High-altitude test task scheduling method based on clustering algorithm and multi-objective optimization

Abstract

The application belongs to the technical field of intelligent manufacturing and industrial scheduling, and provides a high-altitude table multi-test task scheduling method based on a clustering algorithm and multi-objective optimization, which comprises the following steps: automatically calculating the exclusive resource demand of each working condition point in a plurality of aero-engine test tasks based on a multi-parameter coupling model; for each task, a plurality of work packages are divided by using a clustering algorithm based on multi-dimensional features containing exclusive resource demand; a multi-objective model of multi-optimization objectives is constructed, and all work packages are jointly sorted globally, wherein the optimization objectives include equipment utilization; finally, under the constraints of resource capacity, test cabin mutual exclusion and task continuity, available test devices and continuous time windows are allocated to each work package to generate an initial scheduling plan. The application effectively improves scheduling efficiency and resource utilization, enhances plan robustness, and is suitable for high-complexity aero-engine test scheduling scenarios.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing and industrial scheduling technology, and relates to a multi-test task scheduling method based on clustering algorithm and multi-objective optimization. This method can realize the collaborative scheduling of high-altitude simulation test bench test tasks under high complexity, multi-resource coupling and strong constraints, and is particularly suitable for test scheduling management in high-end manufacturing fields such as aviation, energy and automobiles. Background Technology

[0002] In modern high-end manufacturing, especially in industries such as aviation, aerospace, energy, and automotive, testing tasks are a crucial part of product development and quality verification, characterized by high complexity and uncertainty. Testing tasks typically involve the coordinated operation of various equipment, resources, and process conditions, with complex dependencies and resource competition among tasks.

[0003] Domestic and international scholars have proposed various optimization algorithms in the field of production scheduling, including intelligent optimization algorithms such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Simulated Annealing (SA), as well as multi-objective optimization algorithms such as NSGA-II and MOEA / D. These algorithms are widely used in production scheduling, resource allocation, and task planning problems, but they are mainly aimed at batch production or assembly line operations and have not fully considered the characteristics of experimental tasks, such as multi-resource coupling, process complexity, and diverse priorities.

[0004] Currently, most enterprises and research institutions in the field of aero-engine testing still use traditional manual scheduling methods for scheduling test missions. However, with the continuous increase in the number of aero-engine test missions and test benches / equipment, the traditional manual scheduling method has gradually revealed the following problems:

[0005] 1. Manual scheduling requires a lot of time for task decomposition, resource matching, and time arrangement, resulting in low scheduling efficiency and difficulty in responding quickly to task changes;

[0006] 2. Lack of real-time monitoring and dynamic scheduling of equipment and resource status leads to resource idleness or conflicts, reducing resource utilization;

[0007] 3. Production scheduling plans are easily affected by human factors and lack unified standards, resulting in inconsistent scheduling results and low stability;

[0008] 4. In the face of unexpected situations such as last-minute order insertions, equipment failures, and task delays, manual production scheduling is difficult to adjust quickly and lacks flexibility;

[0009] 5. Existing production scheduling methods mostly focus on "completing tasks" and neglect the need for multi-objective optimization such as equipment utilization, task completion rate, and plan robustness.

[0010] Specifically, in aero-engine testing scenarios, a typical test mission usually involves dozens of operating points at different altitudes and Mach numbers, requiring a large amount of resources such as air supply, air extraction, fuel, and cooling water. Currently, production scheduling mainly relies on manual experience, which presents the following problems:

[0011] a. The operating points of the same task are treated in isolation without considering their process continuity and resource similarity, resulting in frequent start-ups and shutdowns and large resource switching losses.

[0012] b. Similar operating conditions (such as similar height / temperature combinations) in different tasks are not coordinated, causing auxiliary systems (such as air supply units, air extraction units, heaters, air wave machines, etc.) to start and stop repeatedly, reducing equipment life;

[0013] c. The production scheduling objective is singular, focusing only on the task completion time and ignoring multi-dimensional objectives such as equipment utilization, risk control, and plan robustness.

[0014] d. When faced with equipment failure or temporary order insertion, a global reordering is often required, which is inefficient and can easily lead to a chain of delays.

[0015] Although some studies have attempted to introduce genetic algorithms or rule engines for production scheduling, they have not resolved the contradiction between task integrity and global resource optimization. For example, if cross-task clustering is performed, the process logic of a single experimental task will be disrupted; if only intra-task scheduling is performed, resource reuse cannot be achieved. Summary of the Invention

[0016] To address the significant shortcomings of existing technologies in scheduling test missions in aero-engine testing scenarios, this invention discloses a high-altitude multi-test mission scheduling method based on clustering algorithms and multi-objective optimization. This method, through a two-layer architecture of intra-mission clustering and cross-mission similarity scheduling, can significantly improve resource utilization, reduce switching losses, and enhance the robustness of the scheduling plan to disturbances while ensuring the process integrity of each test mission.

