat flow based on the temperature value measured by the heat flow meter and apply it in a linear correction manner, that is, to apply the heat flux density ,in is the heat flux density, is the emissivity, is the Stefan-Boltzmann constant, The temperature value measured by the heat flow meter.

8. The method according to claim 7, characterized in that In step S5, if a partition corresponds to the measurement values ​​of n heat flow meters, the mean value is calculated according to the fourth power average method, that is, Where n is the number of heat flow meters corresponding to a partition, is the average heat flux density of n heat flux meters. 9 . The method according to claim 8 , wherein the heat flow meter is an adiabatic heat flow meter.

Images