Spray system and its control method, controller and medium for icing wind tunnel tests

Abstract

This application relates to a spray system for icing wind tunnel testing, its control method, controller, and medium. This application addresses multiple water supply branches of the spray system for icing wind tunnel testing by training a target prediction model using historical test data under multiple operating conditions. This model pre-calculates the second target opening degree of the outlet regulating valve after the solenoid valve opens the nozzle. Then, based on the on / off state signal of the solenoid valve, the outlet regulating valve is controlled to adjust to the second target opening degree. In this way, the coupling relationship between the opening ratio of the outlet regulating valve before and after the solenoid valve opens the nozzle and the steady-state parameters before the solenoid valve opens the nozzle can be learned under different test conditions. This allows for adaptation to multi-condition icing wind tunnel test scenarios, shortens the time for water supply pressure to stabilize, thereby reducing the deterioration of cloud parameters caused by excessive adjustment time and ensuring the accuracy of icing shape simulation.

Description

Technical Field

[0001] This application relates to the field of icing wind tunnel testing technology, specifically to a spray system for icing wind tunnel testing, its control method, controller, and medium. Background Technology

[0002] The spray system is a crucial component of icing wind tunnel testing, used to simulate low-temperature cloud and fog environments at high altitudes. It primarily consists of an air supply system, a water supply system, and spray rakes. The spray rake is the structural component that supports the multiple branches of the spray system. To achieve a highly uniform cloud and fog environment, the spray rake is typically designed in a matrix configuration, with these branches distributed along a specific pattern on the rake, and multiple nozzles arranged on each layer of the rake. The main function of the water supply system is to provide the required pressure to the nozzles within the spray rake, thereby simulating cloud and fog environments with different particle sizes and water contents. The quality of water pressure control determines the quality of the cloud and fog environment simulation.

[0003] During the preparation for icing wind tunnel tests, the water supply system of the spray system is typically controlled in a closed loop by adjusting the water pump speed and the opening of the valves on each spray rake branch to maintain the water supply pressure at the same stable parameters. The moment the spray solenoid valve opens, pressure fluctuations occur, which are usually addressed by readjusting the water pump and valves. The adjustment time is determined by the efficiency of the closed-loop control and is generally around ten seconds. For tests with only a few tens of seconds of spray time, excessive adjustment time can lead to deterioration of cloud and fog parameters, affecting icing formation. Summary of the Invention

[0004] The purpose of this application is to provide a spray system for icing wind tunnel testing, as well as its control method, controller, and medium, to solve the problem that the long water supply pressure adjustment time in icing wind tunnel testing affects the icing shape.

[0005] To achieve the above objectives, the first aspect of this application provides a control method for a spray system in an icing wind tunnel test. The spray system includes multiple water supply branches, and in each water supply branch, an outlet regulating valve and a solenoid valve are connected in series along the water flow direction. The solenoid valve is used to control the on / off flow of water through the nozzles in the water supply branch. The control method includes:

[0006] Historical test data of the spray system under multiple operating conditions are obtained to obtain multiple sets of sample datasets. Each set of sample datasets includes a first sample parameter of the spray system in a steady state before the solenoid valve opens the nozzle, and a second sample parameter of the spray system reaching a steady state again after the solenoid valve opens. The first sample parameter includes the sample opening ratio of the nozzle and the first sample opening degree of the outlet regulating valve. The second sample parameter includes the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio.

[0007] Using the first sample parameters as input and the ratio of the second sample opening degree to the first sample opening degree as output, the initial prediction network is trained to obtain the target prediction model.

[0008] The set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system before the solenoid valve opens the nozzle are obtained. The steady-state parameters are input into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve. The second target opening is calculated based on the first target opening of the outlet regulating valve in the steady-state parameters.

[0009] Based on the on / off status signal of the solenoid valve, the outlet regulating valve is controlled to adjust to the second target opening degree so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0010] A second aspect of this application provides a controller for a spray system used in an icing wind tunnel test. The spray system includes multiple water supply branches, and in each water supply branch, an outlet regulating valve and a solenoid valve are connected in series along the water flow direction. The solenoid valve is used to control the on / off state of the water flow through the nozzles in the water supply branch. The controller includes:

[0011] The acquisition module is used to acquire historical test data of the spray system under multiple operating conditions to obtain multiple sets of sample datasets. Each set of sample datasets includes a first sample parameter of the spray system in a steady state before the solenoid valve opens the nozzle, and a second sample parameter of the spray system reaching a steady state again after the solenoid valve opens. The first sample parameter includes the sample opening ratio of the nozzle and the first sample opening degree of the outlet regulating valve. The second sample parameter includes the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio.

[0012] The training module is used to train the initial prediction network with the first sample parameters as input and the ratio of the second sample opening degree to the first sample opening degree as output, so as to obtain the target prediction model.

[0013] The prediction module is used to obtain the set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system in a steady state before the solenoid valve opens the nozzle. The steady-state parameters are input into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve, and the second target opening is calculated based on the first target opening of the outlet regulating valve in the steady-state parameters.

[0014] The control module is used to control the outlet regulating valve to adjust to the second target opening degree based on the on / off status signal of the solenoid valve, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0015] A third aspect of this application provides a spray system for icing wind tunnel testing, comprising:

[0016] One of the controllers described above;

[0017] Multiple water supply branches, each of which includes an outlet regulating valve, a solenoid valve, and a nozzle;

[0018] The outlet regulating valve and the solenoid valve are connected in series along the water flow direction. The outlet regulating valve and the solenoid valve communicate with the controller respectively. The solenoid valve is used to control the water flow of the nozzle in the water supply branch.

[0019] A fourth aspect of this application provides a computer-readable storage medium storing a program that can be loaded by a processor and executed by the above-described control method for a spray system in an icing wind tunnel test.

