A method for acquiring temperature of a distributed electrically heated deicing surface

Abstract

This invention relates to a method for obtaining the surface temperature of distributed electric heating anti-icing and de-icing systems, specifically within the field of distributed electric heating de-icing. The method is applied to a distributed electric heating anti-icing and de-icing system. In the distributed anti-icing and de-icing area, temperature sensors are deployed in certain areas to obtain the corresponding area temperature. The method includes: collecting first data at the current moment, combining the first data with relevant historical data, and preprocessing it into time series input data required by a surface temperature prediction network to predict the area temperature of the distributed anti-icing and de-icing area without temperature sensors at the next moment; combining the predicted area temperature of the distributed anti-icing and de-icing area without temperature sensors with the measured area temperature of the area with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing and de-icing system. While ensuring the accuracy of the obtained temperatures of each distributed heating area in the electric heating anti-icing and de-icing system, this method reduces the number of temperature sensors required and lowers onboard energy consumption.

Description

Technical Field

[0001] This invention relates to the field of distributed electric heating de-icing technology, specifically to a method for obtaining the surface temperature of distributed electric heating de-icing. Background Technology

[0002] Electrothermal de-icing, as one of the active anti-icing technologies for aircraft, has been widely used in the field of aerospace engineering due to its rapid response characteristics, precise controllability, and mature technical system. This technology utilizes distributed electrothermal elements on key aerodynamic surfaces such as the wing leading edge, tail assembly, and air intake lips. When the aircraft encounters icing weather conditions, a periodic heating strategy is implemented based on an intelligent control module to regulate the surface temperature field distribution, effectively suppressing the formation of critical conditions for ice crystal phase transition, thereby ensuring the aerodynamic performance and flight safety of the aircraft. However, considering the limited onboard energy, efficient and precise control of electrothermal energy has become an inevitable choice for technological development. Distributed electrothermal de-icing, with its advantages of multiple areas and independent control, can achieve fine-grained surface temperature regulation and high energy efficiency. Surface temperature distribution is both a core evaluation indicator of de-icing effectiveness and a key parameter for the optimized design of the electrothermal system. Therefore, research on the surface temperature of distributed electrothermal de-icing is of significant engineering practical importance for the development of intelligent electrothermal de-icing systems, system optimization, and improving the reliability of protection under icing weather conditions, as well as research value for exploring efficient icing protection methods.

[0003] In current electric heating anti-icing systems, multiple heating zones are set up for heating control in order to achieve more precise control. Therefore, it is also necessary to obtain the surface temperature of multiple heating zones. How to accurately obtain the surface temperature of each heating zone while minimizing airborne energy consumption has become an urgent problem to be solved. Summary of the Invention

[0004] The technical problem to be solved by this application is to provide a method for obtaining the surface temperature of distributed electric heating anti-icing system, which can reduce the number of temperature sensors while ensuring the accuracy of the temperature obtained in each distributed heating area of ​​the electric heating anti-icing system, thereby reducing airborne energy consumption.

[0005] One embodiment provides a method for obtaining the surface temperature of a distributed electric heating anti-icing and de-icing system. In this system, temperature sensors are arranged in certain areas of the distributed anti-icing and de-icing zone to obtain the corresponding area temperature. The surface temperature acquisition method includes:

[0006] Collect the first data at the current moment, which includes wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current of each distributed anti-icing zone, temperature of the area with temperature sensors and predicted temperature of the area without temperature sensors.

[0007] The first data is combined with the corresponding historical data and preprocessed into the time series input data required by the surface temperature prediction network to predict the area temperature of the distributed anti-icing area without temperature sensors at the next moment; the surface temperature prediction network is a neural network trained on LSTM to predict the area temperature of the distributed anti-icing area without temperature sensors.

[0008] By combining the predicted area temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured area temperature with temperature sensors, the surface temperature of the distributed electric heating anti-icing and de-icing system is obtained.

[0009] In one embodiment, the arrangement of temperature sensors in the partial area includes: for any distributed anti-icing area that does not have a temperature sensor, at least one adjacent distributed anti-icing area has a temperature sensor.

[0010] In one embodiment, the distributed anti-icing and de-icing area includes a first area Z1, a second area Z2, and a third area Z3 that are sequentially adjacent to each other, and the training method of the surface temperature prediction network includes:

[0011] Components equipped with distributed electric heating anti-icing and de-icing systems will have wind tunnel parameters, electric heating system parameters, and surface temperature data collected based on icing wind tunnel tests, and converted into time-series LSTM input data. The wind tunnel parameters include wind speed, ambient temperature, flight altitude, liquid water content, and average liquid water particle size. The electric heating system parameters include the current in each distributed anti-icing and de-icing zone. The surface temperature data includes the temperatures of the first zone Z1, the second zone Z2, and the third zone Z3.

