Active control method and system for surge of aero-engine

Abstract

The invention discloses an aero-engine surge active control method and system, and relates to the technical field of aero-engine control. The engine state and the environment state are monitored in real time, and according to the engine state and the environment state, control parameters are output through a surge prediction model to conduct active control to reduce the surge risk. Through the two stages of pre-generation and self-adaptive control, whether surge occurs can be predicted according to the engine state and the environment state, when the surge is predicted to occur, the pre-control parameters are output through the surge prediction model, and whether surge exists can be directly judged according to the engine state and the environment state. When surge exists, the control parameters are output through the surge prediction model, active control is carried out through the prepared control parameters and the control parameters, the surge risk can be reduced in advance or in real time, the service life of an engine is prolonged, and normal operation of the aircraft is prevented from being affected.

Description

Technical Field

[0001] The present invention relates to the technical field of aero-engine control, and particularly to an active control method and system for aero-engine surge. Background Technique

[0002] Aero-engine surge refers to the unstable fluctuation of air flow between the compressor and the combustion chamber during the operation of the engine, which leads to a decline in engine performance and may even cause engine damage. This phenomenon usually occurs in the operation of high-pressure-ratio compressors. By actively controlling aero-engine surge, not only can the safety and reliability of the engine be improved, but also the overall performance can be enhanced, the maintenance cost can be reduced, thereby providing a safer and more efficient flight experience for airlines and passengers.

[0003] An active control system for aero-engine surge based on backstepping sliding mode control with the patent publication number of CN115981160A includes establishing a compressor model with an actuator, coordinate transformation of the compressor model, and designing a backstepping sliding mode controller. Based on the principle of backstepping sliding mode control, the present invention can effectively simplify the design steps of the backstepping controller, expand the applicable range of the system, enhance the robustness of the system, and achieve active control of surge with large range, self-adaptation, and performance optimization. The designed backstepping sliding mode controller adds sliding mode control on the basis of backstepping control, which not only simplifies the design of the backstepping controller, but also can expand the effective working range of the controller and enhance the robustness of the system to unmatched uncertainties; the designed controller can be applied to the active control of surge caused by various incentives, improve the self-adaptability of the controller, and be closer to the actual working conditions of the engine; it can be slightly modified on the basis of the existing controller to achieve the required control effect, and compared with the existing intelligent control methods, the present invention is simpler and easier to implement.

[0004] The above-mentioned active control process and similar principles have limitations in predicting and real-time monitoring of surge. When performing reconnaissance missions at high altitudes, the flight training and operation processes need to frequently change flight altitude and speed. In this environment, due to changes in air pressure and temperature, it may pose challenges to the operating stability of aero-engines, especially prone to the occurrence of surge. Based on the above scenarios in the case of drastic environmental changes, the above methods have the drawback of being unable to quickly adapt to new aerodynamic characteristics, that is, there are limitations in predicting and real-time monitoring of surge, which is likely to increase the risk of surge. When a surge phenomenon occurs in an aero-engine, there are also individual factors that contribute to the surge phenomenon. If the factors contributing to the surge phenomenon are not monitored and processed in real time, it will lead to the growth of surge, thereby reducing the service life of the engine, and in severe cases, it may affect the normal operation of the aircraft. Therefore, the present invention is proposed. Summary of the Invention

[0005] The object of the present invention is to provide a method and system for active control of aeroengine surge to solve the problems raised in the above background art.

[0006] To achieve the above object, the present invention provides the following technical solution: A method for active control of aeroengine surge, the method comprising: Active control: Real-time monitor the engine state and the environmental state, and output control parameters through a surge prediction model according to the engine state in cooperation with the environmental state for active control to reduce the surge risk; Information collection: Obtain past flight information, extract the information with surge conditions in the past flight information to obtain relevant information, and classify the relevant information according to the flight state through a classification method to obtain target information and the corresponding target category of the target information; Information processing: Judge the factors causing the surge situation in the target information through a deep association method to obtain target factors and the surge situation, integrate the target factors and the surge situation to obtain reference information, and establish a reference library to store the reference information with the target category as the name; Preliminary generation: Preset a similarity value, judge whether there is information in the engine state and the environmental state that exceeds the similarity value with the target factors based on the reference library through a judgment method to obtain a judgment result, and generate preliminary control parameters through the surge prediction model based on the judgment result for active control to reduce the surge risk; Adaptive control: Establish a surge prediction model, and import the engine state and the environmental state into the surge prediction model to obtain control parameters; Monitoring feedback: Obtain feedback address information, obtain and monitor the components that will contribute to the surge through a monitoring method to obtain a monitoring result, and when the monitoring result feedback is that the component is abnormal, find a solution to the monitoring result through an abnormal handling method to obtain result information; Feedback output: Based on the feedback address information, real-time feedback the monitoring result and the result information to the manager; The deep association method includes: The past flight information includes the past engine state and the past environmental state. Intercept the information before the surge occurs in the target information to obtain processed information, intercept the information after the surge occurs in the target information to obtain comparison information, judge the difference items in the processed information and the comparison information through an exploration method to obtain target factors, and extract the surge that occurs in the target information to obtain the surge situation.