[0020] The beneficial effects of this application are:

[0021] This application addresses multiple water supply branches of a spray system used in icing wind tunnel tests. It trains a target prediction model using historical test data from various operating conditions to pre-calculate the second target opening degree of the outlet regulating valve after the solenoid valve opens the nozzle. This allows the model to learn the coupling relationship between the opening ratio of the outlet regulating valve before and after the solenoid valve opens the nozzle and the steady-state parameters before the solenoid valve opens the nozzle under different test conditions, making it adaptable to multi-condition icing wind tunnel test scenarios. Then, based on the on / off state signal of the solenoid valve, the outlet regulating valve is controlled to adjust to the second target opening degree. Compared to traditional post-event closed-loop regulation, this shortens the time for water supply pressure to stabilize, thereby reducing the deterioration of cloud and fog parameters caused by excessive adjustment time and ensuring the accuracy of icing shape simulation.

[0022] Other features and advantages of this application will be described in detail in the following detailed description section. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the structure of a spray system for an icing wind tunnel test provided in an embodiment of this application;

[0024] Figure 2 This is a schematic diagram of the water supply system of a spray system provided in the embodiments of this application;

[0025] Figure 3 This is a flowchart illustrating a control method for an icing wind tunnel spray system provided in an embodiment of this application.

[0026] Figure 4 This is a schematic diagram of the structure of a controller provided in an embodiment of this application.

[0027] Explanation of reference numerals in the attached figures

[0028] 1. Controller; 2. Water supply branch; 3. Outlet regulating valve; 4. Solenoid valve; 5. Nozzle. Detailed Implementation

[0029] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0030] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified. In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to implement and use this application. In the following description, details are set forth for illustrative purposes. It should be understood that those skilled in the art will recognize that this application can be implemented without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessary detail that would obscure the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0031] The spray system of this application embodiment may include a controller and an execution module for multiple water supply branches. The execution module communicates with the controller and performs corresponding operations based on the controller's instructions. The execution module may include an outlet regulating valve, a solenoid valve, and nozzles. Figure 1 This is a schematic diagram of the structure of a spray system for an icing wind tunnel test provided in an embodiment of this application. Figure 1 As shown, the spray system includes a controller 1 and multiple water supply branches 2. In each water supply branch 2, an outlet regulating valve 3 and a solenoid valve 4 are connected in series along the water flow direction. The solenoid valve 4 is used to control the flow of water through the nozzles 5 in the water supply branch.

[0032] Figure 2 This is a schematic diagram of the water supply system of a spray system provided in an embodiment of this application. Figure 2 As shown, the water storage device S1 is used to store water and is the water source for the water supply system in the spray system. The water pump B1 is a device connected to outlet a of the water storage device S1, used to provide power to transport the water from the water storage device S1. The spray rake may include multiple water supply branches 2, and the outlet regulating valve 3 is typically installed on each water supply branch 2. Figure 2 Upstream of (not shown), it is responsible for regulating the water pressure of the water supply branch. Each water supply branch 2 is equipped with a solenoid valve 4, which is used to control the opening and closing of each nozzle 5 in the branch, thereby regulating the water flow of each water supply branch. The return pipeline connects the outlet of the water supply branch 2 to the inlet b of the water storage device S1 to form a circulation system, realizing the functions of water recycling.

[0033] As an example, the outlet regulating valve 3 can be an electrically operated proportional regulating valve. It receives opening commands from the controller 1 and changes the flow area of ​​the pipeline by driving the valve core, thereby regulating the water supply pressure and flow rate of the branch. The solenoid valve 4 can receive on / off control signals from the controller 1. For example, when the digital control signal (DC) of the solenoid valve 4 is 0, it indicates that the solenoid valve 4 is in the closed state. At this time, the water path of the corresponding water supply branch is blocked, and the nozzle will not spray water. When DC = 1, it indicates that the solenoid valve 4 is in the open state, the water path of the corresponding water supply branch is open, and water can be sprayed out through the nozzle to participate in the cloud and fog environment simulation of the icing wind tunnel test.

[0034] Based on the structure of the spray system in the icing wind tunnel test described above, this application provides a control method for the spray system in the icing wind tunnel test. This method aims to solve the problems of large pressure fluctuations and slow adjustment speed of traditional spray systems due to the opening of solenoid valves. It does not involve complex mathematical calculations, can improve the efficiency of water supply pressure regulation, and improve the quality of cloud and fog environment simulation.

[0035] Figure 3 This is a flowchart illustrating a control method for an icing wind tunnel spray system provided in an embodiment of this application. Figure 3 As shown, this control method may include steps 301-304, which will be described in detail below.

[0036] Step 301: Obtain historical test data of the spray system under multiple operating conditions to obtain multiple sets of sample datasets. The operating conditions can cover different seasons, different stages of nozzle wear, different air source pressure fluctuations, different water pump speed levels, different target water supply pressures, and different nozzle opening ratios, etc.

[0037] The multiple sets of operating condition data extracted from historical test data need to include parameters from two key stages: the steady state before the solenoid valve opens the nozzle and the steady state after the solenoid valve opens the nozzle. Steady state refers to a stable state where key parameters of the spray system (such as water supply pressure, valve opening, and pump speed) are in relative equilibrium with minimal fluctuations. For example, the criteria for determining steady state could be that the fluctuation in actual water supply pressure or the change in valve opening is less than a certain set value, and the duration is greater than a set time. Therefore, each sample dataset can include the first set of parameters showing the spray system in a steady state before the solenoid valve opens the nozzle, and the second set of parameters showing the spray system reaching a steady state again after the solenoid valve opens.

[0038] As an example, the first sample parameter may include the sample opening ratio of the nozzle and the first sample opening degree of the outlet regulating valve, and the second sample parameter may include the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio.

[0039] The sample opening ratio refers to the proportion of nozzles actually open out of the total number of nozzles, reflecting the strong spray demand. The first sample opening degree refers to the stable opening value of the outlet regulating valve to maintain the set water supply pressure when the solenoid valve closes the nozzles. The set water supply pressure is the target pressure value preset for this operating condition. The second sample opening degree is the opening value of the outlet regulating valve when the spray system regains stability after the solenoid valve opens the nozzles according to the sample opening ratio.

