An aircraft maintenance data visualization method, system, electronic device and product

Abstract

The present invention belongs to the technical field of aircraft maintenance, and its purpose is to provide an aircraft maintenance data visualization method, system, electronic device and product. The method includes: obtaining multi-dimensional operation data of an aircraft; inputting the multi-dimensional operation data into a preset equipment health perception model to obtain health status characterization data; obtaining an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data; and displaying the health status characterization data and the optimal maintenance strategy in a visual manner on an AI prediction dashboard. The present invention can realize the dynamic analysis of aircraft maintenance data and the dynamic decision-making of maintenance strategies, and can simultaneously realize the visual display of the optimal maintenance strategy in a visual manner.

Description

Technical Field

[0001] The present invention belongs to the technical field of aircraft maintenance, and specifically relates to an aircraft maintenance data visualization method, system, electronic device and product. Background Art

[0002] With the rapid development of the air transportation industry, the operation safety and maintenance efficiency of aircraft have become key factors to ensure the smooth completion of flight missions. At present, in the field of aircraft maintenance, a strategy combining regular inspections (scheduled inspections) and preventive maintenance is widely adopted, and maintenance plans are formulated according to fixed rules such as flight hours, takeoff and landing times or calendar time.

[0003] In recent years, with the development of sensor technology and data acquisition systems, the operation data during the scheduled inspection and maintenance process of aircraft has been continuously collected and stored. The visualization analysis system constructed based on these data has gradually been applied to assist maintenance work, and displays aircraft status information through charts, dashboards, etc., providing certain decision-making references for operation and maintenance personnel. For example, a visualization method, device, equipment and medium for aircraft scheduled inspection and maintenance data disclosed in a Chinese patent with the publication number CN119515346A. However, in the process of using the existing technology, the inventor found that there are at least the following problems in the existing technology:

[0004] The existing technology mostly stays at the level of static display, fails to deeply explore the potential laws in historical data and real-time data, lacks intelligent analysis and prediction capabilities, and cannot fully support the optimization and upgrade of maintenance strategies; at the same time, the existing maintenance strategies still rely on manual rule-making, with low automation, which is not conducive to realizing dynamic decision-making. Summary of the Invention

[0005] The present invention aims to solve the above technical problems at least to a certain extent, and provides an aircraft maintenance data visualization method, system, electronic device and product.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides an aircraft maintenance data visualization method, including:

[0008] Obtaining multi-dimensional operation data of the aircraft;

[0009] Inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data;

[0010] Obtaining an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data;

[0011] Displaying the health status characterization data and the optimal maintenance strategy in a visual manner on an AI prediction dashboard.

[0012] In a possible design, the health status characterization data includes health status time series feature data, current health status information, current predicted remaining life value of the device, and predicted remaining life sequence; correspondingly, inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data, including:

[0013] Inputting the multi-dimensional operation data into a preset device health perception model to obtain health status time series feature data, and using the feature data at the latest moment in the health status time series feature data as the current health status information of the aircraft;

[0014] Based on the health status time series feature data, obtaining the predicted remaining life sequence of the aircraft, and using the predicted remaining life value at the latest moment in the predicted remaining life sequence as the current predicted remaining life value of the aircraft.

[0015] In a possible design, the device health perception model uses a long short-term memory network; correspondingly, in the health status time series feature data, the feature data at time t is:

[0016] ;

[0017] where LSTM() represents the long short-term memory network, x t is the multi-dimensional operation data at time t , t = 1, 2,..., T , T is the time index, T is a natural number greater than 1, h t-1 is the feature data at time t - 1;

[0018] In the predicted remaining life sequence, the predicted remaining life value at time t is:

[0019] ;

[0020] where W is a preset weight matrix, b is a preset correction parameter.

[0021] In a possible design, obtaining an optimal maintenance strategy corresponding to the health status characterization data based on the health status characterization data, including:

[0022] Input the current health status information in the health status characterization data into a preset maintenance strategy optimization module, and calculate the expected return value corresponding to the maintenance actions in the action space based on the value function of multiple state-action pairs formed by the current health status information and the maintenance actions in the action space of the maintenance strategy optimization module;

[0023] Take the maintenance action in the state-action pair with the maximum expected return value as the optimal maintenance strategy.