[0017] Specifically, the method includes the following steps:

[0018] S1. Obtain multiple aero-engine test missions, each test mission contains multiple test operating points, and automatically calculate the exclusive resource requirements of each test operating point based on a preset multi-parameter coupled calculation model.

[0019] S2. For each test task, based on the multidimensional features of the test conditions, a clustering algorithm is used to divide it into several work packages, wherein the multidimensional features include specific resource requirements.

[0020] S3. Construct a multi-objective optimization model with task priority, total test duration, equipment utilization, task completion rate and production scheduling robustness as optimization objectives. Perform joint sorting of all test task work packages in the global scope. The optimization of equipment utilization is achieved by arranging work packages with similar resource demand patterns in temporal or spatial proximity.

[0021] S4. Under the constraints of resource capacity, mutual exclusion of test chambers and task continuity, allocate an available tester and a continuous time window that meets the execution duration requirements for each work package according to the joint sorting results, and generate an initial production schedule.

[0022] Furthermore, in step S1 above, the specific resource requirements for each test condition point are automatically calculated based on a preset multi-parameter coupled calculation model, including:

[0023] S11. Obtain the parameters of the test conditions included in each test task, including Mach number, altitude, air flow rate, intake air temperature, exhaust air temperature, exhaust pressure, and fuel flow rate;

[0024] S12. A multi-parameter coupled calculation model is established by integrating the standard atmospheric model, gas dynamics function, flow conservation equation and heat balance formula. Based on the multi-parameter coupled calculation model, the required gas supply, gas extraction, fuel consumption and cooling water consumption for each test operating point are automatically derived according to the parameters to form a specific resource requirement.

[0025] Furthermore, in step S2 above, for each experimental task, based on the multidimensional features of the experimental conditions, a clustering algorithm is used to divide it into several working packages, including:

[0026] S21. For each test task, construct a feature vector for its test operating point. The feature vector includes process requirement type, dedicated resource requirement, test type, risk level and task priority.

[0027] S22. Within this test task, a clustering algorithm is used to measure the similarity of all test conditions. With the goal of minimizing the Euclidean distance, all test conditions of this test task are divided into several work packages.

[0028] S23. For each test condition point in a work package, determine the execution order according to process logic or resource continuity logic, and ensure that the total execution time of the work package does not exceed the preset upper limit of the single continuous test time.

[0029] Furthermore, the clustering algorithm is either the K-means algorithm or the DBSCAN algorithm, and the similarity metric is calculated using weighted Euclidean distance; wherein, when calculating the similarity metric, the weights are dynamically configured based on the scarcity of resources or scheduling sensitivity in the high-altitude station system.

[0030] Furthermore, in step S3 above, the joint sorting of the work packages of all the experimental tasks on a global scale includes:

[0031] S31. Use work packages as the basic unit of scheduling to form a set of tasks to be scheduled.

[0032] S32. Define multiple optimization objectives, including minimizing the total trial duration, maximizing equipment utilization, minimizing risk, minimizing the probability of task delay, maximizing the task completion rate, and improving the robustness of the production schedule to disturbances; among them, the equipment utilization objective explicitly incorporates the measurement results of the similarity of resource requirements of work packages as an optimization factor.

[0033] S33. Construct a weighted comprehensive objective function based on multiple optimization objectives, normalize each optimization objective, and dynamically adjust the weight of each optimization objective according to any one or more of the following: real-time task priority, resource stress, or historical disturbance frequency.

[0034] S34. Use a multi-objective evolutionary algorithm to perform a global joint sort of all working packages, and solve the weighted synthesis objective function to generate a Pareto optimal solution set.

[0035] Furthermore, the multi-objective evolutionary algorithm is the NSGA-II algorithm, the MOEA / D algorithm, or a combination of both, and the weight coefficients of each objective are adjusted online according to real-time task priority, resource stress, or historical perturbation frequency.

[0036] Furthermore, in step S4 above, generating an initial production schedule includes:

[0037] S41. Establish constraint models for resource capacity constraints, test module mutual exclusion constraints, and mission continuity constraints;

[0038] S42. Based on the heuristic scheduling algorithm, each work package is sequentially assigned to an available tester according to the joint sorting result. During the assignment process, if the current work package is highly similar to other assigned test task work packages in terms of air supply, air extraction, fuel consumption, and cooling water consumption, it is given priority to arrange them in adjacent time periods of the same tester to reuse the same air supply unit, air extraction unit, fuel pump, and cooling water unit, thereby reducing the number of auxiliary system switching and energy consumption.