[0040] By distinguishing between the two steady-state stages before and after opening, the disturbance of the solenoid valve's action on the spray system and its final equilibrium state can be captured, providing complete mapping data for subsequent model training. Collecting data from multiple operating conditions also ensures that the sample dataset includes various operating scenarios that the spray system may face, reducing overfitting of the target prediction model. The mapping relationship between the sample opening ratio of the associated nozzles and the first and second sample opening degrees lays the foundation for learning the implicit correlation between the opening ratio of the outlet regulating valve and operating parameters, solving the problem of fuzzy parameter correlation in the traditional water supply pressure control of spray systems.

[0041] Step 302: Using the first sample parameters as input and the sample opening ratio of the second sample opening to the first sample opening as output, train the initial prediction network to obtain the target prediction model.

[0042] In this embodiment, the initial prediction network refers to an untrained basic neural network. The target prediction model is a model that, after training, can stably output the predicted opening ratio of the outlet control valve. Therefore, the initial prediction network can be trained using the first sample parameters as input and the sample opening ratio of the second sample opening and the first sample opening as output to obtain the target prediction model. For example, a loss function can be calculated based on the difference between the predicted opening ratio and the actual sample opening ratio, and the network weights can be iteratively adjusted using an Adam optimizer or similar method until the loss function converges.

[0043] By capturing the nonlinear relationship between the first sample parameter and the sample opening ratio using a neural network model, the prediction accuracy is higher than that of traditional linear formulas. The target prediction model, trained on multi-condition data, can also adapt to new, unseen conditions without requiring manual adjustment of control parameters. Compared to traditional techniques that require recalibration when changing conditions, it exhibits stronger generalization ability. The output is a predicted opening ratio rather than a direct opening value, and it can also adapt to different first sample openings, enhancing the flexibility of the control logic.

[0044] Step 303: Obtain the set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system before the solenoid valve opens the nozzle. Input the steady-state parameters into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve, and calculate the second target opening based on the first target opening of the outlet regulating valve in the steady-state parameters.

[0045] In each new icing wind tunnel test, the steady-state parameters of the current test condition are first obtained. These steady-state parameters are those of the spray system at steady state before the solenoid valve opens the nozzles, and may include the current nozzle opening ratio (i.e., the planned opening ratio for this test condition), the first target opening degree of the outlet regulating valve (the opening degree to ensure the actual water supply pressure stabilizes at the set water supply pressure), etc. Then, using a trained target prediction model, the predicted opening ratio of the output outlet regulating valve is obtained. Next, based on the product of the first target opening degree and the predicted opening ratio of the outlet regulating valve, the second target opening degree is obtained. The first target opening degree is the opening degree of the outlet regulating valve that stabilizes the spray system before the solenoid valve opens the nozzles. The second target opening degree is the predicted opening degree of the outlet regulating valve that returns the spray system to steady state after the solenoid valve opens the nozzles. In this way, the predicted opening degree of the outlet regulating valve after opening is calculated before the solenoid valve opens, reducing the lag of traditional closed-loop control where there is a disturbance before adjustment, laying the foundation for instantaneous stability. Based on the specific parameters of the current test conditions, the second target opening is calculated in real time through the target prediction model, which solves the problem of accuracy fluctuation caused by experience dependence in manual prediction.

[0046] Step 304: Based on the on / off status signal of the solenoid valve, control the outlet regulating valve to adjust to the second target opening degree so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0047] The solenoid valve's on / off status signal refers to the state signal indicating whether the solenoid valve opens or closes the nozzle. When the solenoid valve opens the nozzle, it indicates that the solenoid valve is in the open state, DC=1. Conversely, when the solenoid valve closes the nozzle, it indicates that the solenoid valve is in the closed state, DC=0. When the solenoid valve's on / off status signal switches from DC=0 to DC=1, it indicates that the signal jump occurs at the instant the solenoid valve opens the nozzle. Based on the jump state of the solenoid valve's on / off status signal, the outlet regulating valve can be controlled to adjust to the predicted second target opening degree, so that the deviation between the adjusted actual water supply pressure and the set water supply pressure is less than the set deviation value. The set deviation value refers to the small deviation that allows the actual water supply pressure to stabilize at the set water supply pressure, indicating that the spray system has returned to a pressure-stabilized state. By synchronizing the solenoid valve's action with the opening adjustment of the outlet regulating valve, the system vibration problem caused by the solenoid valve opening the nozzle in traditional control is reduced, and the water supply pressure stabilization time is shortened, meeting the requirements of short-duration tests in icing wind tunnel experiments.

[0048] This application's embodiments train a target prediction model using historical test data under multiple operating conditions. It pre-calculates the second target opening degree of the outlet regulating valve after the solenoid valve opens the nozzle. This allows the model to learn the coupling relationship between the opening ratio of the outlet regulating valve before and after the solenoid valve opens the nozzle and the steady-state parameters before the solenoid valve opens the nozzle under different test conditions. This approach extends from single-condition adaptation to full-condition generalization, enabling adaptation to multi-condition icing wind tunnel test scenarios. Then, based on the solenoid valve's on / off state signal, the outlet regulating valve is controlled to adjust to the second target opening degree. Compared to traditional post-event closed-loop regulation, this shift from passive regulation to active prediction shortens the time for water supply pressure to stabilize, thereby reducing the deterioration of cloud and fog parameters caused by excessive adjustment time and ensuring the accuracy of icing shape simulation. Furthermore, by moving from single-parameter control to multi-parameter coupling optimization, and comprehensively considering the synergistic effects of factors such as the nozzle opening ratio and the initial first target opening degree, the control accuracy of the spray system's water supply pressure can be improved.