[0024] In a possible design, the maintenance strategy optimization module adopts a deep Q-network; correspondingly, the state space of the maintenance strategy optimization module is S={ s i ∣ i =1,2,..., I}, s i is the i th state element in the state space, I is the total number of state elements in the state space, the action space of the maintenance strategy optimization module is A={ a 1: Immediate maintenance, a 2: Deferred maintenance, a 3: Ignore}, the reward function of the maintenance strategy optimization module is:

[0025] ;

[0026] In the formula, R ( s , a ) is the reward value for executing the maintenance action s in the health state a , Cmaint ( a ) is the direct maintenance cost brought by executing the maintenance action a , Crisk ( s , a ) is the potential failure risk loss caused by executing the maintenance action s in the health state a , Cavailability ( a ) is the flight mission availability benefit brought by the maintenance.

[0027] In a possible design, the value function is:

[0028] ;

[0029] In the formula, Q ( s , a ) is in the health states Perform maintenance actions a The cumulative expected benefits brought about α is the preset learning rate γ is the preset discount factor s ' is the action executed a The new health state transferred to after a ' is in the new health state s ' is the best action among all possible actions in the new health state Q ( s ', a ') is the maximum expected benefit that can be obtained by executing the action a ' in the health of the new health state;

[0030] Correspondingly, the optimal maintenance strategy is:

[0031] = argmax a Q ( h T , a )

[0032] where argmax a is the argmax function that finds the maintenance action that maximizes the value function among all maintenance actions in the action space Q ( h T , a ) is the cumulative expected benefit brought about by executing the maintenance action h T under the current health state information a ;

[0033] In a possible design, the method further includes:

[0034] Generating maintenance assistance decision-making suggestions according to the optimal maintenance strategy;

[0035] Generating strategy explanation information according to the maintenance assistance decision-making suggestions, and outputting a manual confirmation prompt message bound with the maintenance assistance decision-making suggestions and the strategy explanation information;

[0036] Judging whether a manual confirmation instruction is received within a specified duration. If so, generating a maintenance task plan according to the maintenance assistance decision-making suggestions, and feeding back the execution result after executing the maintenance task plan to the device health perception model for self-learning update of the device health perception model.

[0037] In a second aspect, the present invention provides an aircraft maintenance data visualization system, including:

[0038] A data acquisition module, configured to obtain multi-dimensional operation data of an aircraft;

[0039] A health data extraction module, communicatively connected to the data acquisition module, configured to input the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data;

[0040] A maintenance strategy generation module, communicatively connected to the health data extraction module, configured to obtain an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data;

[0041] A visualization module, communicatively connected to the maintenance strategy generation module, configured to visually display the health status characterization data and the optimal maintenance strategy on an AI prediction dashboard.

[0042] In a third aspect, the present invention provides an electronic device, including:

[0043] A memory, configured to store computer program instructions; and,

[0044] A processor, configured to execute the computer program instructions to complete the operations of an aircraft maintenance data visualization method as described in any one of the above.

[0045] In a fourth aspect, the present invention provides a computer program product, including a computer program or instructions, and the computer program or the instructions, when executed by a computer, implement an aircraft maintenance data visualization method as described in any one of the above.

[0046] The beneficial effects of the present invention are as follows:

[0047] The present invention discloses an aircraft maintenance data visualization method, system, electronic device and product, which can realize the dynamic analysis of aircraft maintenance data and the dynamic decision-making of maintenance strategies, and can also realize the visual display of the optimal maintenance strategy in a visual manner. In the implementation process of the present invention, by obtaining the multi-dimensional operation data of the aircraft and inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data, then, according to the health status characterization data, an optimal maintenance strategy corresponding to the health status characterization data is obtained, and finally, the health status characterization data and the optimal maintenance strategy are visually displayed on an AI prediction dashboard. The present invention can realize the dynamic analysis of aircraft maintenance data and the dynamic decision-making of maintenance strategies, and can also realize the visual display of the optimal maintenance strategy in a visual manner, which helps to improve the aircraft maintenance efficiency and is applicable to the application scenarios of aircraft scheduled inspection and preventive maintenance.