[0039] S43. When all constraints are satisfied through the conflict detection and backtracking mechanism, the initial production schedule is output. The initial production schedule includes a task Gantt chart, a resource occupancy matrix, and a risk warning list.

[0040] Furthermore, the constraints described in step S43 include:

[0041] At any given time, the cumulative demand for each resource does not exceed its maximum available capacity;

[0042] Testers belonging to the same mutually exclusive set cannot be used simultaneously at the same time.

[0043] All test conditions within the same work package must be executed continuously without interruption or interleaving with other test tasks.

[0044] In an improved embodiment of the high-altitude platform multi-test mission scheduling method based on clustering algorithms and multi-objective optimization, the method further includes:

[0045] S5. Based on user needs, manually modify any one or more of the following: the work package division result, the joint sorting result, and the initial production schedule.

[0046] In an improved embodiment of the high-altitude platform multi-test mission scheduling method based on clustering algorithms and multi-objective optimization, the method further includes:

[0047] S6. When equipment failure, temporary order insertion, or schedule delay occurs, the initial production schedule will be partially rescheduled, specifically including:

[0048] S61. Identify the affected work packages and freeze the production schedules of unaffected test tasks;

[0049] S62. For affected work packages, reschedule them within the idle time window released by the freeze operation. The scheduling objective is to minimize the weighted sum of the total task delay time and the adjustment range of the production schedule. The weight coefficient is dynamically set according to the urgency of the task and the system stability requirements.

[0050] This invention proposes a method that establishes a two-layer intelligent scheduling mechanism combining intra-task clustering and cross-task multi-objective collaborative scheduling. First, within each aero-engine test mission, based on the process attributes and resource requirements of the operating point, a clustering algorithm automatically generates highly cohesive work packages to ensure the continuity and integrity of mission execution. Second, at the global level, all work packages are used as scheduling units to construct a multi-objective optimization model that integrates task priority, equipment utilization, completion rate, and plan robustness. The similarity of resource requirements between work packages is explicitly introduced as an optimization factor to achieve efficient resource reuse and proximity arrangement across missions. Simultaneously, the system supports manual intervention and local rescheduling to ensure rapid recovery to stable operation under dynamic disturbances.

[0051] Compared with the prior art, the method of the present invention has the following beneficial effects:

[0052] 1. Significantly improves production scheduling efficiency: The clustering algorithm automatically groups test conditions and generates work packages, greatly reducing the workload of manual task decomposition and scheduling design; according to actual tests at an aero-engine test base, the human participation rate is less than 20%, and the overall production scheduling efficiency is improved by more than 90%.

[0053] 2. Significantly improves resource utilization: Under a multi-objective optimization framework, combining a resource matching model with a similarity-guided proximity arrangement strategy, it achieves efficient allocation and reuse of key resources such as gas supply, extraction, fuel oil, and cooling water; actual test data shows that the overall resource utilization rate is improved by more than 10%, and the number of auxiliary system start-ups and shutdowns is reduced by 30%.

[0054] 3. Enhance the robustness of production scheduling: Explicitly introduce "planning robustness" indicators (such as sensitivity to disturbances, buffer time distribution, etc.) into the optimization objectives to make the generated production scheduling plan more resistant to interference and reduce the risk of cascading delays caused by minor disturbances.

[0055] 4. The system has multiple built-in rescheduling trigger mechanisms to respond to scenarios such as temporary order insertion, equipment failure, task delays and manual adjustments. It adopts a freeze-rescheduling strategy to reschedule only the affected tasks locally, which improves the rescheduling speed by more than 50% and ensures the stability of the overall plan.

[0056] 5. This solution fully considers the characteristics of strong coupling of multiple resources, complex process logic, and diverse task priorities. It has been successfully deployed at a national-level aero-engine test base and can support the collaborative production of multiple medium and large test vehicles. It has good engineering scalability and industry universality.

[0057] 6. The method of the present invention is not only applicable to high-altitude simulation test benches for aero-engines, but can also be extended to high-complexity multi-mission test scheduling scenarios for gas turbines, aerospace propulsion systems, or new energy power plants. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a flowchart of the high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to the present invention;

[0060] Figure 2 The main logic flowchart for scheduling multiple test missions on high-altitude test platforms;

[0061] Figure 3 This is a flowchart illustrating window scanning using a heuristic algorithm.

[0062] Figure 4 The logical flowchart for production rescheduling;

[0063] Figure 5 This is a schematic diagram illustrating the sorting and clustering of test operating points. Detailed Implementation

[0064] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0065] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features of the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0066] This invention discloses a high-altitude test mission scheduling method based on clustering algorithms and multi-objective optimization. This method, through a two-layer architecture of intra-mission clustering and cross-mission similarity scheduling, can significantly improve resource utilization, reduce switching losses, and enhance the robustness of the scheduling plan to disturbances while ensuring the process integrity of each test mission.

