Method and system for detecting health state of aircraft hydraulic system

Abstract

The invention discloses a health state detection method and system for an aircraft hydraulic system, and relates to the field of aerospace engineering, and the method comprises the steps: dividing a multi-modal data flow into an operation parameter subset and an acoustic emission subset, carrying out the preprocessing, respectively extracting features, generating feature vectors, splicing the two feature vectors into a fusion feature vector, and carrying out the recognition of the fusion feature vector; the fusion feature vector code is mapped into a quantum initial state, a quantum annealing model is established, a state characterization result is obtained, a pre-trained deep neural network is used for processing the state characterization result, a health index is calculated, and a structured diagnosis result of the aircraft hydraulic system is obtained; according to the method, a nonlinear coupling relationship among multiple variables is processed by using a quantum annealing technology, state probability distribution of the aircraft hydraulic system is generated through quantum optimization, and potential characteristics of a complex state are captured.

Description

Technical Field

[0001] The present invention relates to the field of aerospace engineering, and particularly to a method and system for detecting the health status of an aircraft hydraulic system. Background Art

[0002] The aircraft hydraulic system is an important part of modern aircraft, and it is responsible for realizing core functions such as flight control, landing gear retraction, and braking system. With the increasing progress of aviation technology, the requirements for detecting the health status of the hydraulic system are also rising, and more efficient and accurate monitoring means are expected to ensure flight safety.

[0003] In most cases, traditional machine learning or statistical methods are usually adopted when dealing with multi-modal data. However, when facing high-dimensional and non-linearly coupled multi-modal data, they often seem powerless. When dealing with strongly correlated variables, overfitting is likely to occur, and it is difficult to capture microscopic fault modes. Secondly, the resolution of traditional time-frequency analysis technology is insufficient, and it is very difficult to detect nanosecond-level transient signals, which seriously affects the reliability of the hydraulic system. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for detecting the health status of an aircraft hydraulic system to solve the problem that the resolution of traditional time-frequency analysis technology is insufficient and it is very difficult to detect nanosecond-level transient signals.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for detecting the health status of an aircraft hydraulic system, which includes:

[0008] Deploy a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream;

[0009] Divide the multi-modal data stream into two parts, an operating parameter subset and an acoustic emission subset, for preprocessing, and extract features to generate feature vectors respectively. At the same time, splice the two feature vectors into a fusion feature vector;

[0010] Encode and map the fusion feature vector into a quantum initial state, establish a quantum annealing model, and obtain a state characterization result;

[0011] Use a pre-trained deep neural network to process the state characterization result and calculate a health index to obtain a structured diagnosis result of the aircraft hydraulic system;

[0012] Present the diagnosis result in real time through an on-board display module, and record it in the on-board black box and the ground maintenance server.

[0013] As a preferred solution of the health status detection method for the aircraft hydraulic system of the present invention, the following steps are included: Deploy a multi-modal sensor array to collect real-time data of the aircraft hydraulic system, and form a structured multi-modal data stream, specifically including the following steps,

[0014] Unify the collected real-time data by sequence and timestamp, compress it and transmit it to the on-board processing unit to obtain a structured multi-modal data stream.

[0015] As a preferred solution of the health status detection method for the aircraft hydraulic system of the present invention, the following steps are included: Divide the multi-modal data stream into two parts, an operating parameter subset and an acoustic emission subset, for preprocessing, and extract features to generate feature vectors respectively. At the same time, splice the two feature vectors into a fused feature vector, specifically including the following steps,

[0016] Use a predefined mapping table to divide the multi-modal data stream into an operating parameter subset and an acoustic emission subset;

[0017] Perform compressive sensing analysis on the acoustic emission subset to obtain a reconstructed acoustic wave signal;

[0018] Apply a dual-modal triggering strategy, set a sliding window with a fixed width, and preset a change rate threshold and an offset limit;

[0019] Calculate the energy mutation rate of the reconstructed acoustic wave signal. When the energy mutation rate exceeds the preset energy threshold, record the trigger time point. Use the fast Fourier transform to convert the reconstructed acoustic wave signal into the frequency domain, detect the offset of the peak frequency, and when the offset exceeds the offset limit, record another trigger time point;

[0020] When the two trigger time points coincide, extract the features of the reconstructed acoustic wave signal centered on this time point to obtain an acoustic emission feature vector;

[0021] Calculate the mean and standard deviation of the operating parameter subset through statistical analysis to obtain an operating parameter feature vector;

[0022] Combine the acoustic emission feature vector and the operating parameter feature vector into a fused feature vector by splicing.