[0048] Other beneficial effects of the present invention will be further described in the specific implementation manner. Brief Description of the Drawings

[0049] Figure 1 is a flowchart of the aircraft maintenance data visualization method in Embodiment 1;

[0050] Figure 2 is a block diagram of the modules of the aircraft maintenance data visualization system in Embodiment 2;

[0051] Figure 3 is a block diagram of the modules of the electronic device in Embodiment 3. Detailed Implementation Manner

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the present invention in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structures of the drawings is only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. It should be noted here that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation on the present invention.

[0053] Embodiment 1:

[0054] This embodiment discloses an aircraft maintenance data visualization method, which can be, but is not limited to, executed by a computer device or virtual machine with certain computing resources, such as an electronic device such as a personal computer, a smart phone, a personal digital assistant, or a wearable device, or executed by a virtual machine.

[0055] As Figure 1 shown, an aircraft maintenance data visualization method can, but is not limited to, include the following steps:

[0056] S1. Obtain the multi-dimensional operation data of the aircraft X ={ x t ∣ t =1, 2,..., T}, where x t is the t -dimensional operation data at time n , which reflects the multi-dimensional operation state of the aircraft at this moment, T is the time index, n is the dimension of the multi-dimensional operation data, n and T are natural numbers greater than 1 respectively. The multi-dimensional operation data X represents at TThe operation data collected at consecutive time steps. It should be noted that the multi-dimensional operation data includes sensor data such as temperature, pressure, and vibration, and may also include, but is not limited to, flight record data, historical maintenance records, and fault logs, etc., which are not restricted here. In this embodiment, the system obtains in real time the multi-dimensional operation data of the previous T consecutive time steps X , so as to realize the dynamic analysis of the operation data and the dynamic generation of maintenance strategies.

[0057] It should be understood that in this embodiment, after the initial multi-dimensional operation data is collected, the initial multi-dimensional operation data is also preprocessed to obtain the preprocessed multi-dimensional operation data and perform subsequent data processing, thereby improving the quality of the data and the efficiency of analysis. Specifically, the preprocessing process of the multi-dimensional operation data includes, but is not limited to, outlier removal, missing value filling (such as using KNN interpolation or linear interpolation), normalization processing, and normalization processing, which are not restricted here.

[0058] S2. Input the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data. Among them, the health status characterization data includes the predicted value of the remaining life of the device and the current health status information h T and other data, which are used to comprehensively reflect the current and future operation status of the device. It should be understood that in this embodiment, the device health perception model is pre-trained through sample data, and the device health perception model has the ability to automatically extract key health features from the original operation data and perform life prediction, which is the core basis for realizing intelligent maintenance decision-making.

[0059] In step S2, the health status characterization data includes health status time series feature data, current health status information, current predicted value of the remaining life of the device, and remaining life prediction sequence; correspondingly, inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data includes:

[0060] S201. Input the multi-dimensional operation data into a preset device health perception model to obtain health status time series feature data, and use the feature data at the latest moment in the health status time series feature data as the current health status information of the aircraft. It should be understood that the health status time series feature data is the hidden state vector encoded by the device health perception model for the multi-dimensional operation data X , representing the health status of the aircraft at T consecutive time steps.

[0061] S202. Based on the time-series feature data of the health status, obtain the remaining life prediction sequence of the aircraft, and use the remaining life prediction value at the latest moment in the remaining life prediction sequence as the current remaining life prediction value of the aircraft.

[0062] Specifically, in step S2, the device health perception model uses a long short-term memory network (LSTM); correspondingly, in the time-series feature data of the health status, the feature data at time t is:

[0063] ;

[0064] In the formula, LSTM() represents the long short-term memory network, x t is the multi-dimensional operation data at time t , t = 1, 2,..., T , T is the time index, T is a natural number greater than 1, h t-1 is the feature data at time t -1, t When h = 1,

[0065] 0 is the preset initial feature data; h t | t = 1, 2,..., T}, the current health status information is also the feature data at time T in the time-series feature data of the health status h T .

[0066] In the remaining life prediction sequence, the remaining life prediction value at time t is:

[0067] ;

[0068] In the formula, W is the weight matrix of the output layer of the preset device health perception model, with a dimension of 1×n, b is the preset correction parameter, that is, the bias term of the output layer of the device health perception model, which is a scalar used to adjust the output prediction value.