[0012] Construct an LSTM basic network architecture and use the LSTM input data as the input of the basic network architecture to obtain the predicted temperature of the required partially distributed anti-icing area;

[0013] Based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed and backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network and determine the surface temperature prediction network.

[0014] In one embodiment, a temperature sensor is provided in the second region Z2, while no temperature sensors are provided in the first region Z1 and the third region Z3; the predicted temperature of the required partially distributed anti-icing and de-icing region includes the temperature of the first region Z1 and the temperature of the third region Z3.

[0015] In one embodiment, the acquisition of the first data at the current moment includes: acquiring the wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, and temperature in the second region Z2 at the current moment.

[0016] The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing and de-icing area without temperature sensors at the next moment includes: predicting the temperature of the first region Z1 and the temperature of the third region Z3 at the next moment.

[0017] The method of combining the predicted regional temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured regional temperature with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing includes: combining the predicted temperature TMP1 of the first region Z1 and the temperature TMP3 of the third region Z3, and the collected temperature TMP2 of the second region Z2 to obtain the surface temperatures TMP1, TMP2 and TMP3 of the distributed electric heating anti-icing.

[0018] In one embodiment, the second region Z2 is not equipped with a temperature sensor, while the first region Z1 and the third region Z3 are equipped with temperature sensors; the predicted temperature of the required partially distributed anti-icing area includes the temperature of the second region Z2.

[0019] In one embodiment, the temperature sensor in the third region Z3 is located near the boundary between the first region Z1 and the third region Z3.

[0020] In one embodiment, the acquisition of the first data at the current moment includes: acquiring the wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, temperature in the first region Z1, and temperature in the third region Z3 at the current moment.

[0021] The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing area without temperature sensors includes: predicting the temperature of the second region Z2.

[0022] The method of combining the predicted regional temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured regional temperature with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing includes: combining the predicted temperature TMP2 of the second region Z2 at the next moment, and the collected temperature TMP1 of the first region Z1 and the temperature TMP3 of the third region Z3 to obtain the surface temperatures TMP1, TMP2 and TMP3 of the distributed electric heating anti-icing.

[0023] In one embodiment, the distributed anti-icing and de-icing area includes a fifth region Z5, a fourth region Z4, a first region Z1, a second region Z2, and a third region Z3 that are sequentially adjacent. The training method for the surface temperature prediction network includes:

[0024] Components equipped with distributed electric heating anti-icing and de-icing systems will have wind tunnel parameters, electric heating system parameters, and surface temperature data collected based on icing wind tunnel tests, and converted into time-series LSTM input data. The wind tunnel parameters include wind speed, ambient temperature, flight altitude, liquid water content, and average liquid water particle size. The electric heating system parameters include the current in each distributed anti-icing and de-icing zone. The surface temperature data includes the temperatures of the fifth zone Z5, the fourth zone Z4, the first zone Z1, the second zone Z2, and the third zone Z3.

[0025] Construct an LSTM basic network architecture and use the LSTM input data as the input of the basic network architecture to obtain the predicted temperature of the required partially distributed anti-icing area;

[0026] Based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed and backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network and determine the surface temperature prediction network.

[0027] In one embodiment, temperature sensors are provided in the second region Z2 and the fourth region Z4, while temperature sensors are not provided in the first region Z1, the third region Z3, and the fifth region Z5; the predicted temperature of the required partially distributed anti-icing and de-icing areas includes the temperature of the first region Z1, the temperature of the third region Z3, and the temperature of the fifth region Z5.

[0028] The acquisition of the first data at the current moment includes: wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, current in the fourth region Z4, current in the fifth region Z5, temperature in the second region Z2, and temperature in the fourth region Z4.

[0029] The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing area without temperature sensors includes: predicting the temperature of the first region Z1, the temperature of the third region Z3, and the temperature of the fifth region Z5 at the next moment.

