Helicopter maintenance task intelligent decision-making method
By constructing a resource and rule base for maintenance task implementation support, and combining multi-objective optimization algorithms and particle swarm optimization algorithms, intelligent decision-making for helicopter maintenance tasks was achieved, solving the problems of long decision-making time, insufficient accuracy, and insufficient flexibility, and realizing efficient, accurate, and flexible execution of maintenance tasks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-09
- Publication Date
- 2026-03-27
AI Technical Summary
Helicopter maintenance tasks are time-consuming, lack accuracy and flexibility, are heavily influenced by human decision-making, and are difficult to respond quickly to emergencies.
By employing structured processing and multi-objective optimization algorithms, a resource library and rule library for maintenance task implementation are constructed. Combining multi-objective optimization algorithms and particle swarm optimization algorithms, decision data is automatically collected and analyzed to generate intelligent maintenance task decision schemes.
It shortens decision-making time, improves decision-making accuracy and flexibility, and can automatically adjust decision-making plans to cope with emergencies, ensuring the efficient execution of maintenance tasks.
Smart Images
Figure CN121745899A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of helicopter comprehensive support, and particularly relates to a helicopter maintenance task intelligent decision method. BACKGROUND
[0002] Helicopter maintenance task decision is the process in which the ground crew leader (usually a mechanic) decides which planned maintenance tasks to perform, whether to perform troubleshooting work according to the support site conditions, support resource availability, personnel quantity and technical ability, and feeds back the maintenance personnel and support resource coordination and scheduling requirements to the superior or relevant department.
[0003] Maintenance task decision, on the one hand, requires the ground crew leader to obtain helicopter fault information from helicopter PFL / MFL / ground inspection results / crew report / flight parameter analysis report and other information, to determine the troubleshooting idea based on personal experience or query technical data, and to determine whether the troubleshooting work can be performed according to the current and forecast weather conditions, support resource supply situation, helicopter flight training / task plan, support site conditions, and to feed back the support personnel, support resource coordination and scheduling requirements to the superior according to the situation; on the other hand, it is necessary to analyze the necessary conditions for executing the maintenance task according to the planned maintenance task instructions issued by the superior, such as life part replacement, modification, special inspection, mechanical day, technical report implementation, periodic maintenance work, etc., to determine whether the planned maintenance can be performed in combination with the current and forecast weather conditions, support resource supply situation, helicopter flight training / task plan, support site conditions, and to feed back the support personnel, support resource coordination and scheduling requirements to the superior.
[0004] Helicopter maintenance task decision requires the ground crew leader to query and analyze a large amount of information, and to have sufficient decision-making experience, so as to make the decision-making result fully guarantee the accuracy and efficiency of maintenance task execution.
[0005] The existing helicopter maintenance task decision is made by manually collecting and summarizing meteorological, support resource, flight training / task plan, support site and other information, comprehensively analyzing and judging by the relevant leaders of maintenance work organization and implementation, and manually making the decision, which is then executed by the maintenance personnel.
[0006] The main technical problems existing in the prior art are:
[0007] (1) The time for decision-making is long. Due to the large amount of information and many factors affecting the maintenance task decision, involving multiple departments and fields, the content of collection, analysis and judgment is large, although the maintenance task decision maker is required to have corresponding professional and leadership, it still needs a long time to form a correct decision. At present, it usually needs to be advanced for about half a day to a day, and a meeting form is adopted to form the maintenance task decision scheme of this time.
[0008] (2) Decision accuracy is lacking. Since the existing maintenance task decision relies entirely on manual implementation, the maintenance task decision scheme formed is subjective and the accuracy needs to be improved due to the influence of factors such as the individual ability level of the decision maker, the timeliness and integrity of information.
[0009] (3) The flexibility of the decision is not enough. Since the maintenance task is affected by many factors, once real-time conditions such as flight task adjustment, sudden weather conditions, and failure of support resource scheduling occur, the original decision scheme needs to be adjusted in time, and there is a lack of flexibility. SUMMARY
[0010] The purpose of the present application is to realize the goal of changing the maintenance task decision from relying on manual to intelligent decision.