[0023] As a preferred solution of the health status detection method for the aircraft hydraulic system of the present invention, the following steps are included: Encode and map the fused feature vector into an initial quantum state, and establish a quantum annealing model, specifically including the following steps,

[0024] Normalize the fused feature vector and map it into an initial quantum state;

[0025] Set the energy function as the core of the quantum annealing model, and select a linear scheduling method to complete the configuration of the quantum annealing model;

[0026] As a preferred embodiment of the method for detecting the health status of the aircraft hydraulic system according to the present invention, wherein: obtaining the state characterization result specifically includes the following steps,

[0027] Obtain the target state by minimizing the target problem term, adjust the initial quantum state according to the selected linear scheduling method, transition to the target state, and use iteration to generate the probability distribution of the hydraulic system state;

[0028] Use a post-processing tool to analyze the evolution trajectory of the quantum state, calculate the weight value of each feature in the fusion feature vector, and form a weight vector;

[0029] Combine the weight vector with the probability distribution of the hydraulic system state to obtain the state characterization result.

[0030] As a preferred embodiment of the method for detecting the health status of the aircraft hydraulic system according to the present invention, wherein: use a pre-trained deep neural network to process the state characterization result and calculate the health index to obtain the structured diagnosis result of the aircraft hydraulic system, specifically including the following steps,

[0031] Establish a historical frequency spectrum database and set a fault threshold according to historical fault data;

[0032] Input the state characterization result into the pre-trained deep neural network;

[0033] Extract the probability values of each probability distribution in the state characterization result and calculate to obtain a preliminary state judgment of the aircraft hydraulic system;

[0034] Compare the sum of the probability values of the microscopic fault state and the severe fault state with the fault threshold, and obtain the state label and trigger signal after exceeding the fault threshold;

[0035] Extract the weight vector from the state characterization result, determine the features that contribute greatly to the abnormal state, and match them with the patterns in the historical frequency spectrum database to identify the faulty components and obtain the structured diagnosis result of the aircraft hydraulic system.

[0036] As a preferred embodiment of the method for detecting the health status of the aircraft hydraulic system according to the present invention, wherein: the structured diagnosis result of the aircraft hydraulic system specifically includes the following steps,

[0037] Calculate the health index according to the probability value of the normal state, and integrate the state label and the faulty components to generate a diagnosis report of the aircraft hydraulic system.

[0038] In a second aspect, the present invention provides a system for detecting the health status of an aircraft hydraulic system, including,

[0039] The acquisition module deploys a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream.

[0040] The fusion module divides the multi-modal data stream into two parts, namely the operating parameter subset and the acoustic emission subset, for preprocessing, extracts features respectively to generate feature vectors, and simultaneously concatenates the two feature vectors into a fused feature vector.

[0041] The quantum processing module encodes and maps the fused feature vector into an initial quantum state, establishes a quantum annealing model, and obtains a state representation result.

[0042] The diagnosis module processes the state representation result using a pre-trained deep neural network and calculates the health index to obtain a structured diagnosis result of the aircraft hydraulic system.

[0043] The recording module presents the diagnosis result in real time through the on-board display module and records it in the on-board black box and the ground maintenance server.

[0044] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the health state detection method of the aircraft hydraulic system as described in the first aspect of the present invention is implemented.

[0045] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the health state detection method of the aircraft hydraulic system as described in the first aspect of the present invention is implemented.

[0046] The beneficial effects of the present invention are as follows: The quantum annealing technology is used to process the non-linear coupling relationship between multi-variables, and then the state probability distribution of the aircraft hydraulic system is generated through quantum optimization to capture the potential features of complex states. Through this step, the ability to efficiently solve high-dimensional non-linear problems is achieved, and the computational complexity and dimensionality limitation are significantly reduced compared with classical computing methods, the accuracy and real-time performance of state representation are improved, and accurate technical support is provided for the ultra-early detection of microscopic faults; in addition, by detecting the energy mutation rate and frequency domain offset of the acoustic emission subset, transient fault features can be captured, and a rapid response to faults can be realized. Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0048] Figure 1 It is a schematic diagram of the overall process of the aircraft hydraulic system health state detection method in Embodiment 1.