[0030] The method of combining the predicted regional temperatures of the distributed anti-icing and de-icing areas without temperature sensors and the measured regional temperatures with temperature sensors to obtain the surface temperatures of the distributed electric heating anti-icing and de-icing includes: combining the predicted temperatures of the first region Z1 (TMP1), the third region Z3 (TMP3), and the fifth region Z5 (TMP5), and the collected temperatures of the second region Z2 (TMP2) and the fourth region Z4 (TMP4) to obtain the surface temperatures of the distributed electric heating anti-icing and de-icing, namely TMP1, TMP2, TMP3, TMP4, and TMP5.

[0031] The beneficial effects of this invention are:

[0032] By using a trained surface temperature prediction network to predict the area temperature of distributed anti-icing zones without temperature sensors, the number of temperature sensors can be reduced, thereby lowering onboard energy consumption. Within the distributed anti-icing zone, temperature sensors are deployed in some areas to obtain their corresponding temperatures. Based on these measured temperatures, the area temperature of distributed anti-icing zones without temperature sensors is predicted, improving the accuracy of the acquired temperatures for each distributed heating zone. Furthermore, by predicting the area temperature of distributed anti-icing zones without temperature sensors based on the current in each distributed anti-icing zone, the accuracy of the acquired temperatures for each distributed heating zone is further ensured. Therefore, while maintaining the accuracy of the acquired temperatures for each distributed heating zone of the electric heating anti-icing system, the number of temperature sensors required is reduced, thus lowering onboard energy consumption. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of an airfoil component according to one embodiment of this application;

[0034] Figure 2 This is a schematic diagram of the heating area according to one embodiment of this application;

[0035] Figure 3 This is a schematic flowchart of a distributed electric heating anti-icing and de-icing surface temperature acquisition method according to an embodiment of this application;

[0036] Figure 4 This is a schematic diagram of an LSTM network structure according to an embodiment of this application;

[0037] Figure 5This is a schematic diagram of the training method of a surface temperature prediction network according to an embodiment of this application;

[0038] Figure 6 This is a schematic diagram of a surface temperature prediction network architecture according to an embodiment of this application.

[0039] In the diagram, 01 is the leading edge region, 02 is the heating region, and 03 is the temperature measuring point. Detailed Implementation

[0040] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0041] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0042] The serial numbers assigned to components in this article, such as "first" and "second", are used only to distinguish the objects being described and have no sequential or technical meaning.

[0043] To facilitate the explanation of the inventive concept of this application, the following is a brief description of the distributed electric heating anti-icing surface temperature acquisition technology.

[0044] In current electrically heated anti-icing systems, multiple heating zones (including a heating control matrix) are set up for heating control to achieve more precise control, thus requiring the surface temperature of these multiple heating zones. To obtain the surface temperature of each heating zone, current methods require temperature measurement of each zone, meaning a temperature sensor is installed for each zone. However, this increases onboard energy consumption.

[0045] In light of this, the applicant found in the research that an LSTM (Long Short-Term Memory) prediction model for predicting the surface temperature of each region can be trained using existing parameters. However, the applicant also found in the research that if the temperature of all regions is predicted based solely on a simple historical temperature control method, the accuracy of the predicted surface temperature cannot meet the requirements, and the predictive controllability is poor. Taking a distributed anti-icing zone comprising sequentially adjacent regions Z1, Z2, and Z3 as an example, based on an LSTM surface temperature prediction model and the evaluation criteria of the airfoil component electrothermal anti-icing wind tunnel test, and considering the control requirements and applications of the electrothermal anti-icing system, selective predictions were made for the corresponding area temperatures TMP1, TMP2, and TMP3 in regions Z1, Z2, and Z3. Three prediction scenarios were implemented: ① Using wind tunnel parameters combined with the measured area temperature TMP2 as input, predictions for area temperatures TMP1 and TMP3 were made; ② Using wind tunnel parameters combined with the measured area temperature TMP1 as input, predictions for area temperatures TMP2 and TMP3 were made; ③ Using wind tunnel parameters as input, area temperatures TMP1, TMP2, and TMP3 were directly predicted. The prediction input data was a segment of data arbitrarily selected from the test set. The results are shown in Table 1.

[0046] Table 1. Correlation coefficient R and root mean square error C of the prediction results RMSE

[0047]

[0048] From the prediction results in Table 1, the correlation coefficient R and root mean square error C RMSE It is evident that the prediction results obtained by this prediction method have a correlation of over 89% with the actual measurements, and the variation patterns of the predicted data and the actual measured data are basically consistent. Furthermore, comparing the root mean square error C... RMSE The difference lies in the input C, which includes the measured area temperature TMP2 and the predicted area temperatures TMP1 and TMP3. RMSE The minimum value is the input of the predicted temperature TMP2 and TMP3, which includes the actual measured temperature TMP1. Then, the input of the predicted temperature TMP1, TMP2, and TMP3 is the input of the measurement points without actual measured temperature.