[0007] Furthermore, the exploration method includes: presetting influencing factors, obtaining difference items by comparing the difference items in the processed information and the comparison information, judging the relationship between the difference items and the influencing factors, and when the influencing factors include the difference items, extracting the change information of the difference items before the occurrence of surge in the target information to obtain pre-factors, presetting a similarity threshold and the number of expressions, recording the number of pre-factors with similarity exceeding the similarity threshold under the same difference item to obtain the number of factors, and when the number of factors exceeds the number of expressions, integrating the pre-factors with similarity exceeding the similarity threshold under the same difference item to obtain the target factor.

[0008] Furthermore, the classification method includes: presetting flight state categories to obtain target categories, where the target categories include takeoff state, altitude increase state, altitude decrease state, and cruise state, extracting the flight state in which surge occurs in the relevant information to obtain the target state, classifying the target state based on the target categories to obtain classification information, extracting the relevant information corresponding to the classification information to obtain the target information, and establishing an association relationship between the target information and the target categories based on the classification information.

[0009] Furthermore, the judgment method includes: obtaining the real-time flight state to obtain the target state, traversing the reference library based on the target state to find the target category corresponding to the target state to obtain the specific category, extracting the reference information corresponding to the specific category in the reference library to obtain the specific information, monitoring and recording the engine state and the environmental state in real time to obtain the real-time information, comparing the similarity between the specific information and the real-time information based on the reference library to obtain the judgment result, and when the judgment result feedbacks that the similarity of the information in the specific information and the real-time information exceeds the similarity value, generating a preliminary control parameter based on the real-time information through the surge prediction model.

[0010] Furthermore, the method for establishing a surge prediction model includes: Data acquisition and processing: obtaining the parameter catalog, obtaining the data of successfully eliminating surge in the past flight information to obtain the first target data set, marking the data process of eliminating surge in the first target data set to obtain the first training data set, extracting the relevant information to obtain the second target data set, and marking whether surge occurs in the second target data set to obtain the second training data set; Model selection and training: selecting a suitable deep learning model as the model matrix, and importing the first training data and the second training data into the model matrix for the model to be trained to obtain the initial model; Model adjustment and output: obtaining the verification data and the verification information, obtaining the result information by importing the verification data into the initial model, where the result information includes the existence of surge and the control parameter or the non-existence of surge, comparing the result information with the verification information to obtain the comparison result, and adjusting the initial model according to the comparison result to obtain the surge prediction model.

[0011] Furthermore, the monitoring method includes: obtaining the factors that promote surge to get the promotion items, determining the corresponding components based on the promotion items to obtain the target components, obtaining the information under the corresponding operating conditions of the target components to get the standard information, monitoring and recording the operating conditions of the components to get the real-time component information, comparing whether the standard information is consistent with the real-time component information to obtain the monitoring result, and when the standard information is inconsistent with the real-time component information, the monitoring result feedback is that the component is abnormal.

[0012] Furthermore, the anomaly handling method includes: obtaining the adjustable parameters of the promotion items based on the promotion items to get the parameter information, obtaining the difference between the standard information and the real-time component information to get the difference information, adjusting the parameters based on the parameter information and recording the process of eliminating the difference information to get the process information, obtaining the method for eliminating component anomalies based on the monitoring result to get the target method, and integrating the process information and the target method to get the result information.