[0011] The present application provides a helicopter maintenance task intelligent decision method, which comprises:
[0012] Step 1: According to the helicopter maintenance technical data, the necessary support resources required by the maintenance task and the necessary conditions required by the maintenance task implementation are extracted and structured, and a maintenance task implementation support resource library and a rule library are constructed;
[0013] Step 2: According to the helicopter maintenance technical data and historical maintenance record data, the time required for maintenance task implementation, the type and quantity of support resources, the technical level and quantity of maintenance personnel, and the weight are calculated to generate a maintenance task influence factor database;
[0014] Step 3: Construct a query-rule matching algorithm, dynamically read the state data of the relevant business information system, query the dynamically read state data of the relevant business information system, and determine whether the maintenance task can be executed;
[0015] Step 4: The helicopter maintenance task intelligent decision model based on the multi-objective optimization algorithm reads the maintenance task information and the maintenance task execution influence factor data for the executable maintenance task determined in step 3, and generates a helicopter maintenance task decision scheme.
[0016] Preferably, step 1 specifically comprises:
[0017] On the one hand, the helicopter maintenance technical data is extracted, including the demand for spare parts, the demand for tools and equipment, the demand for station resources, and the demand for maintenance personnel, and is structured and standardized to be stored in the maintenance task implementation support resource library; on the other hand, the necessary conditions required for maintenance task execution are extracted, the relevant information sources are determined, and the relevant criteria are designed, and are stored in the maintenance task implementation rule library.
[0018] Preferably, the helicopter maintenance technical data refers to the maintenance single card or the reference maintenance manual and maintenance procedure;
[0019] Structured processing refers to the process of identifying data information in paper documents, document photos, audio recordings, video recordings, and PDF files in technical materials through manual input or OCR recognition methods, and storing the data in the relevant database according to the provided data input template;
[0020] Standardization processing refers to constructing a semantically similar thesaurus and data dictionary for cases with the same semantics but different textual descriptions, automatically identifying semantically detailed words, and merging them into standard words;
[0021] The necessary conditions for performing maintenance tasks specifically refer to the necessary conditions for carrying out maintenance tasks, in addition to the support resources required for the maintenance tasks.
[0022] Preferably, step 2 specifically includes:
[0023] A combination of manual and algorithmic extraction methods was used to extract the influencing factors of maintenance task implementation from maintenance technical data; the influencing factors of maintenance task implementation were read from historical maintenance record data in the support information system; the weights of the influencing factors in the technical data and the influencing factors in the historical maintenance record data were set; and the final values of the influencing factors of maintenance task implementation were calculated.
[0024] Preferably, the influencing factors for the implementation of maintenance tasks are the duration required for the implementation of maintenance tasks, the type and quantity of support resources, and the technical level and number of maintenance personnel;
[0025] The purpose of setting the weights of the influencing factors is to use historical maintenance task implementation data to compensate for the lack of specificity in the data specified in the technical documents; and to use the data specified in the technical documents to compensate for the insufficient sample size and breadth of historical maintenance record data.
[0026] Preferably, step 3 specifically includes:
[0027] Constructing a query-rule matching algorithm involves the following two steps:
[0028] Query: Using search statements, retrieve the necessary conditions for performing a maintenance task from the database based on the task name;
[0029] Rule matching: Dynamically reads status data from relevant business information systems, performs rule matching based on constraints, and outputs the matching results.
[0030] Preferably, step 4 specifically includes:
[0031] The maintenance task decision-making based on multi-objective optimization adopts the particle swarm optimization algorithm. A massless particle is designed to simulate birds in a flock. The particle has only two attributes: velocity and position. Velocity represents the speed of movement, and position represents the direction of movement. Each particle searches for the optimal solution in the search space independently and records it as the current individual extreme value. The individual extreme value is shared with other particles in the entire particle swarm. The optimal individual extreme value is found as the current global optimal solution for the entire particle swarm. All particles in the particle swarm adjust their velocity and position according to their current individual extreme value and the current global optimal solution shared by the entire particle swarm.
[0032] Preferably, a massless particle is used as an influencing factor for the implementation of maintenance tasks.