[0049] Figure 2 It is a schematic diagram of the collaborative work of multiple modules in the health state detection system in Embodiment 1.

[0050] Figure 3 It is a flowchart of multi-modal feature extraction and fusion processing in Embodiment 1.

[0051] Figure 4 It is an architecture diagram of the joint processing of quantum annealing and deep neural network in Embodiment 1. Specific Embodiments

[0052] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the accompanying drawings of the specification.

[0053] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0054] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0055] Embodiment 1, referring to Figure 1 、 Figure 2 、 Figure 3 and Figure 4 , is the first embodiment of the present invention. This embodiment provides a method for detecting the health state of an aircraft hydraulic system, including the following steps:

[0056] S1. Deploy a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream.

[0057] Specifically, it includes the following steps.

[0058] Deploy a four-dimensional heterogeneous sensing matrix at key parts of the aircraft hydraulic system (such as pumps, valves, pipeline joints, actuators, etc.), including pressure sensors, flow sensors, temperature sensors, and acoustic emission sensors. The sensors are firmly installed on the surface of the components through aviation-grade fixed fixtures or high-temperature adhesives to ensure anti-vibration (acceleration < 20g) and high-temperature resistance (< 150°C) performance. It is used to collect pressure data, flow data, temperature data, and broadband acoustic emission signal data of the aircraft hydraulic system.

[0059] Preferably, compared with the traditional detection method that uses a single type of sensor (such as pressure / temperature), this solution integrates a four-dimensional sensing matrix (pressure / flow / temperature / acoustic emission) to achieve complementary detection of physical quantities. The acoustic emission sensor fills the detection blind spot of traditional vibration sensors for early faults such as microcracks / cavitation.

[0060] Unify the numbering and timestamp marking of the real-time data of the aircraft hydraulic system collected in time series. The data of each sensor is marked with an accurate timestamp for subsequent data synchronization and analysis. For example, pressure data, flow data, temperature data, and acoustic emission signal data are all marked at the same time point to ensure data consistency and traceability.

[0061] In order to improve data transmission efficiency and reduce bandwidth occupancy, the collected raw data is compressed. An efficient compression algorithm (such as LZ77) is used to significantly reduce the data volume on the premise of ensuring data integrity. The compressed data is transmitted to the on-board processing unit through wired or wireless communication methods. In the on-board processing unit, the compressed data is decompressed and further processed to form a structured multi-modal data stream.

[0062] Preferably, through unified numbering in time series, timestamp marking, and decompression and recombination by the on-board processing unit, a multi-modal data stream is formed, which can well support subsequent deep learning analysis. The traditional processing method is to use discrete storage, and the data correlation is poor, making it difficult to support subsequent joint fault diagnosis.

[0063] S2. Divide the multi-modal data stream into two parts, the operating parameter subset and the acoustic emission subset, for preprocessing, extract features to generate feature vectors respectively, and at the same time concatenate the two feature vectors into a fusion feature vector.

[0064] Specifically, it includes the following steps:

[0065] Based on predefined logical rules (judged by signal type), divide the multi-modal data stream into two types of subsets. The operating parameter subset contains three types of data: pressure, flow, and temperature, reflecting the working state trend of the hydraulic system; the acoustic emission subset independently processes high-frequency acoustic emission signals to capture the transient fault characteristics of the equipment.

[0066] S2.1. Perform compressive sensing analysis on the acoustic emission subset to reconstruct the acoustic wave signal. Compressive sensing technology can maintain the main information of the signal while reducing the sampling rate, which is particularly important for processing high-frequency acoustic emission signals. Through this technology, high-quality acoustic wave signals can be recovered from limited data, thereby improving the accuracy of subsequent analysis.

[0067] Set a sliding window with a fixed width (e.g., 5 milliseconds) for segment-by-segment analysis of the acoustic wave signal;

[0068] Set the change rate threshold of the energy mutation rate and the peak frequency offset limit according to historical data (the values of the change rate threshold and the offset limit can be customized according to specific usage scenarios and actual situations).

[0069] Within each sliding window, calculate the energy mutation rate of the reconstructed acoustic wave signal. If the energy mutation rate exceeds the preset energy threshold, record the current time point as the first trigger time point. This step helps to identify significant fluctuations in the signal, which may indicate a potential fault.