[0013] An active control system for aero-engine surge uses the above-mentioned active control method for aero-engine surge.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: The active control method and system for aero-engine surge can predict whether surge will occur according to the engine state and environmental state through the two stages of preliminary generation and adaptive control set. When it is predicted that surge will occur, preliminary control parameters are output through the surge prediction model, and it can also directly judge whether there is surge according to the engine state and environmental state. When there is surge, control parameters are output through the surge prediction model. Active control is carried out through the preliminary control parameters and the control parameters, so as to reduce the surge risk in advance or immediately, extend the service life of the engine, and avoid affecting the normal operation of the aircraft.

[0015] At the same time, through the classification method set, when predicting surge subsequently, the target category can be matched according to the real-time flight state, and the corresponding information can be selected for matching to complete the surge prediction, which is beneficial to reducing the system occupancy. Through the judgment method set, the similarity can be matched according to the real-time monitored engine state and environmental state. Whether surge will occur is predicted according to the similarity. When surge will occur, preliminary control parameters are generated to facilitate reducing the surge risk through active control. By recording the number of pre-factors whose similarity exceeds the similarity threshold and comparing the expression times, when the expression times are exceeded, the pre-factors are integrated to obtain the target factor. Through the setting of the similarity threshold and the expression times, it is convenient to improve the accuracy of target factor collection, reduce the probability of special situations occurring, and facilitate subsequent prediction of whether surge will occur.

[0016] Meanwhile, the final surge prediction model can be integrated into the engine control system to capture the engine operating state and environmental information in real time for judgment, realizing real-time feedback and adjustment. According to the information output by the model, the engine operating parameters are adjusted in cooperation with active control to reduce the surge risk. Through the set exception handling method, the process information and target methods for facilitating the manager to solve the factors contributing to surge can be generated, and the process information and target methods are integrated and fed back to the manager to facilitate the manager to solve the factors contributing to surge and reduce the surge risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the main process structure of the present invention; Figure 2 It is a schematic diagram of the auxiliary process structure of the present invention; Figure 3 It is a schematic diagram of the structure for establishing a surge prediction model of the present invention; Figure 4 It is a schematic diagram of the operation structure of the surge prediction model of the present invention; Figure 5 It is a schematic diagram of the reference library structure of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] The surge phenomenon may cause serious safety hazards during the operation of the engine. If not controlled, the engine may fail during flight, affecting flight safety. Active control can prevent or mitigate these risks and provide reliable guarantee for flight. Surge will cause a decrease in engine thrust and fuel efficiency. Through active control, the performance of the engine under different flight conditions can be optimized, the overall efficiency and economy can be improved, and thus the service life of the engine can be extended.