[0033] This application has the following technical effects:
[0034] (1) Reduce decision-making time
[0035] Based on advanced technologies such as automatic acquisition, analysis, and processing of multi-source data and artificial intelligence algorithms, this invention can effectively reduce the decision-making time for helicopter maintenance tasks and improve the informatization and intelligence level of maintenance task decision-making.
[0036] (2) Improve decision-making accuracy
[0037] This invention ensures the integrity, timeliness, and objectivity of key aspects of maintenance task decision-making, such as data, processes, and nodes, while avoiding the subjectivity and arbitrariness present in existing maintenance task decision-making technologies, thus improving decision-making accuracy.
[0038] (3) Improve decision-making adjustability
[0039] Once a decision-making plan is formed, if unforeseen circumstances arise during execution, such as flight mission adjustments, sudden weather conditions, or resource allocation issues, information about these unforeseen circumstances can be input to automatically adjust the maintenance task decision-making plan, ensuring the maintenance task continues. Attached Figure Description
[0040] Figure 1 This is a flowchart of the intelligent maintenance task decision-making process of the present invention;
[0041] Figure 2 This is a flowchart of the intelligent decision-making algorithm for helicopter maintenance tasks based on multi-objective optimization, as described in this invention. Detailed Implementation
[0042] Please see Figure 1 and Figure 2 The present invention adopts the following technical solution:
[0043] The system automatically collects various support resource information, technical data, meteorological conditions, and maintenance personnel information required for maintenance task execution. Using an autonomous decision-making algorithm, it generates intelligent maintenance task decision plans, including executable maintenance tasks, support resource requirements, and maintenance personnel requirements. It also provides feedback on reasons for unexecutable maintenance tasks, allowing decision-makers to adjust and confirm the maintenance task decision plans. The system automatically generates support resource and personnel requirements and scheduling plans, which are then fed back to the agile support information system terminals of higher-level leaders, aircraft parts warehouses, and tool rooms for approval and allocation of support resources and personnel. This achieves intelligent maintenance task decision-making. (See process details below.) Figure 1 .
[0044] Through business process analysis, maintenance task decision-making primarily addresses the question of whether or not to execute a maintenance task. The execution of a maintenance task depends mainly on the following conditions:
[0045] Impact of the malfunction: The minimum release list can be used to determine whether the malfunction can be released. If it cannot be released, it must be performed. If it can be released under conditions, the release conditions and associated maintenance tasks need to be determined.
[0046] Helicopter health status: The worse the health status, the more urgent the maintenance mission. The health status output by the ground health management system is a quantifiable indicator.
[0047] Flight mission plan: In principle, all maintenance missions must be carried out if the execution of maintenance missions does not affect flight missions.
[0048] Fault-related maintenance methods: For the same fault, there may be multiple possible maintenance methods, such as repair, debugging, component replacement, component swapping, etc. Choosing the appropriate maintenance method is of great significance for the execution of maintenance tasks in the current scenario;
[0049] Status of support resources (including personnel) required for maintenance: Under the same conditions, the fewer support resources (including personnel) used, the better. If the support resources (including personnel) cannot meet the requirements, the maintenance task cannot be performed.
[0050] Repair method duration: Different repair methods correspond to different repair task durations. Under the same conditions, the shorter the duration, the better.
[0051] Environmental factors: Some repair methods have specific requirements for weather, repair site, etc. Repair methods that do not meet the environmental conditions cannot be selected.
[0052] This invention should be able to intelligently output maintenance task decision schemes, the contents of which are shown in Table 1.
[0053] Table 1 Maintenance Task Decision-Making Scheme
[0054]
[0055]
[0056] Based on the above analysis, intelligent maintenance task decision-making is a multi-objective optimization problem; therefore, a task decision-making algorithm based on multi-objective optimization can be adopted. The intelligent maintenance task decision-making algorithm consists of two parts: rule-based decision-making and dynamic programming-based decision optimization. Rule-based decision-making is used to calculate the necessary conditions for executing the maintenance task; the dynamic programming-based optimization algorithm is used to dynamically optimize and adjust the maintenance task decision-making scheme. Rule-based decision-making is the foundation of maintenance decision-making, while decision-making based on multi-objective optimization algorithms is the key to achieving intelligent and accurate decision-making.