[0069] It should be understood that the remaining life prediction sequence can be expressed as , the current remaining life prediction value is also the remaining life prediction value at the moment in the remaining life prediction sequence T of the remaining life prediction value .

[0070] It should be noted that the device health perception model can also adopt a Transformer network based on the attention mechanism. Relatively speaking, the long short-term memory network has unique advantages in dealing with the time series characteristics of device operation data. Specifically, the long short-term memory network has a gating structure, which can effectively capture the long-term dependencies and short-term dynamic changes in the sequence, and has stronger robustness and stability for dealing with the implicit trend degradation, mutation signals, etc. in the device state. Especially in industrial data scenarios with limited sample size or strong time continuity, the long short-term memory network can better retain key time information, avoid the information dilution problem that may be brought by the attention mechanism, and thus improve the reliability of accurate modeling of the device health state and remaining life prediction.

[0071] S3. Obtain the optimal maintenance strategy corresponding to the health state characterization data according to the health state characterization data . Among them, the maintenance strategy optimization module is used to evaluate the long-term benefits under different maintenance strategies to realize the recommendation of the optimal maintenance strategy.

[0072] Specifically, in step S3, the optimal maintenance strategy corresponding to the health state characterization data is obtained according to the health state characterization data , including:

[0073] S301. Input the current health state information in the health state characterization data into a preset maintenance strategy optimization module, and calculate the expected return value corresponding to the maintenance actions in the action space based on the value function of multiple state-action pairs formed by the current health state information and the maintenance actions in the action space of the maintenance strategy optimization module.

[0074] S302. Take the maintenance action in the state-action pair with the maximum expected return value as the optimal maintenance strategy.

[0075] Specifically, in this embodiment, the maintenance strategy optimization module adopts a Deep Q-Network (DQN); correspondingly, the state space of the maintenance strategy optimization module is S={ s i ∣ i =1,2,..., I}, s i is the i th state element in the state space, Iis the total number of state elements in the state space, and the action space of the maintenance strategy optimization module is A = { a 1: Immediate maintenance, a 2: Deferred maintenance, a 3: Ignore}. The reward function of the maintenance strategy optimization module is:

[0076] ;

[0077] In the formula, R ( s , a ) is the reward value for executing the maintenance action s in the healthy state a , that is, the direct benefit obtained after executing the maintenance action s in the healthy state a . Cmaint ( a ) is the direct maintenance cost brought by executing the maintenance action a . Crisk ( s , a ) is the potential failure risk loss caused by executing the maintenance action s in the healthy state a . Cavailability ( a ) is the flight mission availability benefit brought by the maintenance.

[0078] It should be noted that the state space S represents all the states that the aircraft may be in, such as the current healthy state; the action space A represents all the maintenance strategies that the agent can take in any state; the reward function R ( s , a ) is used to evaluate the immediate benefit or loss brought by taking the maintenance action s in the state a . In this embodiment, the maintenance strategy optimization module can select a strategy by maximizing the expected reward R ( s , a ) so that the cumulative reward is maximized in future interactions, thereby finding an optimal balance among maintenance costs, failure risks, and availability benefits and achieving an economical and efficient intelligent maintenance decision.

[0079] Specifically, in this embodiment, the direct maintenance cost a brought by executing the maintenance action Cmaint ( a ) is defined as:

[0080] ;

[0081] Among them, C 0 indicates no cost for non - repair, C 0 = 0, C minor indicates the cost of minor repair, C major indicates the cost of comprehensive or emergency repair, and C major > C minor .

[0082] The loss of failure risk that may be caused by performing a maintenance action s under the healthy state a is: Crisk ( s , a ) is:

[0083] Crisk ( s , a ) = P fail ( s , a ) · L fail ;

[0084] In the formula, P fail ( s , a ) is the failure probability after performing the maintenance action s under the healthy state a , which can be estimated by the model or set by manual experience value. L fail is the expected loss caused by a single failure, including comprehensive costs such as maintenance, flight interruption, and personal injury.