[0020] As Figures 1-5 shown, the present invention provides a technical solution: an active control method for surge of an aeroengine, the method comprising: Active control: real-time monitoring of the engine state and environmental state, and active control to reduce the surge risk by outputting control parameters through a surge prediction model according to the engine state in cooperation with the environmental state; Information collection: Obtain past flight information, extract information on engine surges from the past flight information to obtain relevant information, and classify the relevant information according to flight status through a classification method to obtain target information and the corresponding target categories; Information processing: Determine the factors causing engine surges in the target information through a deep association method to obtain target factors and surge conditions, integrate the target factors and surge conditions to obtain reference information, and establish a reference library to store the reference information with the target category as the name; Preliminary generation: Preset a similarity value, based on the reference library, determine whether there is information in the engine state and environmental state whose similarity to the target factor exceeds the similarity value through a judgment method to obtain a judgment result, and generate preliminary control parameters through a surge prediction model based on the judgment result for active control to reduce the surge risk; Adaptive control: Establish a surge prediction model, and import the engine state and environmental state into the surge prediction model to obtain control parameters; Monitoring feedback: Obtain feedback address information, obtain and monitor components that can contribute to surges through a monitoring method to obtain a monitoring result. When the monitoring result indicates that a component is abnormal, find a solution to handle the abnormal component in the monitoring result through an abnormal handling method to obtain result information; Feedback output: Based on the feedback address information, real-time feedback the monitoring result and result information to the manager, and the manager eliminates the factors contributing to surges according to the result information; It should be noted that: In the active control stage, the engine states include rotational speed, intake air temperature, oil temperature, pressure, etc., and the environmental states include air temperature, humidity, altitude, wind speed, etc. The surge prediction model is set to judge whether surge occurs based on the engine states in combination with the environmental states and output control parameters when surge occurs. Active control is carried out through the control parameters to reduce the surge risk. In the information collection stage, the past flight information can be obtained by collecting the past flight logs of relevant aircraft. The relevant information is the information indicating the occurrence of engine surge in the flight information. The relevant information is classified according to the flight states of the aircraft through the set classification method to facilitate subsequent comparison corresponding to different flight states, so as to reduce the difficulty of information processing and the system occupancy rate. In the information processing stage, the factors causing the engine to surge in the target information are judged through the deep correlation method to obtain the target factors and the surge conditions corresponding to the target factors. The target factors and the surge conditions are the reference information, and the reference information is stored with the target category as the name to facilitate subsequent retrieval of the reference information. In the preliminary generation stage, the similarity value is preset according to the actual usage. Usually, the similarity value can be set to 80%. It is judged whether there is information in the engine states and environmental states whose similarity to the target factors exceeds the similarity value through the judgment method to obtain the judgment result. In the preliminary generation stage, the engine states and environmental states are paragraph information. The preliminary control parameters are generated through the surge prediction model according to the feedback of the judgment result. Active control is carried out through the preliminary control parameters to reduce the surge risk. In the adaptive control stage, the function of the surge prediction model is to import the engine states and environmental states into the surge prediction model to obtain the control parameters for reducing the surge risk. Whether surge occurs under the real-time engine states and environmental states can be judged through both the preliminary generation and adaptive control stages. Moreover, in the preliminary generation stage, it is also possible to predict whether surge will occur subsequently according to the fluctuations of the real-time engine states and environmental states, and then through the surge prediction model, output the control parameters for reducing the surge risk, which is conducive to reducing the surge risk and prolonging the service life of the engine. In the monitoring and feedback stage, the feedback address information is the address information for feedbacking the information to the manager. The feedback address information includes but is not limited to telephone numbers and email addresses, etc. The components and factors that will contribute to surge are obtained and monitored through the set monitoring method to obtain the monitoring result. Then, according to the feedback of the monitoring result, the solution method for finding and processing the abnormal components in the monitoring result is obtained through the set abnormal handling method to obtain the result information. In the feedback output stage, the monitoring result and the result information are feedbacked to the manager according to the feedback address information, and then the manager's feedback is listened to. The factors contributing to surge are solved according to the manager's feedback. Through the preliminary generation and adaptive control stages, it is possible to predict whether surge will occur according to the engine states and environmental states. When it is predicted that surge will occur, the preliminary control parameters are output through the surge prediction model. Moreover, it is also possible to directly judge whether there is surge according to the engine states and environmental states. When there is surge, the control parameters are output through the surge prediction model.Active control is performed using preliminary control parameters and control parameters, which can reduce the surge risk in advance or immediately, extend the service life of the engine, and avoid affecting the normal operation of the aircraft.

[0021] As Figure 1 shown, the in-depth correlation method includes: the past flight information includes the past engine state and the past environmental state, the target information includes part of the past engine state and part of the past environmental state, the information before surge occurrence in the target information is intercepted to obtain the processed information, the information after surge occurrence in the target information is intercepted to obtain the comparison information, the target factors are obtained by exploring the difference items in the processed information and the comparison information, the surges that occurred in the target information are extracted to obtain the surge situation, and the correlation between the target factors and the surge situation is established.

[0022] It should be noted that: the past flight information includes the engine state of the past aircraft and the environmental state where the past aircraft is located, the target information is paragraph information, that is, part of the past engine state and part of the past environmental state, the processed information is the information before surge occurrence in the target information, the comparison information is the information after surge occurrence in the target information, the target factors are obtained by setting an exploration method based on the processed information and the comparison information and judging the difference items between the two, the surge situation is the surge that occurred in the target information, which can be specifically expressed as the specific magnitude of the surge. By establishing the correlation between the target factors and the surge situation, it is convenient for subsequent search and retrieval.

[0023] The exploration method includes: presetting influencing factors, comparing the difference items in the processed information and the comparison information to obtain the difference items, judging the relationship between the difference items and the influencing factors. When the influencing factors contain the difference items, the change information of the difference items before surge occurrence is extracted from the target information to obtain the preliminary factors, presetting a similarity threshold and an expression count, recording the number of preliminary factors with a similarity exceeding the similarity threshold under the same difference item to obtain the factor count. When the factor count exceeds the expression count, the preliminary factors with a similarity exceeding the similarity threshold under the same difference item are integrated to obtain the target factors.