[0057] In other embodiments of this application, a helicopter maintenance task intelligent decision-making method provided in this application includes the following steps:
[0058] Step 1: Based on helicopter maintenance technical data, extract the support resources required for the maintenance mission and the necessary conditions for the implementation of the maintenance mission, and perform structured processing to build a support resource library and rule library for the implementation of the maintenance mission;
[0059] Step 2: Based on helicopter maintenance technical data and historical maintenance records, extract the time required for the implementation of maintenance tasks, the type and quantity of support resources, and the technical level and number of maintenance personnel, and calculate them according to weights to generate a database of maintenance task influencing factors;
[0060] Step 3: Construct a query-rule matching algorithm to dynamically read the status data of relevant business information systems, and query the status data of relevant business information systems to determine whether the maintenance task can be executed;
[0061] Step 4: Intelligent decision-making model for helicopter maintenance tasks based on multi-objective optimization algorithm. For the executable maintenance tasks determined in Step 3, read the maintenance task information and maintenance task execution influencing factor data, and generate a helicopter maintenance task decision scheme.
[0062] Step 1 specifically includes:
[0063] On the one hand, the requirements for aviation materials, tools and equipment, airfield resources, and maintenance personnel in helicopter maintenance technical data are extracted, and then structured and standardized and stored in the maintenance task implementation support resource library. On the other hand, the necessary conditions required for the execution of maintenance tasks are extracted, the relevant information sources are determined, and relevant criteria are designed and stored in the maintenance task implementation rule library.
[0064] Among them, helicopter maintenance technical data refers to maintenance slips or reference maintenance manuals, maintenance procedures, etc.
[0065] Structured processing refers to the process of identifying data information in technical documents, such as paper documents, document photos, audio recordings, video recordings, and PDF files, through manual input or OCR recognition, and storing the data in the relevant database according to the provided data entry template.
[0066] Standardization processing refers to constructing a semantically similar thesaurus and data dictionary for cases with the same meaning but different textual descriptions, automatically identifying semantically detailed words, and merging them into standard words.
[0067] The necessary conditions for performing maintenance tasks specifically refer to the necessary conditions for carrying out maintenance tasks other than the resources required to support the maintenance task. For example, replacing the fuel pump in the fuel system is not allowed to be done for electrical or oxygen-related maintenance tasks.
[0068] Step 2 specifically includes:
[0069] A combination of manual and algorithmic extraction methods was used to extract the influencing factors of maintenance task implementation from maintenance technical data; the influencing factors of maintenance task implementation were read from historical maintenance record data in the support information system; the weights of the influencing factors in the technical data and the influencing factors in the historical maintenance record data were set; and the final values of the influencing factors of maintenance task implementation were calculated.
[0070] Among them, the influencing factors for the implementation of maintenance tasks are the duration required for the implementation of maintenance tasks, the type and quantity of support resources, and the technical level and number of maintenance personnel.
[0071] The purpose of setting weights for influencing factors is to compensate for the lack of specificity in the data specified in the technical documentation by using historical maintenance task implementation data (for example, differences in maintenance capabilities and support resource allocation levels among different helicopter user units, which affect the final efficiency and quality of maintenance task implementation); and to compensate for the insufficient sample size and breadth of historical maintenance record data by using the data specified in the technical documentation (for example, in the early stages of helicopter use, the sample size of maintenance records is small and cannot broadly represent the true level of maintenance task implementation).
[0072] Step 3 specifically includes:
[0073] Constructing a query-rule matching algorithm involves the following two steps:
[0074] Query
[0075] Using a search query, based on the maintenance task name (code), retrieve the necessary conditions for performing the maintenance task from the database, such as weather restrictions, number of personnel, personnel technical requirements, tool and equipment requirements, and aircraft material requirements. The query is as follows:
[0076] SELECT * FROM taskDB WHERE task = task
[0077] taskDB is the name of the maintenance task table, and the task is the name (code) of the task being queried.
[0078] Rule matching
[0079] Dynamically read status data from relevant business information systems, such as weather status, personnel status, tool and equipment status, and aircraft material status. Based on constraints, perform rule matching and output the matching results. The matching results include executable maintenance task names, non-executable maintenance task names, and the reason for non-executability (rules with a false matching result). The statement is as follows:
[0080] SELECT * FROM statusDB / / Retrieve relevant business information from the status database
[0081] If the required weather conditions for the task are equal to the actual weather conditions:
[0082] weather=True
[0083] else:
[0084] weather=False
[0085] ...