[0049] In view of this, this application provides a method for obtaining the surface temperature of distributed electric heating anti-icing and de-icing systems. On one hand, by using a trained surface temperature prediction network to predict the area temperature of distributed anti-icing and de-icing areas without temperature sensors, the number of temperature sensors can be reduced, thereby lowering airborne energy consumption. On the other hand, in the distributed anti-icing and de-icing areas, temperature sensors are deployed in some areas to obtain the corresponding area temperature, and the area temperature of distributed anti-icing and de-icing areas without temperature sensors is predicted based on the measured temperature, improving the accuracy of the obtained temperatures of each distributed heating area. Thirdly, the area temperature of distributed anti-icing and de-icing areas without temperature sensors is predicted based on the current of each distributed anti-icing and de-icing area, further ensuring the accuracy of the obtained temperatures of each distributed heating area. Thus, while ensuring the accuracy of the obtained temperatures of each distributed heating area in the electric heating anti-icing and de-icing system, the number of temperature sensors is reduced, thereby lowering airborne energy consumption.

[0050] To fully illustrate the proposed solution, the following description uses a distributed electric heating anti-icing and de-icing system arranged on the leading edge of an aircraft airfoil component as an example.

[0051] Please refer to Figure 1 and Figure 2 , Figure 1 The airfoil component employs a distributed electric heating protection method. An electric heating anti-icing system is installed in the leading edge region 01. By deploying the electric heating elements along the chord of the component at the leading edge section, [the following can be observed]. Figure 2 The five heating zones 02 include the first zone Z1 in the middle, and the second zone Z2, the third zone Z3, the fourth zone Z4, and the fifth zone Z5 on both sides. Each zone can be heated and controlled independently. In the traditional arrangement, each zone has a temperature measuring point 03, and the temperature of each zone is obtained by placing temperature sensors at the temperature measuring points.

[0052] One embodiment of this application provides a method for obtaining the surface temperature of a distributed electric heating anti-icing system, which can be applied to the aforementioned distributed electric heating anti-icing system. Temperature sensors are arranged in certain areas of the distributed anti-icing area to obtain the corresponding area temperature; please refer to... Figure 3 Methods for obtaining surface temperature include:

[0053] Step S10: Collect the first data at the current moment.

[0054] The first data includes wind speed, ambient temperature, flight altitude, liquid water content, average liquid water particle size, current in each distributed anti-icing zone, temperature predictions for areas with temperature sensors and areas without temperature sensors.

[0055] The first data collected includes the wind speed at the current moment. Ambient temperature Flight altitude Liquid water content Average particle size of liquid water Current in each distributed anti-icing zone Temperature in areas where temperature sensors are installed Temperature prediction values ​​for areas without temperature sensors , where t represents the current time, i, p and q represent the indices of the distributed electric heating region, and p ≠ q.

[0056] The applicant found in the study that in the distributed anti-icing and de-icing area, some areas are equipped with temperature sensors to obtain the corresponding area temperature, and the area temperature of the distributed anti-icing and de-icing area without temperature sensors is predicted based on the measured temperature. Although this can improve the accuracy of the temperature of each distributed heating area, it is still difficult to guarantee the accuracy of the temperature of each distributed heating area.

[0057] Therefore, in this application, the current of each distributed anti-icing zone is combined to predict the temperature of the distributed anti-icing zone without temperature sensors, thereby further ensuring the accuracy of the obtained temperatures of each distributed heating zone.

[0058] In one embodiment, to further improve the accuracy of the predicted regional temperature of distributed anti-icing areas without temperature sensors, the temperature sensor placement must satisfy the following condition: for any distributed anti-icing area without a temperature sensor, at least one adjacent distributed anti-icing area must have a temperature sensor. For example, if the distributed anti-icing areas include a first area Z1, a second area Z2, and a third area Z3 that are sequentially adjacent, the temperature sensor placement scheme includes placing temperature sensors in the first area Z1 and the second area Z2, or placing temperature sensors in the first area Z1 and the third area Z3, or placing temperature sensors only in the second area Z2, or placing temperature sensors in the second area Z2 and the third area Z3.

[0059] In summary, as long as the above relationship conditions are met, it is acceptable. To minimize the number of temperature sensors required, we can choose to place temperature sensors only in the second region Z2, thereby predicting the regional temperatures of the first region Z1 and the third region Z3. In this case, i=1, 2, 3, p=2, q=1, 3.