[0070] For each trigger time point, use the Fast Fourier Transform (FFT) to transform the reconstructed acoustic wave signal into the frequency domain. In the frequency domain, detect the offset of the peak frequency. If the offset exceeds the preset offset limit, record this time point as the second trigger time point. This time-frequency combination method can more comprehensively capture the subtle changes in the acoustic emission signal.

[0071] When the two trigger time points coincide, extract the characteristics of the reconstructed acoustic wave signal centered on this time point, including peak frequency, energy amplitude, duration, and frequency band width, to form an acoustic emission feature vector. The characteristics of the reconstructed acoustic wave signal can effectively describe the abnormal patterns in the acoustic emission signal and provide a basis for subsequent fault diagnosis.

[0072] Perform statistical analysis on the operating parameter subset (pressure, flow rate, temperature data), calculate its mean and standard deviation, and obtain an operating parameter feature vector. Statistical features can reflect the overall operating condition of the hydraulic system and help identify potential abnormal trends.

[0073] Combine the acoustic emission feature vector and the operating parameter feature vector into a fusion feature vector by splicing. This fusion not only integrates different types of feature information but also enhances the integrity of the feature vector.

[0074] Furthermore, the traditional dual-modal trigger strategy has only a single energy threshold, and the trigger is vulnerable to environmental noise interference. Moreover, it is difficult to capture frequency domain anomalies only by relying on time domain features. However, the energy-frequency joint trigger mechanism of this scheme can simultaneously capture transient impact signals and identify harmonic offsets caused by component wear.

[0075] S3. Map the fused feature vector through encoding to an initial quantum state, establish a quantum annealing model, and obtain a state representation result.

[0076] Specifically, it includes the following steps:

[0077] Normalize the fused feature vector extracted and concatenated from the multimodal data stream. Normalization is to ensure the consistency of scales among different features and avoid some features dominating the optimization process due to overly large numerical values. Subsequently, use a quantum state representation method (such as qubits or quantum registers) to map the normalized fused feature vector to an initial quantum state.

[0078] Furthermore, by mapping the fused feature vector to an initial quantum state, the limitations of traditional methods in high-dimensional spaces are overcome. The quantum state representation method can handle complex non-linear relationships in higher-dimensional spaces, thereby more accurately capturing minute changes in the aircraft hydraulic system.

[0079] S3.1. In the quantum annealing model, set an energy function as the core. The energy function consists of an initial energy term and a target problem term. The initial energy term is usually used to describe the ground state energy of the aircraft hydraulic system, while the target problem term reflects the objective function that the aircraft hydraulic system needs to optimize.

[0080] Complete the configuration of the quantum annealing model by selecting a linear scheduling method (i.e., the process of gradually adjusting parameters to guide the aircraft hydraulic system from the initial state to the target state). The linear scheduling method can effectively control the evolution path of the quantum state and ensure that the aircraft hydraulic system can smoothly reach the optimal solution.

[0081] Use the quantum annealing algorithm to minimize the target problem term through an iterative process and find the target state of the aircraft hydraulic system. During this process, dynamically adjust the initial quantum state according to the linear scheduling method to gradually transition it to the target state. Each iteration will generate a probability distribution of the aircraft hydraulic system state, including normal state, microscopic fault states (such as nanoscale wear of the sealing ring) and severe fault states (such as pipeline leakage). The probability distribution reflects the likelihood of the aircraft hydraulic system in different health states.

[0082] Preferably, classical sensor signal analysis (such as threshold alarm, spectrum analysis) has insufficient sensitivity to micron- and nanoscale early faults (such as nanoscale wear of the sealing ring, tiny oil bubbles) and a high false alarm rate. The present invention can detect microscopic energy changes that cannot be recognized by traditional methods through quantum annealing, directly output the probability of microscopic faults instead of binary judgment, and support preventive maintenance decisions.

[0083] To further refine the state representation results, a post-processing tool is used to analyze the evolution trajectory of the quantum state. Specifically, based on the energy function, the weight value of each feature in the fused feature vector is calculated through gradient calculation to form a weight vector. This process involves quantitative analysis of the contribution degree of each feature during the quantum state evolution process, which can help identify which features have a greater impact on the state change of the aircraft hydraulic system. The weight vector not only provides information on the importance of features but also provides a basis for subsequent state classification and fault diagnosis.