[0086] Check each condition one by one. If all conditions are True, the maintenance task can be executed; if any condition is False, the task cannot be executed.
[0087] Step 4 specifically involves...
[0088] The maintenance task decision-making based on multi-objective optimization employs a particle swarm optimization (PSO) algorithm. A massless particle (i.e., the influencing factor for maintenance task implementation) is designed to simulate birds in a flock. The particle has only two attributes: velocity and position. Velocity represents the speed of movement, and position represents the direction of movement. Each particle independently searches for the optimal solution in the search space and records it as its current individual extreme value. This individual extreme value is shared with all other particles in the swarm. The optimal individual extreme value is then used as the current global optimal solution for the entire swarm. All particles in the swarm adjust their velocity and position based on their own current individual extreme value and the shared global optimal solution.
[0089] The flowchart of the particle swarm optimization algorithm is as follows: Figure 2 As shown.
[0090] initialization
[0091] First, a maximum speed range is set to prevent the search speed from exceeding this range. The location information constitutes the entire search space; we randomly initialize speeds and positions within both the speed range and the search space. The population size is then set.
[0092] Suppose there are m particles in a D-dimensional space, and the attribute information of each particle is as follows:
[0093] Position of particle i: xi = (xi1, xi2, ..., xiD)
[0094] The velocity of particle i is: vi = (vi1, vi2, ..., viD), 1 ≤ i ≤ m, 1 ≤ d ≤ D.
[0095] The best historical position experienced by particle i: pi = (pi1, pi2, ..., piD)
[0096] The best positions experienced by all particles within the group (or region):
[0097] pg = (pg1, pg2, ..., pgD)
[0098] Generally speaking, the position and velocity of a particle are determined in a continuous space of real numbers.
[0099] Individual extreme values and global optimal solutions
[0100] The individual extreme value is the best historical position information found for each particle. From these individual historical best solutions, a global optimal solution is found and compared with the historical best solution. The best one is selected as the current historical optimal solution.
[0101] Formulas for updating speed and position
[0102] The speed update formula is:
[0103] V id =ωV id +C1random(0,1)(P id -X id )+C2random(0,1)(P gd -X id )
[0104] The position update formula is:
[0105] X id =X id +V id
[0106] Where ω is called the inertia factor, and C1 and C2 are called acceleration constants, typically taken as 2. Random(0,1) represents a random number in the interval (0,1). Pid P represents the d-th dimension of the individual extreme value of the i-th variable. gd Let d represent the d-th dimension of the global optimal solution.
[0107] Termination conditions
[0108] There are two termination conditions to choose from:
[0109] First, there are two maximum number of generations; second, the algorithm stops when the deviation between two adjacent generations falls within a specified range. Specific termination conditions need to be determined during algorithm verification after importing actual historical maintenance records and technical maintenance data.
[0110] This invention employs intelligent maintenance task decision-making technology, which can automatically collect, analyze, and process decision data, and intelligently generate decision-making schemes. It realizes the informatization and intelligentization of maintenance task decision-making, effectively avoids the subjectivity of human experience, greatly reduces the decision-making time for maintenance tasks, and plays an important role in the high-quality execution of maintenance tasks.
[0111] (1) Reduce decision-making time
[0112] Based on advanced technologies such as automatic acquisition, analysis, and processing of multi-source data and artificial intelligence algorithms, this invention can effectively reduce the decision-making time for helicopter maintenance tasks and improve the informatization and intelligence level of maintenance task decision-making.
[0113] (2) Improve decision-making accuracy
[0114] This invention ensures the integrity, timeliness, and objectivity of key aspects of maintenance task decision-making, such as data, processes, and nodes, while avoiding the subjectivity and arbitrariness present in existing maintenance task decision-making technologies, thus improving decision-making accuracy.
[0115] (3) Improve decision-making adjustability
[0116] Once a decision-making plan is formed, if unforeseen circumstances arise during execution, such as flight mission adjustments, sudden weather conditions, or resource allocation issues, information about these unforeseen circumstances can be input to automatically adjust the maintenance task decision-making plan, ensuring the maintenance task continues.