[0085] The benefit of flight mission availability brought by maintenance Cavailability ( a ) is:

[0086] Cavailability ( a ) = β · Tavailabile ( a ) ;

[0087] In the formula, β is the preset benefit weight coefficient of unit - time availability, Tavailabile ( a ) is the expected time for the aircraft to remain in an available state a after performing the maintenance action Tavailabile ( a1) = 10 days, Tavailabile ( a 2) = 5 days, Tavailabile ( a 3) = 0.

[0088] In this embodiment, the maintenance strategy optimization module guides the agent to learn which maintenance strategy should be adopted in what state by evaluating the impact of the combination of state and action on the long-term benefit of the system, so as to achieve an optimal balance among cost reduction, risk control, and equipment availability improvement. Based on the setting of the maintenance strategy optimization module in this embodiment, the maintenance strategy no longer depends on manual experience, but is adaptively optimized by data-driven.

[0089] In this embodiment, the value function is:

[0090] ;

[0091] In the formula, Q ( s , a ) is the cumulative expected benefit brought by executing the maintenance action s in the healthy state a . During the implementation process, the agent estimates and updates this Q value through continuous learning; α is the preset learning rate, α and the value of γ is in the range of (0, 1); γ is the preset discount factor, which is used to measure the importance of future rewards, γ and the value of s is in the range of (0, 1), a and the larger the value of a , the more importance is attached to the long-term benefit; s ′ is the new healthy state transferred to after executing the action Q ( s ′, a ′), a ′ is the best action among all possible actions in the new healthy state

[0092] Q s ( a , ) is corrected by the currently obtained reward and the expected benefit of the future optimal action to gradually approach the optimal strategy. Finally, the agent can learn to select the optimal action that can bring the maximum long-term benefit in any state, thus realizing intelligent decision-making.

[0093] In this embodiment, the value function adopts the Q-value function, which is used to evaluate the cumulative expected return that the agent can obtain in the future according to the current policy after performing a certain action in a specific state. Different from the state value function that only evaluates the quality of the state s when performing a certain action a , the Q-value function takes into account the impact of the currently selected action on future returns. Therefore, it is more suitable for policy optimization and action selection, and is the key basis for implementing the optimal decision-making strategy for aircraft maintenance (such as the greedy strategy or V ( s ) epsilon-greedy strategy). ε

[0094] Correspondingly, the optimal maintenance strategy is:

[0095] = argmax a Q ( h T , a )

[0096] where argmax a is the argmax function that finds the maintenance action that maximizes the value function among all maintenance actions in the action space, Q ( h T , a ) is the cumulative expected return brought by performing the maintenance action h T in the current health state information a .

[0097] It should be noted that the optimal maintenance strategy is the maintenance action selected by the agent by comparing the Q-values of all candidate actions in the current state, which can bring the maximum long-term return. Based on this, the optimal decision-making of the maintenance strategy can be realized.

[0098] S4. Display the health state characterization data and the optimal maintenance strategy visually on the AI (Artificial Intelligence) prediction dashboard.

[0099] It should be noted that the AI prediction dashboard supports the comparative display of the state information of multiple devices and has a function of warning about abnormal maintenance trends. In this embodiment, the AI prediction dashboard can visually display the health state characterization data and the optimal maintenance strategy to maintenance personnel, which is conducive to realizing human-machine collaborative decision-making.

[0100] ​Specifically, the display content of the AI prediction dashboard includes the remaining life trend chart of the device (such as set as a line chart), the life confidence interval prediction chart, and the comparison chart of the expected benefits of maintenance strategies (such as showing the expected benefits of each maintenance action in the form of a bar chart) obtained by aggregating and structuring the health status characterization data and the optimal maintenance strategy. In this embodiment, the AI prediction dashboard can be constructed using the Web front-end technology of React (an open-source JavaScript library for building user interfaces) + ECharts (an open-source visualization chart library), or commercial BI (Business Intelligence) tools (such as Tableau, PowerBI), and supports user interaction viewing and parallel presentation of multi-device information. In this embodiment, the method further includes:

[0101] S5. Generate maintenance assistance decision-making suggestions according to the optimal maintenance strategy. It should be noted that the maintenance assistance decision-making suggestions include an explanation of the recommended reasons, adjustable parameters of the strategy, and an interactive work order generation function, which are not limited here.

[0102]

[0103] S6. Generate strategy explanation information according to the maintenance assistance decision-making suggestions, and output an artificial confirmation prompt message bound with the maintenance assistance decision-making suggestions and the strategy explanation information.