[0084] S4. Use a pre-trained deep neural network to process the state representation results and calculate the health index to obtain a structured diagnosis result of the aircraft hydraulic system.

[0085] Specifically, it includes the following steps:

[0086] Based on the historical fault data of the hydraulic system, a historical frequency spectrum database is constructed. By analyzing past fault cases, the characteristic patterns of each component (such as hydraulic pumps, main control valves) during faults are extracted. For example, the corresponding relationship between the specific frequency peak of the acoustic emission signal and component damage. Then, according to the statistical analysis of the historical fault data, a fault threshold is set, which reflects the critical point where the sum of the probabilities of the microscopic fault state and the severe fault state is abnormal.

[0087] Furthermore, by establishing a historical frequency spectrum database and setting a fault threshold, the induction of fault characteristics and the quantification of the abnormal judgment standard are realized, providing a reliable basis for fault location and state classification.

[0088] S4.1. First, construct a deep neural network. The first layer is the input layer, the second and third layers are hidden layers, each layer contains a fixed number of neurons, and the fourth layer is the output layer, designed with three output nodes, corresponding to the three state categories of the hydraulic system: normal state, microscopic fault state, and severe fault state. The training process adopts the supervised learning method, and the goal is to make the output of the deep neural network match the labeled state categories (normal, microscopic fault, severe fault) as much as possible. The specific method uses the backpropagation algorithm, combined with the stochastic gradient descent optimizer, to adjust the weights and biases of the network through multiple rounds of iteration. The loss function selects the cross-entropy loss, which is used to measure the difference between the predicted probability and the true label, driving the model to converge.

[0089] S4.2. The deep neural network extracts the probability values in the state probability distribution, corresponding to the normal state, microscopic fault state, and severe fault state respectively. Through internal weights and activation functions, the confidence of each state is calculated to generate a preliminary state judgment of the aircraft hydraulic system. The calculation results are stored as a state vector, indicating the tendency of the current state.

[0090] Check the sum of the probability values of the micro fault state and the severe fault state, and determine whether it exceeds the fault threshold. If it exceeds the threshold, it is determined that there is an abnormality in the aircraft hydraulic system, and a trigger signal is generated to indicate whether to perform fault location. If it does not exceed the threshold, it is determined to be in a normal state, and the trigger signal value is no need to locate.

[0091] Preferably, the present invention utilizes the dynamic classification ability of the neural network, surpasses the singularity of the traditional threshold judgment, achieves the effect of improving the sensitivity and accuracy of state recognition, and provides more superior support for ultra-early fault detection.

[0092] S4.3. Check the trigger signal. If it indicates the need for location, extract the weight vector from the state characterization result, analyze the magnitude of each weight value, and determine which features contribute more to the abnormal state. For example, the weight of the acoustic emission peak frequency may be relatively high, indicating that it has a significant impact on the fault. Load the established historical spectrum database and call the feature pattern data therein, such as the correspondence between specific frequency peaks and component failures. Then, match the features with larger weight values with the patterns in the database, and compare the feature values with the fault patterns through the pattern recognition algorithm to identify the possible faulty components. For example, if the acoustic emission peak frequency matches the fault characteristics of the hydraulic pump, it is located as the hydraulic pump. The location result is stored as a component identifier, and the faulty component identifier is output.

[0093] Further explanation, compared with the prior art, combining the weight vector and the historical spectrum database surpasses the rough location relying on a single feature of the traditional method, and improves the accuracy of fault location and the depth of feature analysis.

[0094] S4.4. Extract the probability value of the normal state from the state probability distribution, multiply it by one hundred, convert it into a percentage value, and store it as the health index, which represents the health degree of the aircraft hydraulic system. Subsequently, compare the health index with a health threshold preset based on historical fault data analysis. If it is lower than the health threshold, it is determined to be an ultra-early fault; if it is higher than or equal to the threshold, confirm the state category according to the state label, such as normal or micro fault. Then, integrate the state label, the health index, and the faulty component identifier to form a structured diagnostic report, which contains three fields: the state category describes the current state, the health index quantifies the health degree, and the faulty component identifier indicates the abnormal location. If the location is not triggered, this field is empty.