Claims
1. A method for intelligent decision-making in helicopter maintenance tasks, characterized in that, The method includes: Step 1: Based on helicopter maintenance technical data, extract the support resources required for the maintenance mission and the necessary conditions for the implementation of the maintenance mission, and perform structured processing to build a support resource library and rule library for the implementation of the maintenance mission; Step 2: Based on helicopter maintenance technical data and historical maintenance records, extract the time required for the implementation of maintenance tasks, the type and quantity of support resources, and the technical level and number of maintenance personnel, and calculate them according to weights to generate a database of maintenance task influencing factors; Step 3: Construct a query-rule matching algorithm to dynamically read the status data of relevant business information systems, and query the status data of relevant business information systems to determine whether the maintenance task can be executed; Step 4: Intelligent decision-making model for helicopter maintenance tasks based on multi-objective optimization algorithm. For the executable maintenance tasks determined in Step 3, read the maintenance task information and maintenance task execution influencing factor data, and generate a helicopter maintenance task decision scheme.
2. The method according to claim 1, characterized in that, Step 1, specifically: On the one hand, the requirements for aviation materials, tools and equipment, airfield resources, and maintenance personnel are extracted from helicopter maintenance technical data, and then structured and standardized and stored in the maintenance task implementation support resource library; on the other hand, the necessary conditions required for the execution of maintenance tasks are extracted, the relevant information sources are determined, and relevant criteria are designed and stored in the maintenance task implementation rule library.
3. The method according to claim 2, characterized in that, Helicopter maintenance technical data refers to maintenance slips or reference maintenance manuals and procedures; Structured processing refers to the process of identifying data information in paper documents, document photos, audio recordings, video recordings, and PDF files in technical materials through manual input or OCR recognition methods, and storing the data in the relevant database according to the provided data input template; Standardization processing refers to constructing a semantically similar thesaurus and data dictionary for cases with the same semantics but different textual descriptions, automatically identifying semantically detailed words, and merging them into standard words; The necessary conditions for performing maintenance tasks specifically refer to the necessary conditions for carrying out maintenance tasks, in addition to the support resources required for the maintenance tasks.
4. The method according to claim 1, characterized in that, Step 2, specifically: A combination of manual and algorithmic extraction methods was used to extract the influencing factors of maintenance task implementation from maintenance technical data; the influencing factors of maintenance task implementation were read from historical maintenance record data in the support information system; the weights of the influencing factors in the technical data and the influencing factors in the historical maintenance record data were set; and the final values of the influencing factors of maintenance task implementation were calculated.
5. The method according to claim 4, characterized in that, The influencing factors for the implementation of maintenance tasks are the duration required for the maintenance task, the type and quantity of support resources, and the technical level and number of maintenance personnel. The purpose of setting the weights of the influencing factors is to use historical maintenance task implementation data to compensate for the lack of specificity in the data specified in the technical documents; and to use the data specified in the technical documents to compensate for the insufficient sample size and breadth of historical maintenance record data.
6. The method according to claim 1, characterized in that, Step 3, specifically: Constructing a query-rule matching algorithm involves the following two steps: Query: Using search statements, retrieve the necessary conditions for performing a maintenance task from the database based on the task name; Rule matching: Dynamically reads status data from relevant business information systems, performs rule matching based on constraints, and outputs the matching results.
7. The method according to claim 1, characterized in that, Step 4, specifically: The maintenance task decision-making based on multi-objective optimization adopts the particle swarm optimization algorithm. A massless particle is designed to simulate birds in a flock. The particle has only two attributes: velocity and position. Velocity represents the speed of movement, and position represents the direction of movement. Each particle searches for the optimal solution in the search space independently and records it as the current individual extreme value. The individual extreme value is shared with other particles in the entire particle swarm. The optimal individual extreme value is found as the current global optimal solution for the entire particle swarm. All particles in the particle swarm adjust their velocity and position according to their current individual extreme value and the current global optimal solution shared by the entire particle swarm.
8. The method according to claim 7, characterized in that, A massless particle is an influencing factor for the implementation of maintenance tasks.