[0024] It should be noted that: the influencing factors are preset according to the actual usage situation, specifically the main reasons for surge occurrence. The difference items are the items with differences in the processed information and the comparison information. The similarity threshold and the expression count are preset according to the actual usage situation. The similarity threshold is the similarity threshold, and the expression count is the number of occurrences of the same preliminary factor. By recording the number of preliminary factors with a similarity exceeding the similarity threshold and comparing the expression count, when it exceeds the expression count, the preliminary factors are integrated to obtain the target factors. By setting the similarity threshold and the expression count, it is convenient to improve the accuracy of target factor collection, reduce the probability of special situations occurring, and facilitate subsequent prediction of whether a surge will occur.

[0025] As Figure 1 and Figure 5As shown, the classification method includes: presetting flight state categories to obtain target categories, where the target categories include takeoff state, height increase state, height decrease state, and cruise state; extracting the flight state in the relevant information where surge occurs to obtain the target state; classifying the target state based on the target categories to obtain classification information; extracting relevant information corresponding to the classification information to obtain target information; and establishing an association relationship between the target information and the target categories based on the classification information.

[0026] It should be noted that: the flight state categories are set according to actual usage requirements. Extract the flight state in the relevant information where surge occurs to obtain the target state, classify the target state according to the target categories to obtain classification information. Through the set classification method, when predicting surge subsequently, the target category can be matched according to the real-time flight state and the corresponding information can be selected for matching to complete the surge prediction, which is beneficial to reducing the system occupancy. Extract relevant information corresponding to the target state in the classification information to obtain target information. By establishing an association relationship between the target information and the target categories, it is convenient to retrieve and find the other information according to one of the information subsequently. For example Figure 4 As shown, target category 1, target category 2, target category 3, and target category n all represent an actual target category, which can specifically be the takeoff state, height increase state, height decrease state, and cruise state.

[0027] For example Figure 1 As shown, the judgment method includes: obtaining the real-time flight state to obtain the target state; traversing the reference library based on the target state to find the target category corresponding to the target state to obtain the specific category; extracting the reference information corresponding to the specific category in the reference library to obtain the specific information; monitoring and recording the engine state and environmental state in real time to obtain the real-time information; comparing the similarity between the specific information and the real-time information based on the reference library to obtain the judgment result; when the judgment result feedback indicates that the similarity of the information in the specific information and the real-time information exceeds the similarity value, generating a preliminary control parameter based on the real-time information through a surge prediction model.

[0028] It should be noted that: the target state is the real-time flight state, such as the takeoff state, height increase state, height decrease state, and cruise state. Find the target category corresponding to the target state in the reference library according to the target state to obtain the specific category, extract the reference information corresponding to the specific category in the reference library to obtain the specific information, compare the similarity between the specific information and the real-time information to obtain the judgment result, and decide whether to generate a preliminary control parameter according to the judgment result. Through the set judgment method, the similarity can be matched according to the engine state and environmental state monitored and recorded in real time, and whether surge will occur can be predicted according to the similarity magnitude. When surge will occur, a preliminary control parameter is generated to facilitate reducing the surge risk through active control.

[0029] For exampleFigure 3 and Figure 4 As shown, the method for establishing a surge prediction model includes: Data acquisition and processing: Obtain a parameter catalog, obtain the data of successfully eliminating surge in past flight information to obtain a first target data set, mark the process of eliminating surge data in the first target data set to obtain a first training data set, extract relevant information to obtain a second target data set, and mark whether surge occurs in the second target data set to obtain a second training data set; Model selection and training: Select a suitable deep learning model as the model matrix, and import the first training data and the second training data into the model matrix for the model to be trained to obtain an initial model; Model adjustment and output: Obtain verification data and verification information, obtain result information by importing the verification data into the initial model. The result information includes the existence of surge and control parameters or the non-existence of surge. Compare the result information with the verification information to obtain a comparison result, and adjust the initial model according to the comparison result to obtain a surge prediction model.