[0095] Preferably, through the calculation of the health index and the generation of the structured diagnostic report, the quantitative evaluation and multi-dimensional integration of the state of the aircraft hydraulic system are combined, providing comprehensive health information. Compared with the prior art, this step surpasses the limitations of the traditional single index through neural network classification and health index quantification, and provides a reliable basis for real-time monitoring and fault warning.

[0096] S5. Present the diagnostic results in real time through the on-board display module and record them in the on-board black box and the ground maintenance server.

[0097] Specifically, it includes the following steps:

[0098] Transmit the diagnostic report to the on-board display module. The on-board display module is a display device installed in the cockpit, such as a high-resolution liquid crystal screen. Load the diagnostic results onto the display interface. The status category is displayed in text form, such as micro-failure. The health index is presented as a percentage bar chart, and the faulty components are labeled with specific names, such as hydraulic pump. The on-board display module refreshes the data at a fixed frequency to ensure that the pilot can view the latest status of the aircraft hydraulic system in real time. Subsequently, send the same diagnostic report to the on-board black box, which is a storage device resistant to high temperatures and impacts. Write the data through the internal interface and record it as a timestamped log entry. At the same time, transmit the diagnostic report to the ground maintenance server through an aviation communication link, such as a satellite or 5G network. After the server receives the data, it stores it in the database for maintenance personnel to view. The transmission process ensures low latency, and the data is saved in encrypted form to prevent loss and tampering.

[0099] This embodiment also provides a health status detection method system for an aircraft hydraulic system, including:

[0100] An acquisition module that deploys a multi-modal sensor array to acquire real-time data of the aircraft hydraulic system and form a structured multi-modal data stream;

[0101] A fusion module that divides the multi-modal data stream into two parts: an operating parameter subset and an acoustic emission subset for preprocessing, extracts features to generate feature vectors respectively, and concatenates the two feature vectors into a fusion feature vector;

[0102] A quantum processing module that encodes and maps the fusion feature vector into a quantum initial state, establishes a quantum annealing model, and obtains a state characterization result;

[0103] A diagnostic module that uses a pre-trained deep neural network to process the state characterization result and calculate the health index to obtain a structured diagnostic result of the aircraft hydraulic system;

[0104] A recording module that presents the diagnostic results in real time through the on-board display module and records them in the on-board black box and the ground maintenance server.

[0105] This embodiment also provides a computer device applicable to the health status detection method of an aircraft hydraulic system, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the health status detection method of the aircraft hydraulic system proposed in the above embodiment.

[0106] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball, or touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0107] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for detecting the health state of an aircraft hydraulic system as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disc.

[0108] In summary, the present invention: uses quantum annealing technology to process the non-linear coupling relationship between multiple variables, and then generates the state probability distribution of the aircraft hydraulic system through quantum optimization to capture the potential characteristics of complex states. Through this step, it achieves the ability to efficiently solve high-dimensional non-linear problems, significantly reduces the computational complexity and dimensionality limitations compared with classical computing methods, improves the accuracy and real-time performance of state representation, and provides accurate technical support for the ultra-early detection of microscopic faults; in addition, by detecting the energy mutation rate and frequency domain offset of the acoustic emission subset, it can capture transient fault characteristics and achieve a rapid response to faults.

[0109] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for detecting the health state of an aircraft hydraulic system, characterized in that: including, deploying a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream; dividing the multi-modal data stream into two parts, namely an operating parameter subset and an acoustic emission subset, for preprocessing, respectively extracting features to generate feature vectors, and at the same time splicing the two feature vectors into a fused feature vector; encoding and mapping the fused feature vector into an initial quantum state, establishing a quantum annealing model, and obtaining a state representation result; using a pre-trained deep neural network to process the state representation result and calculate a health index to obtain a structured diagnosis result of the aircraft hydraulic system; presenting the diagnosis result in real time through an on-board display module and recording it in the on-board black box and the ground maintenance server.

2. The health state detection method of the aircraft hydraulic system according to claim 1, characterized in that: Deploying a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream, specifically including the following steps, unifying the collected real-time data according to the sequence and timestamp, compressing it and transmitting it to the on-board processing unit to obtain a structured multi-modal data stream.