[0104] As an example, the strategy explanation information is "Based on the current predicted remaining life of 120 hours, the system recommends a regular inspection within 24 hours to avoid the risk of failure."

[0105] In this embodiment, the artificial confirmation prompt message can be displayed visually on the AI prediction dashboard or output in the form of voice, as long as it can be received by the maintenance personnel.

[0106] Correspondingly, this embodiment should provide an artificial confirmation interface (such as a touch button, confirmation button, etc.) corresponding to the artificial confirmation prompt message to receive an artificial confirmation instruction or a rejection instruction.

[0107] S7. Determine whether an artificial confirmation instruction is received within a specified duration. If so, generate a maintenance task plan according to the maintenance assistance decision-making suggestions, and feedback the execution result after executing the maintenance task plan to the device health perception model for self-learning and updating of the device health perception model.

[0108] ​​It should be noted that the maintenance personnel can also fine-tune the strategy interpretation information based on experience, so as to finally generate a maintenance task plan and a maintenance scheduling work order for executing the maintenance task plan.

[0109] In this embodiment, based on the above step S6, the role of combining the AI prediction result with manual experience for auxiliary decision-making can be realized, which is beneficial to further improve the maintenance safety on the premise of ensuring the maintenance efficiency.

[0110] In this embodiment, based on the above step S7, the execution result after executing the maintenance task plan includes the latest remaining life prediction value T+1 and the health status information h T+1 , by using it as a training sample and flowing it back to the device health perception model, the parameters of the device health perception model can be updated, the self-learning closed-loop of the model can be realized, and further it helps to continuously optimize the system performance.

[0111] This embodiment can realize the dynamic analysis of aircraft maintenance data and the dynamic decision-making of maintenance strategies, and can also realize the visual display of the optimal maintenance strategy in a visual manner. During the implementation of this embodiment, by obtaining the multi-dimensional operation data of the aircraft and inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data, then, according to the health status characterization data, the optimal maintenance strategy corresponding to the health status characterization data is obtained, and then the health status characterization data and the optimal maintenance strategy are visually displayed on the AI prediction dashboard, and at the same time, maintenance auxiliary decision-making suggestions are generated according to the optimal maintenance strategy. This embodiment can realize the dynamic analysis of aircraft maintenance data and the dynamic decision-making of maintenance strategies, and can also realize the visual display of the optimal maintenance strategy in a visual manner, which helps to improve the aircraft maintenance efficiency and is applicable to the application scenarios of aircraft scheduled inspection and preventive maintenance.

[0112] Embodiment 2:

[0113] This embodiment discloses an aircraft maintenance data visualization system for implementing the aircraft maintenance data visualization method in Embodiment 1; as Figure 2 shown, the aircraft maintenance data visualization system includes:

[0114] A data acquisition module, configured to obtain the multi-dimensional operation data of the aircraft;

[0115] A health data extraction module, communicatively connected to the data acquisition module, and configured to input the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data;

[0116] A maintenance strategy generation module, communicatively connected to the health data extraction module, for obtaining an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data;

[0117] A visualization module, communicatively connected to the maintenance strategy generation module, for visually displaying the health status characterization data and the optimal maintenance strategy on an AI prediction dashboard.

[0118] It should be noted that for the working process, working details, and technical effects of the aircraft maintenance data visualization system provided in this Embodiment 2, reference can be made to Embodiment 1, which will not be elaborated here.

[0119] Embodiment 3:

[0120] Based on Embodiment 1 or 2, this embodiment discloses an electronic device, which can be a smart phone, a tablet computer, a notebook computer, or a desktop computer, etc. The electronic device may be referred to as a user terminal, a portable terminal, a desktop terminal, etc. As Figure 3 shown, the electronic device includes:

[0121] A memory for storing computer program instructions; and,

[0122] A processor for executing the computer program instructions to complete the operations of a method for visualizing aircraft maintenance data as described in any one of Embodiment 1.

[0123] Specifically, the processor 301 may include one or more processing cores, such as a 4-core processor or an 8-core processor. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen.