[0030] It should be noted that: In the data acquisition and processing stage, in cases of successfully eliminating surge, record in detail the relevant parameter changes and control measures to form a first target data set. These data will show the operations and state changes related to eliminating surge. Mark each sample in the extracted second target data set as whether surge occurs, and use the knowledge of domain experts or historical data for judgment to form a second training data set. In the model selection and training stage, according to the characteristics and complexity of the data, select a suitable deep learning framework. Commonly used models include: Convolutional Neural Network, which is suitable for processing time series data and feature extraction; Recurrent Neural Network or Long Short-Term Memory Network, which is suitable for processing sequence data and can capture long-range dependencies. Import the first training data set and the second training data set into the selected deep learning model for training to obtain an initial model. During the training process, appropriate optimization algorithms and loss functions can also be used for training to improve the model training efficiency. In the model adjustment and output stage, independently select a part of the data that has not participated in the training as verification data to ensure that these data represent different working conditions and environments and enhance the robustness of the model. Input the verification data into the initial model to obtain real-time result information, including whether each sample has surge and the corresponding control parameters (such as engine throttle, intake adjustment amount, etc.). Compare the result information generated by the model with the verification information to evaluate the accuracy and reliability. After training is completed, continue to update the model with real-time data to ensure its adaptability and accuracy to continuously optimize the judgment ability of the model. Moreover, the final surge prediction model can be integrated into the engine control system to capture the engine working state and environmental information in real time for judgment, realize real-time feedback and adjustment, and cooperate with active control to adjust the engine working parameters according to the information output by the model, thereby reducing the surge risk.

[0031] As Figure 2 shown, the monitoring method includes: obtaining the factors that promote surge to get the promotion term, determining the corresponding component based on the promotion term to get the target component, obtaining the information under the corresponding operating conditions of the target component to get the standard information, monitoring and recording the operating conditions of the component to get the real-time component information, comparing whether the standard information is consistent with the real-time component information to get the monitoring result, and when the standard information is inconsistent with the real-time component information, the monitoring result feedback is that the component is abnormal.

[0032] It should be noted that: the process of obtaining the factors that promote surge can be directly obtained by retrieving on the search platform or by establishing a model for deduction. Usually, the factors that promote surge include insufficient fuel quantity, unstable supply or poor fuel quality, all of which may cause uneven gas mixing and promote surge. By determining the corresponding component through the promotion term and obtaining the standard information by obtaining the information under the corresponding operating conditions of the target component, that is, obtaining the normal operating information of the target component under the current conditions to get the standard information, and obtaining the monitoring result by comparing the real-time component information and the standard information through the monitoring method.

[0033] As Figure 2 shown, the abnormal handling method includes: obtaining the adjustable parameters of the promotion term based on the promotion term to get the parameter information, obtaining the difference between the standard information and the real-time component information to get the difference information, performing parameter adjustment based on the parameter information and recording the process of eliminating the difference information to get the process information, obtaining the method of eliminating component abnormality based on the monitoring result to get the target method, and integrating the process information and the target method to get the result information.

[0034] It should be noted that: obtaining the adjustable parameters of the promotion term, for example, obtaining the controllable parameter information of the fuel component, the difference information is the difference between the standard information and the real-time component information, and the process of performing parameter adjustment based on the parameter information and recording the process of eliminating the difference information to get the process information, for example, directly connecting the auxiliary fuel tank in real time to supplement fuel to ensure sufficient fuel supply, and the process of obtaining the method of eliminating component abnormality based on the monitoring result to get the target method, obtaining the target method by directly retrieving the method of eliminating component abnormality, integrating the process information and the target method to get the result information, and by feeding back the result information to the manager, it is convenient for the manager to solve the promoted surge to reduce the surge risk.

[0035] Embodiment: When the aircraft is performing a reconnaissance mission at high altitude, the active control method for aero-engine surge provided by the present invention is used. During the flight training and operation of the aircraft, it is necessary to frequently change the flight altitude and speed. Due to the changes in air pressure and temperature at different altitudes, it may pose a challenge to the operating stability of the aero-engine, especially prone to the occurrence of surge. Therefore, it is necessary to set an active control method for surge to reduce the surge risk to ensure the normal operation of the aircraft.