[0049] In this embodiment, the first sample parameters also include sample water supply pressure, sample air supply pressure, and sample water pump speed. Water supply pressure refers to the actual water supply pressure of the spray system in steady state before the solenoid valve opens the nozzle. It is a core parameter reflecting the power output of the water supply system and directly affects the atomization effect of the nozzle. This parameter deviates from the set water supply pressure within the allowable steady-state range. Sample water supply pressure refers to a sample of water supply pressure from historical experimental data, serving as a key input for training the target prediction model. Air supply pressure refers to the compressed air pressure that cooperates with the water supply system in the spray system, used to assist in water atomization. Changes in air supply pressure affect the nozzle flow characteristics, thereby affecting the pressure balance of the water supply branch. Sample air supply pressure refers to a sample of air supply pressure from historical experimental data, and therefore can be used as input to the target prediction model to improve prediction accuracy. Water pump speed refers to the operating speed that provides power to the water supply system. Operating speed is positively correlated with water supply pressure; for example, the higher the operating speed, the higher the water supply pressure usually is. Therefore, operating speed allows the target prediction model to learn the mapping relationship between operating speed, water supply pressure, and outlet regulating valve opening. This is suitable for systems with variable pump speed regulation, reducing opening prediction errors caused by fluctuations in operating speed. The sample operating speed is a sample of operating speeds from historical test data.

[0050] Based on this, in step 302, to improve the accuracy of the sample data, preprocessing operations can be performed on multiple sets of sample datasets. These preprocessing operations may include outlier removal, normalization of the first sample parameters, and sample balancing. Outlier removal identifies and removes outliers from the sample data that deviate from the normal range using statistical methods, reducing interference from outliers with model learning, ensuring the reliability of training data, and reducing the risk of model overfitting. Examples include extreme pressure values ​​caused by sensor malfunctions and obviously unreasonable opening ratios. Normalization of the first sample parameters converts first sample parameters of different dimensions and magnitudes to the same numerical range, eliminating the impact of dimensional differences on neural network training and accelerating neural network convergence. Sample balancing addresses the imbalance in the number of samples for different operating conditions in the first sample parameters by adjusting the sample distribution through oversampling or undersampling, reducing model bias towards operating conditions with larger sample sizes, ensuring balanced predictive ability for all operating conditions, and improving prediction accuracy under extreme operating conditions.

[0051] Then, the sample water supply pressure, sample air supply pressure, sample water pump speed, sample opening ratio, and first sample opening degree are input to obtain the sample prediction ratio. The sample prediction ratio is the ratio of the model-predicted second sample opening degree to the first sample opening degree, used to calculate the loss function by comparing it with the actual sample opening degree ratio. The five processed first sample parameters cover the key states of the spray system's water path, air path, power source, and actuators. Multi-parameter input allows the target prediction model to capture more comprehensive system characteristics, thereby improving prediction accuracy.

[0052] Finally, based on the loss function of the sample prediction ratio and the sample opening ratio, the parameters of the initial prediction network are iteratively updated until the iteration stopping condition is met, resulting in the target prediction model. The initial prediction network can include an input layer, hidden layers, and an output layer. The input layer can include five neurons, each corresponding to the first sample parameter after normalization, responsible for receiving the input sample data. The hidden layers can include multiple layers of neurons with progressively decreasing numbers. For example, a progressively decreasing multi-layer structure can be used, extracting features layer by layer through a non-linear activation function (such as ReLU). The output layer can include one neuron, corresponding to the sample opening ratio. For example, a sigmoid or linear activation function can be used to ensure reasonable values. The progressively decreasing hidden layers can balance feature extraction and computational efficiency, reducing overfitting caused by excessively deep networks. The multi-layer non-linear transformation capability can learn complex parameter coupling relationships, overcoming the limitations of traditional linear models. The model training process can use mean squared error to calculate the difference between the sample prediction ratio and the actual sample opening ratio, then use the backpropagation algorithm to pass the loss value from the output layer to the input layer, using an optimizer to adjust the weights and biases of each layer to reduce the loss value. Termination conditions for model training can typically include the loss function value falling below a preset threshold, the number of iterations reaching the maximum limit, or the validation set loss no longer decreasing after multiple consecutive rounds, in order to ensure the model's generalization ability, reduce overfitting or underfitting, and minimize unnecessary computational resources.

[0053] Because the multiple water supply branches of the spray system have varying spray rake heights and initial target openings, the flow rates entering each branch differ, resulting in different impacts on the water supply pressure. This necessitates adjusting the outlet regulating valves of each branch to maintain uniform water supply pressure, leading to lengthy adjustment times. Therefore, in this embodiment, a branch-independent modeling training strategy is adopted for the multiple water supply branches of the spray system. A dedicated target prediction sub-model is trained for each branch, and these sub-models are combined to form a complete target prediction model. This approach fully considers the different adjustment patterns caused by variations in installation location, pipeline characteristics, and nozzle conditions, achieving refined control and reducing adjustment time.

[0054] In this embodiment, the sample dataset may include first sample parameters and second sample parameters for each water supply branch. The first and second sample parameters for each water supply branch terminal are derived from the test records of the same branch to ensure a strong correlation between the parameters and the branch characteristics. In step 302, each initial prediction sub-network in the initial prediction network can be trained using the first sample parameters of each water supply branch as input and the ratio of the second sample opening degree to the first sample opening degree of each water supply branch as output, to obtain multiple target prediction sub-models. The initial prediction sub-network is the untrained initial neural network corresponding to each water supply branch. The initial values ​​of the weights, biases, and other parameters of each initial prediction sub-network can be randomly generated.

[0055] Then, a target prediction model is obtained based on multiple target prediction sub-models. Each target prediction sub-model corresponds to a water supply branch. The training method for each target prediction sub-model can refer to the training method for the target prediction model described above. The training process of each initial prediction sub-network is executed in parallel or serially, without sharing parameters, ensuring that the learned rules only apply to the corresponding water supply branch. As an example, K-fold cross-validation can also be used to evaluate the performance of each target prediction sub-model. If the accuracy of a certain water supply branch is insufficient, it can be retrained for extreme operating condition samples supplementing that water supply branch.