[0124] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include high-speed random access memory, as well as non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage media in the memory 302 is used to store at least one instruction for being executed by the processor 301 to implement the aircraft maintenance data visualization method provided in Embodiment 1 of this application.

[0125] In some embodiments, the terminal may further optionally include: a communication interface 303 and at least one peripheral device. The processor 301, the memory 302, and the communication interface 303 may be connected through a bus or signal lines. Each peripheral device may be connected to the communication interface 303 through a bus, signal lines, or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 304, a display screen 305, and a power supply 306.

[0126] The communication interface 303 may be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 may be implemented on a separate chip or circuit board, and this embodiment does not limit this.

[0127] The radio frequency circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 304 communicates with a communication network and other communication devices through electromagnetic signals.

[0128] The display screen 305 is used to display a UI (User Interface). The UI may include any combination of graphics, text, icons, and videos.

[0129] The power supply 306 is used to supply power to each component in the electronic device.

[0130] Embodiment 4:

[0131] Based on any one of Embodiments 1 to 3, this embodiment discloses a computer program product, including a computer program or instruction, and the computer program or the instruction, when executed by a computer, implements an aircraft maintenance data visualization method as described in any one of Embodiments 1. Wherein, the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0132] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Optionally, they can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or equivalently replace some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for visualizing aircraft maintenance data, characterized in that, Including: Obtain multi-dimensional operation data of the aircraft; Input the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data; Obtain an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data; Display the health status characterization data and the optimal maintenance strategy in a visual manner on an AI prediction dashboard; The health status characterization data includes health status time series feature data, current health status information, current predicted remaining life value of the device, and remaining life prediction sequence; Correspondingly, inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data includes: Input the multi-dimensional operation data into a preset device health perception model to obtain health status time series feature data, and use the feature data at the latest moment in the health status time series feature data as the current health status information of the aircraft; According to the health status time series feature data, obtain the remaining life prediction sequence of the aircraft, and use the remaining life prediction value at the latest moment in the remaining life prediction sequence as the current predicted remaining life value of the aircraft; Obtaining an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data includes: Input the current health status information in the health status characterization data into a preset maintenance strategy optimization module, and calculate the expected return value corresponding to the maintenance actions in the action space based on the value function of multiple state-action pairs formed by the current health status information and the maintenance actions in the action space of the maintenance strategy optimization module; Use the maintenance action in the state-action pair with the maximum expected return value as the optimal maintenance strategy; The maintenance strategy optimization module adopts a deep Q network; correspondingly, the state space of the maintenance strategy optimization module is S={ s i ∣ i =1,2,..., I }, s i is the first i state elements, I is the total number of state elements in the state space, and the action space of the maintenance strategy optimization module is A={ a 1: Repair immediately, a 2: Deferred maintenance, a 3: Ignore}, the reward function of the maintenance strategy optimization module is: ； Wherein, R ( s , a ) is the reward value for performing a maintenance action s in a healthy state a ; Cmaint ( a ) is the direct maintenance cost incurred by performing the maintenance action a ; Crisk ( s , a ) is the loss of failure risk that may be caused by performing the maintenance action s in a healthy state a ; Cavailability ( a ) is the benefit of flight mission availability brought by the maintenance; The value function is: ； Wherein, Q ( s , a ) is the cumulative expected benefit brought by performing a maintenance action s in a healthy state a , α is the preset learning rate, γ is the preset discount factor, s ' is the new healthy state transferred to after performing the action a , a ' is the best action among all possible actions in the new healthy state s ', Q ( s ', a ') is the maximum expected benefit that can be obtained by performing the action a ' in the healthy state of the new healthy state; Correspondingly, the optimal maintenance strategy is: =argmax a Q ( h T , a ); where argmax a is the argmax function that finds the maintenance action that maximizes the value function among all maintenance actions in the action space, Q ( h T , a ) is the cumulative expected benefit brought by performing the maintenance action h T under the current health state information a .

2. The aircraft maintenance data visualization method according to claim 1, wherein The device health perception model adopts a long short-term memory network; correspondingly, in the time-series feature data of the health state, the feature data at time t is as follows: ； Wherein, LSTM() represents the long short-term memory network, x t is the time t of the multi-dimensional operation data, t = 1, 2, ..., T , T is the time index, T is a natural number greater than 1, h t-1 is the characteristic data at the time t -1; In the remaining life prediction sequence, at time t the predicted remaining life value is: ； In the formula, W is a preset weight matrix, b is a preset correction parameter.