[0036] In the specific implementation process: it is necessary to obtain the historical flight information of the aircraft and extract important information from the historical flight information. The important information is relevant information. The relevant information specifically includes whether surge occurs, the situation of surge, and the engine status and environmental status before and after the surge occurs. For example, a flight history record is obtained from the flight data recorder: Flight date: February 10, 2025, Aircraft model: Boeing 737, Flight altitude: 10,000 meters, Flight speed: 850 kilometers per hour, Engine speed: 70% N1, Fuel flow: 2400 kg / hour, Intake temperature: -30°C, External air pressure: 260hPa, Meteorological conditions: Slight air turbulence; Extracted important information: Whether surge occurs: Yes, Occurrence time: When the flight stage is at cruising altitude, about 90 minutes of flight time, Surge situation: Typical high-frequency vibration, Duration: about 15 seconds, Unbalanced engine intake; Engine status before and after surge occurs: Before surge occurs: Speed: 70% N1, fuel flow: 2400 kg / h, intake air temperature: -30°C, after surge occurs: speed: drops to 66% N1, fuel flow: momentarily increases to 2600 kg / h (crew adjusts to stabilize engine), intake air temperature: remains at -30°C (unchanged), environmental conditions when surge occurs: flight altitude: unchanged, still 10,000 meters, weather conditions: external air pressure may momentarily change to 250 hPa (air disturbance), wind speed and wind direction: a short crosswind of 5 knots occurs during the flight. The above historical flight information is the flight information of two time nodes. Specifically, in the actual process, the historical flight information is paragraph information; the factors causing surge are judged according to the difference in factors before and after surge, and the target factors for surge are obtained based on the cause analysis according to the wind speed and direction combined with the air pressure change. The target factors are the air disturbance caused by the sudden strong airflow, and then the relevant information is classified and stored according to the flight status at the time of occurrence. Subsequently, the engine status paragraph process and environmental status paragraph process obtained by real-time monitoring are matched with the stored relevant information for similarity to determine whether surge will occur. By extracting historical flight The process of eliminating surge and related information in the information are used as training data for training the surge prediction model. The corresponding deep learning model is selected, and the training data is imported into the deep learning model for training to obtain the surge prediction model. During the training process, appropriate optimization algorithms (such as Adam, RMSprop) and loss functions (such as binary cross entropy) can also be used for training to improve the model training efficiency. During the verification process, confusion matrix, ROC curve and other methods can be used to analyze the model performance. According to the comparison results, the shortcomings of the model (such as the existence of false positives or false negatives) are analyzed, and the model structure or training parameters are adjusted in a targeted manner. When necessary, data enhancement, regularization and other methods are added to improve the model performance to obtain the surge prediction model. During actual use, by actively monitoring the engine state and the environmental state, importing the engine state and the environmental state into a reference library for data similarity matching, and combining with the similarity value according to the matching result, it is possible to predict whether surge will occur in the engine state and the environmental state. Here, the engine state and the environmental state are paragraph information, that is, not the flight information at a single time node. When it is predicted that surge will occur, the aircraft with the engine state and the environmental state is recorded. At the same time, preliminary control parameters are generated through a surge prediction model, and active control is carried out through the preliminary control parameters to reduce the surge risk. The engine state and the environmental state are continuously monitored in real time, and whether surge occurs in the engine state and the environmental state is verified through the surge prediction model. When it is verified that surge occurs, control parameters are generated through the surge prediction model, and active control is carried out through the control parameters, so as to reduce the surge risk. At the same time, during the flight of the aircraft, the factors that contribute to surge are monitored in real time. When the factor is abnormal, a method to solve the factor is generated and fed back to the manager. The manager uses the method to solve the factor that contributes to surge to complete the handling of the factor that contributes to surge.

[0037] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended embodiments and their equivalents.

Claims

1. An active control method for aircraft engine surge, characterized in that: The method comprises: Active control: monitor the engine status and environmental status in real time, and output control parameters through the surge prediction model according to the engine status and environmental status to actively control and reduce the surge risk; Information collection: Obtain past flight information, extract information about surge conditions in past flight information to obtain relevant information, and classify the relevant information according to the flight status through classification methods to obtain target information and target categories corresponding to the target information; Information processing: The factors causing the surge in the target information are determined by the deep association method to obtain the target factors and surge conditions. The target factors and surge conditions are integrated to obtain reference information. A reference library is established to store the reference information in the name of the target category. Preparatory generation: Preset similarity values, and use judgment methods based on the reference library to determine whether there is information in the engine state and environmental state that is similar to the target factor beyond the similarity value to obtain a judgment result. Based on the judgment result, the surge prediction model generates preliminary control parameters to actively control and reduce the surge risk; Monitoring feedback: Get feedback address information, obtain and monitor the components that promote surge through monitoring methods to obtain monitoring results. When the monitoring results feedback that the components are abnormal, find solutions to the monitoring results through abnormality handling methods to obtain result information; Feedback output: Based on the feedback address information, the monitoring results and result information are fed back to the manager in real time; The deep correlation method includes: the past flight information includes the past engine status and the past environmental status, the information before the surge occurs in the target information is intercepted to obtain the processing information, the information after the surge occurs in the target information is intercepted to obtain the comparison information, the difference items in the processing information and the comparison information are judged by the exploration method to obtain the target factor, and the surge occurring in the target information is extracted to obtain the surge situation.