[0056] Finally, the trained target prediction sub-models are associated and stored according to their numbers, forming a mapping relationship between branch numbers and target prediction sub-models. When it is necessary to predict the target opening of a water supply branch, the corresponding target prediction sub-model can be called from the mapping relationship based on the branch number. The first target opening of the water supply branch is input, and the corresponding predicted opening ratio is output, thus obtaining the second target opening of the water supply branch. In addition, if it is necessary to support the addition of new water supply branches, a new target prediction sub-model can be trained separately and added to the mapping relationship, without retraining the entire target prediction model, thus improving the efficiency of model training.

[0057] By adopting the modeling approach of training each water supply branch separately, this approach overcomes the limitations of traditional single models that cannot adapt to the differences between multiple water supply branches. It provides higher accuracy and more flexible prediction capabilities for multi-path collaborative control of spray systems, and is especially suitable for spray systems in icing wind tunnel tests with complex water supply branches.

[0058] In step 303, the set opening ratio of the nozzles in the current test condition can be determined first, and the spray system can be started. Based on the cloud and fog simulation requirements of the current icing wind tunnel test, the proportion of the number of nozzles to be opened relative to the total number of nozzles can be preset, i.e., the opening ratio can be set, and the power source of the spray system can be started. The set opening ratio can match the test outline, such as the proportion of nozzles to be opened to simulate a high-altitude thin cloud and fog environment. When the spray system is started, the on / off state signal of the solenoid valve can be in the closed state (e.g., DC=0), and only equipment such as water pumps and air compressors are started to establish initial pressure, thereby reducing initial spray interference with the pressure stabilization process.

[0059] Then, based on the set water supply pressure, the pump speed and the opening of the outlet regulating valve are adjusted to bring the spray system to a steady state. The adjustment process can begin by coarsely adjusting the water supply pressure to near the set pressure using the pump speed, and then finely adjusting the water supply pressure to the precise range using the outlet regulating valve. During the adjustment process, fluctuations in the actual water supply pressure, actual pump speed, and the opening of the outlet regulating valve can be monitored in real time. By coarsely adjusting the speed and finely adjusting the opening, adjustment efficiency and accuracy can be balanced, providing stable baseline parameters for subsequent model predictions and reducing data deviations under fluctuating conditions.

[0060] In response to the spray system reaching a steady state, the system acquires the target water supply pressure, target pump speed, target air supply pressure, and the first target opening degree of the outlet regulating valve. The target water supply pressure refers to the actual water supply pressure under steady-state conditions, with the deviation from the set pressure within the allowable range. The target pump speed refers to the actual pump speed under steady-state conditions. The target air supply pressure refers to the stable pressure of compressed air under steady-state conditions. The first target opening degree refers to the stable opening degree of the outlet regulating valve under steady-state conditions. When the spray system reaches a steady state, these key parameters can be collected as input to the target prediction model, ensuring that the parameters input to the model are representative, rather than fluctuating values. Furthermore, it can comprehensively capture the current coordinated state of the spray system's water circuit, air circuit, and power source, providing a comprehensive data foundation for model prediction.

[0061] Finally, the target water supply pressure, target pump speed, target air supply pressure, and the first target opening degree are input into the target prediction model to obtain the predicted opening ratio. Before input, preprocessing operations such as normalization can still be performed on the target water supply pressure, target pump speed, target air supply pressure, and the first target opening degree. Predicting the adjustment amount of the outlet regulating valve opening based on the actual parameters of the current test conditions can reduce the problem of insufficient adaptability of general formulas and pre-calculate the second target opening degree, i.e., the target opening degree of the outlet regulating valve after the solenoid valve opens the nozzle, laying the foundation for synchronous adjustment at the moment the solenoid valve opens.

[0062] As an example, to bring the spray system to a steady state, the initial pump speed matching the set water supply pressure can be determined from a preset first mapping table. This first mapping table can include the mapping relationship between pump speed and water supply pressure. By directly querying the initial pump speed corresponding to the set water supply pressure, the starting point of the spray system's pump speed is determined, skipping the speed adjustment process from zero, allowing the water supply pressure to quickly approach the target value, thus achieving coarse adjustment of the water supply pressure. This reduces energy consumption and mechanical wear caused by frequent pump speed changes.

[0063] Then, the opening of the outlet regulating valve is adjusted using a closed-loop control method to ensure that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value. The set deviation value refers to the allowable error range between the actual water supply pressure and the set water supply pressure. The closed-loop control method refers to correcting the output opening adjustment amount based on the deviation between the actual water supply pressure and the set water supply pressure until the deviation of the water supply pressure is less than the set deviation value.

[0064] However, the outlet control valve has a limit value for its opening, for example, fully open (≥95%) or fully closed (≤5%). If the actual opening of the outlet control valve reaches the limit value and the deviation is not less than the set deviation value, the initial pump speed is corrected until the deviation value is less than the set deviation value. Depending on the system sensitivity, the set pump speed can be corrected each time, for example, by 50 rpm, and then fine-tuned again through the outlet control valve.

[0065] Finally, quantitative indicators of the fluctuation amplitude and duration of multiple parameters are used to determine whether the system has reached a steady state. For example, when the fluctuation amplitude between the actual water supply pressure and the set water supply pressure is less than a first set value, the change in the opening of the outlet regulating valve is less than a second set value, and the fluctuation amplitude of the actual water pump speed is less than a third set value, and the duration exceeds a set time, the spray system is determined to be in a steady state. The first set value, the second set value, and the set time are all thresholds used to determine whether the system has reached a steady state. When all the above conditions are met simultaneously, the spray system can be determined to have reached a steady state, thereby reducing misjudgments caused by the achievement of a single parameter. The quantitative indicators and duration requirements allow the steady-state definition to be reproduced and verified, and also ensure consistency in operation among different test personnel.

[0066] By combining the strategy of looking up the initial speed in the mapping table with closed-loop fine-tuning, the time for the spray system to go from start-up to steady state is significantly shortened compared to the traditional closed-loop regulation alone, making it particularly suitable for scenarios with rapid switching between multiple operating conditions in icing wind tunnel tests.