3. A method for visualizing aircraft maintenance data according to claim 1, wherein, The method further includes: Generate maintenance assistance decision-making suggestions according to the optimal maintenance strategy; Generate strategy explanation information according to the maintenance assistance decision-making suggestions, and output an artificial confirmation prompt message bound with the maintenance assistance decision-making suggestions and the strategy explanation information; Judge whether an artificial confirmation instruction is received within a specified duration. If so, generate a maintenance task plan according to the maintenance assistance decision-making suggestions, and feedback the execution result after executing the maintenance task plan to the device health perception model for self-learning update of the device health perception model.

4. An aircraft maintenance data visualization system, characterized in that, Including: A data acquisition module for obtaining multi-dimensional operation data of the aircraft; A health data extraction module communicatively connected to the data acquisition module for inputting the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data; A maintenance strategy generation module communicatively connected to the health data extraction module for obtaining an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data; A visualization module communicatively connected to the maintenance strategy generation module for displaying the health status characterization data and the optimal maintenance strategy in a visual manner on an AI prediction dashboard; The health status characterization data includes health status time-series feature data, current health status information, current predicted remaining life value of the device, and predicted remaining life sequence; Correspondingly, input the multi-dimensional operation data into a preset device health perception model to obtain health status characterization data, including: Input the multi-dimensional operation data into a preset device health perception model to obtain health status time-series feature data, and use the feature data at the latest moment in the health status time-series feature data as the current health status information of the aircraft; Based on the health status time-series feature data, obtain the predicted remaining life sequence of the aircraft, and use the predicted remaining life value at the latest moment in the predicted remaining life sequence as the current predicted remaining life value of the aircraft; Obtain an optimal maintenance strategy corresponding to the health status characterization data according to the health status characterization data, including: Input the current health status information in the health status characterization data into a preset maintenance strategy optimization module, and calculate the expected return value corresponding to the maintenance actions in the action space based on the value function of multiple state-action pairs formed by the current health status information and the maintenance actions in the action space of the maintenance strategy optimization module; Use the maintenance action in the state-action pair with the maximum expected return value as the optimal maintenance strategy; The maintenance strategy optimization module adopts a deep Q-network; correspondingly, the state space of the maintenance strategy optimization module is S = { s i | i = 1, 2,..., I}, s i being the i th state element in the state space, I being the total number of state elements in the state space, the action space of the maintenance strategy optimization module is A = { a 1: Immediate maintenance, a 2: Deferred maintenance, a 3: Ignore}, and the reward function of the maintenance strategy optimization module is: ； Wherein, R ( s , a ) is the reward value for performing a maintenance action s in a healthy state a ; Cmaint ( a ) is the direct maintenance cost incurred by performing the maintenance action a ; Crisk ( s , a ) is the loss of failure risk that may be caused by performing the maintenance action s in a healthy state a ; Cavailability ( a ) is the flight mission availability benefit brought by the maintenance; The value function is: ； wherein, Q ( s , a ) is the cumulative expected benefit brought by performing a maintenance action s in a healthy state a ; α is a preset learning rate, γ is a preset discount factor, s ′ is the new healthy state transferred to after performing the action a ; a ′ is the best action among all possible actions in the new healthy state s ′; Q ( s ′, a ′) is the maximum expected benefit that can be obtained by performing the action a ′ in the healthy state of the new healthy state; Correspondingly, the optimal maintenance strategy is: =argmax a Q ( h T , a ); Among them, argmax a is the argmax function that finds the maintenance action that maximizes the value function among all maintenance actions in the action space. Q ( h T , a ) is the cumulative expected benefit brought by performing the maintenance action h T under the current health state information a .

5. An electronic device, characterized in that, Including: A memory for storing computer program instructions; And, A processor for executing the computer program instructions to complete the operations of an aircraft maintenance data visualization method as described in any one of claims 1 to 3.

6. A computer program product, comprising a computer program or instructions, characterized in that, The computer program or the instructions, when executed by a computer, implement an aircraft maintenance data visualization method as described in any one of claims 1 to 3.

Images