2. The active control method for aircraft engine surge according to claim 1, characterized in that: The research method includes: presetting influencing factors, comparing difference items in processing information and comparison information to obtain difference items, judging the relationship between difference items and influencing factors, when the influencing factors include difference items, extracting change information of difference items before surge occurs in target information to obtain predictive factors, presetting similarity threshold and expression times, recording the number of predictive factors whose similarity exceeds the similarity threshold under the same difference item to obtain the number of factors, and when the number of factors exceeds the expression times, integrating the predictive factors whose similarity exceeds the similarity threshold under the same difference item to obtain the target factor.

3. The active control method for aircraft engine surge according to claim 1, characterized in that: The classification method includes: presetting a flight state category to obtain a target category, the target category includes a take-off state, an altitude increase state, a altitude decrease state and a cruising state, extracting the flight state in which a surge occurs from relevant information to obtain a target state, classifying the target state based on the target category to obtain classification information, extracting relevant information based on the classification information to obtain target information, and establishing an association relationship between the target information and the target category based on the classification information.

4. The active control method for aircraft engine surge according to claim 1, characterized in that: The judgment method includes: obtaining a real-time flight state to obtain a target state, traversing a reference library based on the target state to find a target category corresponding to the target state to obtain a specific category, extracting reference information corresponding to the specific category in the reference library based on the specific category to obtain specific information, real-time monitoring and recording of an engine state and an environmental state to obtain real-time information, comparing the similarity between the specific information and the real-time information based on the reference library to obtain a judgment result, and when the judgment result is fed back as the similarity between the specific information and the real-time information exceeds a similarity value, generating a preliminary control parameter through a surge prediction model based on the real-time information.

5. The active control method for aircraft engine surge according to claim 1, characterized in that: Methods for establishing surge prediction models include: Data acquisition and processing: obtaining a parameter catalog, obtaining data on successful surge elimination in past flight information to obtain a first target data set, marking the data process of surge elimination in the first target data set to obtain a first training data set, extracting relevant information to obtain a second target data set, and marking whether surge occurs in the second target data set to obtain a second training data set; Model selection and training: Select a suitable deep learning model as the model matrix, import the first training data and the second training data into the model matrix for model training to obtain an initial model; Model adjustment and output: obtain verification data and verification information, and obtain result information by importing the verification data into the initial model. The result information includes the existence of surge and control parameters or the absence of surge. The result information is compared with the verification information to obtain a comparison result. According to the comparison result, the initial model is adjusted to obtain a surge prediction model.

6. The active control method for aircraft engine surge according to claim 1, characterized in that: The monitoring method includes: obtaining factors that promote surge to obtain promoting items, determining corresponding components based on the promoting items to obtain target components, obtaining information under corresponding operating conditions of the target components to obtain standard information, monitoring and recording the operating conditions of the components to obtain real-time component information, comparing whether the standard information is consistent with the real-time component information to obtain monitoring results, and when the standard information is inconsistent with the real-time component information, the monitoring results are fed back as abnormality of the components.

7. The active control method for aircraft engine surge according to claim 6, characterized in that: The exception handling method includes: obtaining adjustable parameters of the promoting items based on the promoting items to obtain parameter information, obtaining the difference between the standard information and the real-time component information to obtain difference information, adjusting the parameters based on the parameter information and recording the process of eliminating the difference information to obtain process information, obtaining a method of eliminating component abnormalities based on monitoring results to obtain a target method, and integrating the process information and the target method to obtain result information.

8. An active control system for aircraft engine surge, characterized in that: An active control method for aircraft engine surge as described in any one of claims 1 to 7 is used.

Images