[0067] In step 304, based on the switching status signal of the solenoid valve, the opening of the outlet regulating valve can be adjusted through synchronous regulation and closed-loop correction control methods, or through pre-adjustment boosting, open-loop transition, and closed-loop stabilization control methods. Examples of both methods are given below.

[0068] Taking the synchronous adjustment and closed-loop correction control method as an example, firstly, the switch status signal is monitored in real time. In the same control cycle when the switch status signal changes from the closed state to the open state, the outlet regulating valve is controlled to adjust to the second target opening degree. Then, the opening degree of the outlet regulating valve is continuously adjusted using the closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0069] Specifically, the controller can continuously monitor the switching signals of the solenoid valve through a digital access interface. For example, a sampling frequency greater than or equal to 100Hz can be set to ensure the capture of millisecond-level state transitions. The threshold for determining a state transition can be a DC signal transitioning from 0 to 1 for a duration greater than or equal to a set time, such as 10ms, which is considered a valid opening action. The same control cycle refers to the smallest control unit of the spray system, ensuring that state recognition and adjustment commands are completed within the same cycle. For example, in the first control cycle after confirming the switch transition, the controller immediately sends an opening command to the outlet regulating valve, switching the outlet regulating valve from the first target opening to the second target opening. The outlet regulating valve can use a high-speed electric actuator to ensure that the opening switch is completed within a single control cycle. Then, a closed-loop control method, such as Proportional Integral Derivative (PID) control, is continuously used. Every certain time interval, such as 50ms, the deviation value is calculated and the adjustment amount is output, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value. The set deviation value is the allowable deviation range. By synchronizing the solenoid valve's action with its opening adjustment, the pressure overshoot caused by the solenoid valve in traditional control is reduced, for example, from greater than 10% to less than 5%. This solves the oscillation problem of the spray system caused by switching action, meeting the requirements of short-duration icing wind tunnel tests. Rapid response is achieved through feedforward regulation. After coarse adjustment of the water supply pressure through synchronous regulation, high precision is achieved through closed-loop control, balancing response speed and control accuracy. Continuous closed-loop regulation can compensate for errors in synchronous regulation and external disturbances, ensuring the stability of the spray system. Furthermore, no pre-stored pressure compensation is required, making it applicable to different nozzle opening ratios and different water supply pressures, suitable for test scenarios with frequent switching between multiple operating conditions.

[0070] Taking the pre-adjusted pressure boosting, open-loop transition, and closed-loop stabilization control method as an example, before the solenoid valve opens the nozzle, the opening degree of the outlet regulating valve is pre-set to the second target opening degree. An open-loop control method is used to ensure the actual water supply pressure is greater than the set water supply pressure. In response to the switch state signal changing from closed to open, a closed-loop control method is used to adjust the opening degree of the outlet regulating valve, ensuring that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0071] Specifically, before the solenoid valve opens the nozzle, the outlet regulating valve is pre-set to the second target opening degree. This second target opening degree is maintained through open-loop control, allowing a reserve pressure higher than the set supply water pressure to be established in the pipeline. The pre-increase is dynamically adjusted according to the nozzle opening ratio; the higher the opening ratio, the larger the increment. When the solenoid valve's on / off state signal changes from closed to open, the reserve pressure drops rapidly due to the release of nozzle flow. Utilizing the dynamic balance between the pre-increase pressure and flow release, the actual supply water pressure naturally decays to the set supply water pressure within a short time. Therefore, in response to the change in the on / off state signal, it automatically switches to closed-loop control. For example, using PID control with a dead zone, the adjustment amount is only output when the deviation exceeds the dead zone, reducing frequent actions. When the deviation between the actual supply water pressure and the set supply water pressure is less than the set deviation value, it indicates that the spray system is in a steady state. By pre-establishing a pressure reserve and using the natural decay of flow release instead of forced regulation, the fluctuation of the supply water pressure can be controlled within a smaller range, and the opening adjustment amount is reduced, lowering the actuator's action amplitude and frequency. The open-loop pre-adjustment stage requires no high-frequency calculations, and the closed-loop stage only activates when deviations exceed limits, significantly reducing energy consumption. This method is suitable for high flow rate demands caused by high on-ramp ratios, as the pre-pressurization can quickly compensate for flow losses and reduce sudden pressure drops. For example, it is suitable for full rake spray tests.

[0072] This application embodiment establishes a target prediction model, trains it using historical experimental data, and then uses the target prediction model to predict the opening degree of the outlet regulating valve corresponding to the water supply pressure after the spray system stabilizes. This rapid prediction of the outlet regulating valve opening improves the response speed of the spray system. Furthermore, by training the prediction model based on historical experimental data of each layer of spray rakes corresponding to each water supply branch, the influence of the height difference between spray rakes and the initial target valve opening is reduced, achieving decoupled control of multiple water supply pressures.

[0073] Figure 4 This is a schematic diagram of the structure of a controller 400 provided in an embodiment of this application. Figure 4 As shown, the controller 400 is applied to, for example Figure 1The spray system shown is for an icing wind tunnel test. The spray system may include multiple water supply branches. In each water supply branch, an outlet regulating valve and a solenoid valve are connected in series along the water flow direction. The solenoid valve is used to control the on / off flow of water to the nozzles in the water supply branch. The controller 400 may include an acquisition module 401, a training module 402, a prediction module 403, and a control module 404.

[0074] The acquisition module 401 is used to acquire historical test data of the spray system under multiple operating conditions, obtaining multiple sets of sample datasets. Each set of sample datasets includes first sample parameters of the spray system in a steady state before the solenoid valve opens the nozzle, and second sample parameters of the spray system reaching a steady state again after the solenoid valve opens. The first sample parameters include the sample opening ratio of the nozzle and the first sample opening degree of the outlet regulating valve. The second sample parameters include the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio.

[0075] The training module 402 is used to train the initial prediction network with the first sample parameters as input and the sample opening ratio of the second sample opening to the first sample opening as output, to obtain the target prediction model.