[0060] Step S20 involves combining the first data with corresponding historical data and preprocessing it into the time-series input data required by the surface temperature prediction network to predict the area temperature of the distributed anti-icing and de-icing area without temperature sensors at the next moment. The surface temperature prediction network is a neural network trained on LSTM to predict the area temperature of the distributed anti-icing and de-icing area without temperature sensors.

[0061] LSTM is derived from the forgetting gate f t Input gate i t Output gate o t Cell state c t For composition, please refer to Figure 4 In the picture, Indicates multiplication. This represents addition. The input is the cell state c at the previous time step t-1. t-1 and the output h of LSTM t-1 and the external input x at the current time t t After processing by LSTM, the output h at the current time t is output. t and cell state c t The specific calculation can be expressed as:

[0062] (1)

[0063] (2)

[0064] (3)

[0065] (4)

[0066] (5)

[0067] (6)

[0068] Where tanh is the hyperbolic tangent function, and σ is the sigmoid activation function. This represents the candidate cell state. W f W i W c W o and b f b i b c b o These represent the weights and biases for the forget gate, input gate, and candidate cell states, respectively.

[0069] The cell state and gate structure characteristics of Long Short-Term Memory (LSTM) networks enable precise control of long-sequence information. Based on the experimental data of electrothermal de-icing for airfoil components, surface temperature objectively reflects the effectiveness of electrothermal de-icing, and its changes are influenced by the duration of electrothermal heating, exhibiting a long-term dependence. Therefore, for the surface temperature of electrothermal de-icing, LSTM can be used to capture long-term temperature change information, and a surface temperature prediction network can be established based on a neural network model framework.

[0070] In one embodiment of this application, the distributed anti-icing and de-icing zone includes a first zone Z1, a second zone Z2, and a third zone Z3 that are sequentially adjacent. Please refer to [reference needed]. Figure 5 The training methods for surface temperature prediction networks include:

[0071] In step S100, the component equipped with the distributed electric heating anti-icing system is subjected to icing wind tunnel testing, during which wind tunnel parameters, electric heating system parameters, and surface temperature data are collected and converted into time-series LSTM input data. The wind tunnel parameters include wind speed, ambient temperature, flight altitude, liquid water content, and average liquid water particle size. The electric heating system parameters include the current in each distributed anti-icing zone. The surface temperature data includes the temperatures of the first zone Z1, the second zone Z2, and the third zone Z3.

[0072] Step S200: Construct the LSTM basic network architecture and use the LSTM input data as the input of the basic network architecture to obtain the predicted temperature of the required distributed anti-icing area.

[0073] Please refer to Figure 6 The surface temperature prediction network of this application mainly includes an input layer, hidden layers, fully connected layers, and an output layer. The input layer consists of multi-dimensional time-series feature vectors, typically encapsulated into a three-dimensional array (S, T, X), where S is the list dimension, T is the time dimension, X is the feature dimension, and Y represents the output. Figure 6 In this context, k represents the number of lists, m represents the number of dimensions of the time series, and n represents the number of feature dimensions. The hidden layer consists of multiple LSTM units, which accept input layer data, perform the calculation process of equations (1) to (6), and output to the next layer. The fully connected layer performs local information integration and data dimension transformation. The output layer then obtains the final prediction result.

[0074] In step S300, based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed, and the loss function is backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network parameters and determine the surface temperature prediction network.

[0075] Therefore, wind tunnel parameters, electrothermal system parameters, and surface temperature data collected during wind tunnel tests of airfoil components for electrothermal de-icing and icing prevention can be converted into multi-dimensional time series input layer data. Feature learning is performed using LSTM, and after processing by a fully connected layer, a preliminary predicted temperature is obtained. The deviation between this predicted temperature and the experimentally measured temperature is fed back into backpropagation to adjust the model parameters. This process is iterated repeatedly to continuously optimize the model parameters and determine the surface temperature prediction network. Finally, the data to be predicted is processed through the network to output an accurate predicted temperature.

[0076] In one embodiment, a temperature sensor is provided in the second region Z2, while no temperature sensors are provided in the first region Z1 and the third region Z3. The predicted temperature of the required partially distributed anti-icing and de-icing area includes the temperature of the first region Z1 and the temperature of the third region Z3. Therefore, in the training method from steps S100 to S300, the temperatures of the first region Z1 and the third region Z3 are used as label data, and the predicted temperatures of the first region Z1 and the third region Z3 (output Y values) are subtracted from the actual temperatures of the first region Z1 and the third region Z3 to obtain the desired temperature. Calculate the loss value.