3. The health status detection method of the aircraft hydraulic system according to claim 2, characterized in that: Dividing the multi-modal data stream into two parts, namely an operating parameter subset and an acoustic emission subset, for preprocessing, respectively extracting features to generate feature vectors, and at the same time splicing the two feature vectors into a fused feature vector, specifically including the following steps, using a predefined mapping table to divide the multi-modal data stream into an operating parameter subset and an acoustic emission subset; performing compressive sensing analysis on the acoustic emission subset to obtain a reconstructed acoustic wave signal; applying a dual-modal triggering strategy, setting a sliding window with a fixed width, and presetting a change rate threshold and an offset limit; calculating the energy mutation rate of the reconstructed acoustic wave signal, recording the trigger time point when the energy mutation rate exceeds the preset energy threshold, using the fast Fourier transform to convert the reconstructed acoustic wave signal into the frequency domain, detecting the offset of the peak frequency, and recording another trigger time point when the offset exceeds the offset limit; when the two trigger time points coincide, extracting the features of the reconstructed acoustic wave signal centered on this time point to obtain an acoustic emission feature vector; calculating the mean and standard deviation of the operating parameter subset through statistical analysis to obtain an operating parameter feature vector; combining the acoustic emission feature vector and the operating parameter feature vector into a fused feature vector by splicing.

4. The health state detection method of the aircraft hydraulic system according to claim 3, characterized in that: Encoding and mapping the fused feature vector into an initial quantum state, establishing a quantum annealing model, specifically including the following steps, normalizing the fused feature vector and mapping it into an initial quantum state; setting an energy function as the core of the quantum annealing model and selecting a linear scheduling method to complete the configuration of the quantum annealing model.

5. The health status detection method of the aircraft hydraulic system according to claim 4, characterized in that: The said obtaining the state representation result specifically includes the following steps, obtaining a target state by minimizing the target problem term, adjusting the initial quantum state according to the selected linear scheduling method, transitioning to the target state, and using iteration to generate the probability distribution of the hydraulic system state; using a post-processing tool to analyze the evolution trajectory of the quantum state, calculating the weight value of each feature in the fused feature vector to form a weight vector; combining the weight vector with the probability distribution of the hydraulic system state to obtain a state representation result.

6. The health status detection method of the aircraft hydraulic system according to claim 5, characterized in that: Using a pre-trained deep neural network to process the state representation result and calculate a health index to obtain a structured diagnosis result of the aircraft hydraulic system, specifically including the following steps, Establish a historical spectrum database based on historical fault data and set a fault threshold; Input the state characterization result into a pre-trained deep neural network; Extract the probability values of each probability distribution in the state characterization result and perform calculations to obtain a preliminary state judgment of the aircraft hydraulic system; Compare the sum of the probability values of the microscopic fault state and the severe fault state with the fault threshold, and obtain a state label and a trigger signal after exceeding the fault threshold; Extract the weight vector from the state characterization result, determine the features that contribute greatly to the abnormal state, match them with the patterns in the historical spectrum database, identify the faulty components, and obtain the structured diagnosis result of the aircraft hydraulic system.

7. The method for detecting the health state of an aircraft hydraulic system according to claim 6, characterized in that: The obtaining of the structured diagnosis result of the aircraft hydraulic system specifically includes the following steps: Calculate the health index according to the probability value of the normal state, and integrate the state label and the faulty components to generate a diagnosis report of the aircraft hydraulic system.

8. A health status detection method and system for an aircraft hydraulic system, based on the health status detection method for an aircraft hydraulic system according to any one of claims 1 to 7, characterized in that: Including: A collection module that deploys a multi-modal sensor array to collect real-time data of the aircraft hydraulic system and form a structured multi-modal data stream; A fusion module that divides the multi-modal data stream into two parts, namely an operating parameter subset and an acoustic emission subset, for preprocessing, extracts features to generate feature vectors respectively, and at the same time splices the two feature vectors into a fused feature vector; A quantum processing module that encodes and maps the fused feature vector into a quantum initial state, establishes a quantum annealing model, and obtains a state characterization result; A diagnosis module that uses a pre-trained deep neural network to process the state characterization result and calculate the health index to obtain the structured diagnosis result of the aircraft hydraulic system; A recording module that presents the diagnosis result in real time through the on-board display module and records it in the on-board black box and the ground maintenance server.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for detecting the health state of the aircraft hydraulic system according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for detecting the health state of the aircraft hydraulic system according to any one of claims 1 to 7.

Images