[0076] The prediction module 403 is used to obtain the set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system before the solenoid valve opens the nozzle. The steady-state parameters are input into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve, and the second target opening is calculated based on the first target opening of the outlet regulating valve in the steady-state parameters.

[0077] The control module 404 is used to control the outlet regulating valve to adjust to the second target opening degree based on the on / off status signal of the solenoid valve, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0078] In this embodiment, the first sample parameters may further include sample water supply pressure, sample air supply pressure, and sample water pump speed. The training module 402 may include a preprocessing unit, a first prediction unit, and an iteration unit.

[0079] The preprocessing unit is used to perform preprocessing operations on multiple sets of sample datasets. The preprocessing operations include outlier removal, normalization of parameters of the first sample, and sample balancing.

[0080] The first prediction unit is used to input the sample water supply pressure, sample air supply pressure, sample water pump speed, sample opening ratio, and first sample opening degree to obtain the sample prediction ratio.

[0081] The iterative unit is used to iteratively update the parameters of the initial prediction network based on the loss function of the sample prediction ratio and the sample opening ratio until the iteration stopping condition is met, thus obtaining the target prediction model.

[0082] The initial prediction network consists of an input layer, a hidden layer, and an output layer. The input layer has five neurons, each corresponding to the parameters of the first sample after normalization. The hidden layer consists of multiple layers of neurons with decreasing order. The output layer consists of one neuron, corresponding to the sample opening ratio.

[0083] In this embodiment, the training module 402 may further include a startup unit, a first adjustment unit, an acquisition unit, and a second prediction unit.

[0084] The start-up unit is used to determine the set opening ratio of the nozzle in the current test conditions and to start the spray system.

[0085] The first regulating unit is used to adjust the water pump speed and the opening of the outlet regulating valve based on the set water supply pressure, so as to keep the spray system in a steady state.

[0086] The acquisition unit is used to acquire the target water supply pressure, target water pump speed, target air supply pressure, and the first target opening degree of the outlet regulating valve of the spray system in response to the spray system being in a steady state.

[0087] The second prediction unit is used to input the target water supply pressure, target water pump speed, target air supply pressure, and the first target opening degree into the target prediction model to obtain the predicted opening degree ratio.

[0088] In this embodiment, the regulating unit is further configured to: determine an initial water pump speed matching the set water supply pressure from a preset first mapping table, the first mapping table including the mapping relationship between water pump speed and water supply pressure; adjust the opening of the outlet regulating valve through a closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value; wherein, if the actual opening of the outlet regulating valve reaches the limit value and the deviation value is not less than the set deviation value, the initial water pump speed is corrected until the deviation value is less than the set deviation value; when the fluctuation range between the actual water supply pressure and the set water supply pressure is less than the first set value, the change in the opening of the outlet regulating valve is less than the second set value, and the fluctuation range of the actual water pump speed is less than the third set value and the duration exceeds the set time, the spray system is determined to be in a steady state.

[0089] In this embodiment, the control module 404 includes a second adjustment unit and a third adjustment unit.

[0090] The second regulating unit is used to monitor the switch status signal in real time, and in the same control cycle when the switch status signal changes from the closed state to the open state, it controls the outlet regulating valve to adjust to the second target opening degree.

[0091] The third regulating unit is used to continuously adjust the opening of the outlet regulating valve using a closed-loop control method, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0092] In this embodiment, the control module 404 further includes a fourth adjustment unit and a fifth adjustment unit.

[0093] The fourth regulating unit is used to pre-set the opening degree of the outlet regulating valve to the second target opening degree before the solenoid valve opens the nozzle, and uses an open-loop control method to control the actual water supply pressure to be greater than the set water supply pressure.

[0094] The fifth regulating unit is used to respond to the switch state signal from the closed state to the open state, and adjusts the opening of the outlet regulating valve through a closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

[0095] In this embodiment, the sample dataset includes first sample parameters and second sample parameters for each water supply branch. The training module 402 may also include a training unit and a combination unit.

[0096] The training unit takes the first sample parameters of each water supply branch as input and the sample opening ratio of the second sample opening and the first sample opening of each water supply branch as output, and trains each initial prediction sub-network in the initial prediction network to obtain multiple target prediction sub-models.

[0097] The combined unit is used to obtain a target prediction model based on multiple target prediction sub-models, where each target prediction sub-model corresponds to a water supply branch.

[0098] This application also provides a computer-readable storage medium storing a program that can be loaded by a processor and executed by a control method for a spray system in any of the icing wind tunnel tests in this application.

[0099] 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.

[0100] The above examples illustrate this application only to aid understanding and are not intended to limit its scope. Those skilled in the art to which this application pertains can make various simple deductions, modifications, or substitutions based on the ideas presented.

Claims

1. A method of controlling a spray system for icing tunnel testing, characterized in that The spray system includes multiple water supply branches. In each water supply branch, an outlet regulating valve and a solenoid valve are connected in series along the water flow direction. The solenoid valve is used to control the on / off state of the water flow through the nozzles in the water supply branch. The control method includes: Historical test data of the spray system under multiple operating conditions are obtained to obtain multiple sets of sample datasets. Each set of sample datasets includes a first sample parameter of the spray system in a steady state before the solenoid valve opens the nozzle, and a second sample parameter of the spray system reaching a steady state again after the solenoid valve opens. The first sample parameter includes the sample opening ratio of the nozzle, the first sample opening degree of the outlet regulating valve, the sample water supply pressure, the sample air supply pressure, and the sample water pump speed. The second sample parameter includes the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio. Preprocessing operations are performed on multiple sets of the sample datasets, including outlier removal, normalization of the first sample parameters, and sample balancing. Input the sample water supply pressure, the sample air supply pressure, the sample water pump speed, the sample opening ratio, and the first sample opening degree to obtain the sample prediction ratio; Based on the loss function of the sample prediction ratio and the sample opening ratio of the second sample opening and the first sample opening, the parameters of the initial prediction network are iteratively updated until the iteration stopping condition is met to obtain the target prediction model. The initial prediction network includes an input layer, a hidden layer and an output layer. The input layer includes five neurons, which correspond to the first sample parameters after normalization. The hidden layer includes multiple layers of neurons with decreasing layer size. The output layer includes one neuron, which corresponds to the sample opening ratio. The set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system before the solenoid valve opens the nozzle are obtained. The steady-state parameters are input into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve. The second target opening is calculated based on the first target opening of the outlet regulating valve in the steady-state parameters. Based on the on / off status signal of the solenoid valve, the outlet regulating valve is controlled to adjust to the second target opening degree so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