[0077] Based on the surface temperature prediction network trained in steps S100 to S300, the first data collected in step S10 includes the current wind speed, ambient temperature, flight altitude, liquid water content, average liquid water particle size, current in the first region Z1, current in the second region Z2, current in the third region Z3, and temperature in the second region Z2. In step S20, the temperatures of the first region Z1 and the third region Z3 at the next moment are predicted.

[0078] In one embodiment, the second region Z2 is not equipped with a temperature sensor, while the first region Z1 and the third region Z3 are equipped with temperature sensors; the predicted temperature of the required distributed anti-icing area includes the temperature of the second region Z2. In this case, the applicant found in the study that, considering the regional characteristics of distributed electrothermal heating, the leading edge protection part is composed of multiple heating areas spliced ​​together. In order to improve the accuracy of the predicted regional temperature, the temperature sensor in the third region Z3 is located close to the boundary line between the first region Z1 and the third region Z3.

[0079] In one embodiment of this application, the distributed anti-icing and de-icing area includes a fifth area Z5, a fourth area Z4, a first area Z1, a second area Z2, and a third area Z3 that are sequentially adjacent. The training method for the surface temperature prediction network includes:

[0080] In step S1000, the components equipped with the distributed electric heating anti-icing and de-icing system are subjected to wind tunnel tests based on icing wind tunnel experiments, which collect wind tunnel parameters, electric heating system parameters, and surface temperature data, and convert them into time series LSTM input data.

[0081] Unlike step S100, the surface temperature data includes the collected temperatures of the fifth region Z5, the fourth region Z4, the first region Z1, the second region Z2, and the third region Z3.

[0082] Step S2000: Construct the LSTM basic network architecture and use the LSTM input data as the input of the basic network architecture to obtain the predicted temperature of the required distributed anti-icing area.

[0083] In step S3000, based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed, and the loss function is backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network parameters and determine the surface temperature prediction network.

[0084] Based on the surface temperature prediction network trained in steps S1000 to S3000, in one embodiment, temperature sensors are provided in the second region Z2 and the fourth region Z4, while temperature sensors are not provided in the first region Z1, the third region Z3, and the fifth region Z5. The predicted temperatures of the required distributed anti-icing and de-icing areas include the temperatures of the first region Z1, the third region Z3, and the fifth region Z5. In the training method of steps S1000 to S3000, the temperature data of the first region Z1, the third region Z3, and the fifth region Z5 are used as label data. The predicted temperature data (output Y values) of the first region Z1, the third region Z3, and the fifth region Z5 are then subtracted from the actual temperature data of the first region Z1, the third region Z3, and the fifth region Z5 to obtain the predicted Y values. Calculate the loss value.

[0085] Based on the surface temperature prediction network trained in steps S1000 to S3000, the first data collected in step S10 includes the current wind speed, ambient temperature, flight altitude, liquid water content, average liquid water particle size, current in the first region Z1, current in the second region Z2, current in the third region Z3, temperature in the second region Z2, and temperature in the fourth region Z4. In step S20, the temperatures of the first region Z1, the third region Z3, and the fifth region Z5 at the next moment are predicted.

[0086] Step S30: Combine the predicted area temperature of the distributed anti-icing and de-icing area without temperature sensors with the measured area temperature with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing and de-icing system.

[0087] In the distributed anti-icing and de-icing area, there are three adjacent areas: a first area Z1, a second area Z2, and a third area Z3. The second area Z2 is equipped with a temperature sensor, while the first area Z1 and the third area Z3 are not equipped with temperature sensors. Given that the predicted temperature of the required portion of the distributed anti-icing and de-icing area includes the temperature of the first area Z1 and the temperature of the third area Z3, the surface temperatures TMP1, TMP2, and TMP3 of the distributed electric heating anti-icing and de-icing area are obtained by combining the predicted temperature TMP1 of the first area Z1 and the temperature TMP3 of the third area Z3, and the collected temperature TMP2 of the second area Z2.

[0088] In the distributed anti-icing and de-icing area, there are three adjacent areas: a first area Z1, a second area Z2, and a third area Z3. Temperature sensors are installed in the first area Z1 and the third area Z3, but no temperature sensor is installed in the second area Z2. When the predicted temperature of the required part of the distributed anti-icing and de-icing area includes the temperature of the second area Z2, the surface temperatures of the distributed electric heating anti-icing and de-icing area TMP1, TMP2, and TMP3 are obtained by combining the predicted temperature TMP2 of the second area Z2 with the collected temperatures TMP1 of the first area Z1 and TMP3 of the third area Z3.