2. The control method according to claim 1, characterized in that, The step of obtaining the steady-state parameters of the spray system before the solenoid valve opens the nozzle under the current test conditions, and inputting the steady-state parameters into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve, includes: Determine the set opening ratio of the nozzle in the current test condition, and start the spray system; Based on the set water supply pressure, the pump speed and the opening of the outlet regulating valve are adjusted to keep the spray system in a steady state. In response to the spray system being in a steady state, the target water supply pressure, target water pump speed, target air supply pressure, and the first target opening degree of the outlet regulating valve of the spray system are obtained. The target water supply pressure, the target water pump speed, the target air supply pressure, and the first target opening degree are input into the target prediction model to obtain the predicted opening degree ratio.

3. The control method according to claim 2, characterized in that, The step of adjusting the water pump speed and the opening of the outlet regulating valve based on the set water supply pressure to keep the spray system in a steady state includes: Based on the set water supply pressure, an initial water pump speed matching the set water supply pressure is determined from a preset first mapping table, wherein the first mapping table includes the mapping relationship between water pump speed and water supply pressure. The opening of the outlet regulating valve is adjusted by a closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value. If the actual opening of the outlet regulating valve reaches the limit value and the deviation value is not less than the set deviation value, the initial water pump speed is corrected until the deviation value is less than the set deviation value. When the fluctuation range between the actual water supply pressure and the set water supply pressure is less than the first set value, the change in the opening of the outlet regulating valve is less than the second set value, and the fluctuation range of the actual water pump speed is less than the third set value and the duration exceeds the set time, the spray system is determined to be in a steady state.

4. The control method according to claim 1, characterized in that, The method of controlling the outlet regulating valve to adjust to the second target opening degree based on the on / off status signal of the solenoid valve, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value, further includes: The switch status signal is monitored in real time, and within the same control cycle when the switch status signal changes from closed to open, the outlet regulating valve is controlled to adjust to the second target opening degree. The opening of the outlet regulating valve is continuously adjusted using a closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

5. The control method according to claim 1, characterized in that, The method of controlling the outlet regulating valve to adjust to the second target opening degree based on the on / off status signal of the solenoid valve, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value, further includes: Before the solenoid valve opens the nozzle, the opening degree of the outlet regulating valve is set to the second target opening degree in advance, and the actual water supply pressure is controlled to be greater than the set water supply pressure using an open-loop control method. In response to the switch status signal changing from closed to open, the opening of the outlet regulating valve is adjusted by a closed-loop control method so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

6. A controller, characterized in that, A spray system for use in icing wind tunnel testing, the spray system comprising multiple water supply branches, wherein in each water supply branch, an outlet regulating valve and a solenoid valve are connected in series along the water flow direction, the solenoid valve being used to control the on / off state of the water flow through the nozzles in the water supply branch, and the controller comprising: The acquisition module is used to acquire historical test data of the spray system under multiple operating conditions to obtain multiple sets of sample datasets. Each set of sample datasets includes a first sample parameter of the spray system in a steady state before the solenoid valve opens the nozzle, and a second sample parameter of the spray system reaching a steady state again after the solenoid valve opens. The first sample parameter includes the sample opening ratio of the nozzle, the first sample opening degree of the outlet regulating valve, the sample water supply pressure, the sample air supply pressure, and the sample water pump speed. The second sample parameter includes the second sample opening degree of the outlet regulating valve after the solenoid valve opens the nozzle according to the sample opening ratio. The training module is used to preprocess multiple sets of sample datasets. The preprocessing operations include outlier removal, normalization of the first sample parameters, and sample balancing. The module takes into account the sample water supply pressure, sample air supply pressure, sample water pump speed, sample opening ratio, and the first sample opening degree as inputs to obtain a sample prediction ratio. Based on the loss function of the sample prediction ratio and the sample opening degree ratio of the second and first sample opening degrees, the parameters of the initial prediction network are iteratively updated until the iteration stopping condition is met to obtain the target prediction model. The initial prediction network includes an input layer, a hidden layer, and an output layer. The input layer includes five neurons, each corresponding to the normalized first sample parameters. The hidden layer includes multiple layers of neurons with progressively decreasing numbers. The output layer includes one neuron, corresponding to the sample opening degree ratio. The prediction module is used to obtain the set water supply pressure of the spray system under the current test conditions and the steady-state parameters of the spray system in a steady state before the solenoid valve opens the nozzle. The steady-state parameters are input into the target prediction model to obtain the predicted opening ratio of the outlet regulating valve, and the second target opening is calculated based on the first target opening of the outlet regulating valve in the steady-state parameters. The control module is used to control the outlet regulating valve to adjust to the second target opening degree based on the on / off status signal of the solenoid valve, so that the deviation between the actual water supply pressure and the set water supply pressure is less than the set deviation value.

7. A spray system for icing wind tunnel testing, characterized in that, include: A controller according to claim 6; Multiple water supply branches, each of which includes an outlet regulating valve, a solenoid valve, and a nozzle; The outlet regulating valve and the solenoid valve are connected in series along the water flow direction. The outlet regulating valve and the solenoid valve communicate with the controller respectively. The solenoid valve is used to control the water flow of the nozzle in the water supply branch.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that can be loaded by a processor and executed as a control method for a spray system for an icing wind tunnel test as described in any one of claims 1 to 6.

Images