[0089] In the distributed anti-icing and de-icing area, there are sequentially adjacent fifth area Z5, fourth area Z4, first area Z1, second area Z2, and third area Z3. Temperature sensors are installed in second area Z2 and fourth area Z4, while temperature sensors are not installed in first area Z1, third area Z3, and fifth area Z5. Given that the predicted temperature of the required distributed anti-icing and de-icing area includes the temperatures of first area Z1, third area Z3, and fifth area Z5, the surface temperatures of the distributed electric heating anti-icing and de-icing system are obtained by combining the predicted temperatures of first area Z1 (TMP1), third area Z3 (TMP3), and fifth area Z5 (TMP5), as well as the collected temperatures of second area Z2 (TMP2) and fourth area Z4 (TMP4).

[0090] Based on the above scheme, the surface temperature prediction network, trained to predict the area temperature of the distributed anti-icing and de-icing zone without temperature sensors, can reduce the number of temperature sensors and thus reduce airborne energy consumption. Within the distributed anti-icing and de-icing zone, temperature sensors are deployed in some areas to obtain their corresponding temperatures, and the area temperature of the distributed anti-icing and de-icing zone without temperature sensors is predicted based on the measured temperatures, improving the accuracy of the acquired temperatures of each distributed heating zone. Furthermore, the area temperature of the distributed anti-icing and de-icing zone without temperature sensors is predicted based on the current of each distributed anti-icing zone, further ensuring the accuracy of the acquired temperatures of each distributed heating zone. Therefore, while ensuring the accuracy of the acquired temperatures of each distributed heating zone in the electric heating anti-icing and de-icing system, the number of temperature sensors is reduced, thereby reducing airborne energy consumption.

[0091] One embodiment of this application provides a computer-readable storage medium storing a program, the stored program including methods that can be loaded by a processor and processed in any of the above embodiments.

[0092] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0093] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A method for obtaining the surface temperature of a distributed electric heating anti-icing and de-icing system, characterized in that, In the distributed anti-icing and de-icing area, temperature sensors are deployed in some areas to obtain the corresponding area temperature; the surface temperature acquisition method includes: Collect the first data at the current moment, which includes wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current of each distributed anti-icing zone, temperature of the area with temperature sensors and predicted temperature of the area without temperature sensors. The first data is combined with the corresponding historical data and preprocessed into the time series input data required by the surface temperature prediction network to predict the area temperature of the distributed anti-icing area without temperature sensors at the next moment; the surface temperature prediction network is a neural network trained on LSTM to predict the area temperature of the distributed anti-icing area without temperature sensors. By combining the predicted area temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured area temperature with temperature sensors, the surface temperature of the distributed electric heating anti-icing and de-icing system is obtained.

2. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 1, characterized in that, The provision of temperature sensors in certain areas includes: for any distributed anti-icing area that does not have a temperature sensor, at least one adjacent distributed anti-icing area has a temperature sensor.

3. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 1 or 2, characterized in that, The distributed anti-icing and de-icing area includes a first area Z1, a second area Z2, and a third area Z3 that are sequentially adjacent. The training method for the surface temperature prediction network includes: Components equipped with distributed electric heating anti-icing and de-icing systems will have wind tunnel parameters, electric heating system parameters, and surface temperature data collected based on icing wind tunnel tests, and converted into time-series LSTM input data. The wind tunnel parameters include wind speed, ambient temperature, flight altitude, liquid water content, and average liquid water particle size. The electric heating system parameters include the current in each distributed anti-icing and de-icing zone. The surface temperature data includes the temperatures of the first zone Z1, the second zone Z2, and the third zone Z3. Construct an LSTM basic network architecture and use the LSTM input data as the input of this basic network architecture to obtain the predicted temperature of the required partially distributed anti-icing area; Based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed and backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network and determine the surface temperature prediction network.

4. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 3, characterized in that, The second region Z2 is equipped with a temperature sensor, while the first region Z1 and the third region Z3 are not equipped with temperature sensors; the predicted temperature of the required partially distributed anti-icing and de-icing areas includes the temperature of the first region Z1 and the temperature of the third region Z3.

5. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 4, characterized in that, The acquisition of the first data at the current moment includes: wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, and temperature in the second region Z2. The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing and de-icing area without temperature sensors at the next moment includes: predicting the temperature of the first region Z1 and the temperature of the third region Z3 at the next moment. The method of combining the predicted regional temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured regional temperature with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing includes: combining the predicted temperature TMP1 of the first region Z1 and the temperature TMP3 of the third region Z3, and the collected temperature TMP2 of the second region Z2 to obtain the surface temperatures TMP1, TMP2 and TMP3 of the distributed electric heating anti-icing.

6. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 3, characterized in that, No temperature sensor is installed in the second region Z2, while temperature sensors are installed in the first region Z1 and the third region Z3; the predicted temperature of the required partially distributed anti-icing area includes the temperature of the second region Z2.

7. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 6, characterized in that, The temperature sensor in the third region Z3 is located near the boundary between the first region Z1 and the third region Z3.

8. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 6, characterized in that, The acquisition of the first data at the current moment includes: wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, temperature in the first region Z1, and temperature in the third region Z3. The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing area without temperature sensors includes: predicting the temperature of the second region Z2. The method of combining the predicted regional temperature of the distributed anti-icing and de-icing area without temperature sensors and the measured regional temperature with temperature sensors to obtain the surface temperature of the distributed electric heating anti-icing includes: combining the predicted temperature TMP2 of the second region Z2 at the next moment, and the collected temperature TMP1 of the first region Z1 and the temperature TMP3 of the third region Z3 to obtain the surface temperatures TMP1, TMP2 and TMP3 of the distributed electric heating anti-icing.

9. The method for obtaining the surface temperature of distributed electric heating anti-icing and de-icing as described in claim 1 or 2, characterized in that, The distributed anti-icing and de-icing areas include sequentially adjacent fifth area Z5, fourth area Z4, first area Z1, second area Z2, and third area Z3. The training method for the surface temperature prediction network includes: Components equipped with distributed electric heating anti-icing and de-icing systems will have wind tunnel parameters, electric heating system parameters, and surface temperature data collected based on icing wind tunnel tests, and converted into time-series LSTM input data. The wind tunnel parameters include wind speed, ambient temperature, flight altitude, liquid water content, and average liquid water particle size. The electric heating system parameters include the current in each distributed anti-icing and de-icing zone. The surface temperature data includes the temperatures of the fifth zone Z5, the fourth zone Z4, the first zone Z1, the second zone Z2, and the third zone Z3. Construct an LSTM basic network architecture and use the LSTM input data as the input of this basic network architecture to obtain the predicted temperature of the required partially distributed anti-icing area; Based on the actual and predicted temperatures of the partially distributed anti-icing and de-icing areas, a loss function is constructed and backpropagated to adjust the model parameters. The network parameters are then iterated repeatedly to continuously optimize the network and determine the surface temperature prediction network.

10. The method for obtaining the surface temperature of distributed electric heating anti-icing as described in claim 9, characterized in that, Temperature sensors are installed in the second region Z2 and the fourth region Z4, while temperature sensors are not installed in the first region Z1, the third region Z3 and the fifth region Z5; the predicted temperature of the required partially distributed anti-icing and de-icing areas includes the temperature of the first region Z1, the temperature of the third region Z3 and the temperature of the fifth region Z5. The acquisition of the first data at the current moment includes: wind speed, ambient temperature, flight altitude, liquid water content, average particle size of liquid water, current in the first region Z1, current in the second region Z2, current in the third region Z3, current in the fourth region Z4, current in the fifth region Z5, temperature in the second region Z2, and temperature in the fourth region Z4. The process of combining the first data with corresponding historical data and preprocessing it into the time series input data required by the surface temperature prediction network to predict the regional temperature of the distributed anti-icing area without temperature sensors includes: predicting the temperature of the first region Z1, the temperature of the third region Z3, and the temperature of the fifth region Z5 at the next moment. The method of combining the predicted regional temperatures of the distributed anti-icing and de-icing areas without temperature sensors and the measured regional temperatures with temperature sensors to obtain the surface temperatures of the distributed electric heating anti-icing and de-icing includes: combining the predicted temperatures of the first region Z1 (TMP1), the third region Z3 (TMP3), and the fifth region Z5 (TMP5), and the collected temperatures of the second region Z2 (TMP2) and the fourth region Z4 (TMP4) to obtain the surface temperatures of the distributed electric heating anti-icing and de-icing, namely TMP1, TMP2, TMP3, TMP4, and TMP5.

Images