[0067] Specifically, see Figure 1 and Figure 2As shown, the method includes the following steps:

[0068] S1. Obtain multiple aero-engine test missions, each test mission contains multiple test operating points, and automatically calculate the exclusive resource requirements of each test operating point based on a preset multi-parameter coupled calculation model.

[0069] S2. For each test task, based on the multidimensional features of the test conditions, a clustering algorithm is used to divide it into several work packages, wherein the multidimensional features include specific resource requirements.

[0070] S3. Construct a multi-objective optimization model with task priority, total test duration, equipment utilization, task completion rate and production scheduling robustness as optimization objectives. Perform joint sorting of all test task work packages in the global scope. The optimization of equipment utilization is achieved by arranging work packages with similar resource demand patterns in temporal or spatial proximity.

[0071] S4. Under the constraints of resource capacity, mutual exclusion of test chambers and task continuity, allocate an available tester and a continuous time window that meets the execution duration requirements for each work package according to the joint sorting results, and generate an initial production schedule.

[0072] Furthermore, in step S1 above, the specific resource requirements for each test condition point are automatically calculated based on a preset multi-parameter coupled calculation model, including:

[0073] S11. Obtain the parameters of the test conditions included in each test task, including Mach number, altitude, air flow rate, intake air temperature, exhaust air temperature, exhaust pressure, and fuel flow rate, etc.

[0074] S12. A multi-parameter coupled calculation model is established by integrating the standard atmospheric model, gas dynamics function, flow conservation equation and heat balance formula. Based on the multi-parameter coupled calculation model, the required gas supply, gas extraction, fuel consumption, cooling water consumption and other requirements for each test operating point are automatically derived according to the parameters to form a specific resource requirement.

[0075] Furthermore, in step S2 above, for each experimental task, based on the multidimensional features of the experimental conditions, a clustering algorithm is used to divide it into several working packages, including:

[0076] S21. For each test task, construct a feature vector for its test operating point. The feature vector includes process requirement type, dedicated resource requirement, test type, risk level and task priority.

[0077] S22. Within this test task, a clustering algorithm is used to measure the similarity of all test conditions. With the goal of minimizing the Euclidean distance, all test conditions of this test task are divided into several work packages.

[0078] S23. For each test condition point in a work package, determine the execution order according to process logic or resource continuity logic, and ensure that the total execution time of the work package does not exceed the preset upper limit of the single continuous test time.

[0079] Furthermore, the clustering algorithm is either the K-means algorithm or the DBSCAN algorithm, and the similarity metric is calculated using weighted Euclidean distance; wherein, when calculating the similarity metric, the weights are dynamically configured based on the scarcity of resources or scheduling sensitivity in the high-altitude station system.

[0080] Furthermore, in step S3 above, the joint sorting of the work packages of all the experimental tasks on a global scale includes:

[0081] S31. Use work packages as the basic unit of scheduling to form a set of tasks to be scheduled.

[0082] S32. Define multiple optimization objectives, including minimizing the total trial duration, maximizing equipment utilization, minimizing risk, minimizing the probability of task delay, maximizing the task completion rate, and improving the robustness of the production schedule to disturbances; among them, the equipment utilization objective explicitly incorporates the measurement results of the similarity of resource requirements of work packages as an optimization factor.

[0083] S33. Construct a weighted comprehensive objective function based on multiple optimization objectives, normalize each optimization objective, and dynamically adjust the weight of each optimization objective according to any one or more of the following: real-time task priority, resource stress, or historical disturbance frequency.

[0084] S34. Use a multi-objective evolutionary algorithm to perform a global joint sort of all working packages, and solve the weighted synthesis objective function to generate a Pareto optimal solution set.

[0085] Furthermore, the multi-objective evolutionary algorithm is the NSGA-II algorithm, the MOEA / D algorithm, or a combination of both, and the weight coefficients of each objective are adjusted online according to real-time task priority, resource stress, or historical perturbation frequency.

[0086] Furthermore, in step S4 above, generating an initial production schedule includes:

[0087] S41. Establish constraint models for resource capacity constraints, test module mutual exclusion constraints, and mission continuity constraints;

[0088] S42. Based on the heuristic scheduling algorithm, each work package is sequentially assigned to an available test unit according to the joint sorting result. During the assignment process, if the current work package is highly similar to the work packages of other assigned test tasks in terms of requirements such as air supply, air extraction, fuel consumption, and cooling water consumption, they are given priority to be arranged in adjacent time periods of the same test unit to reuse the same air supply unit, air extraction unit, fuel pump, cooling water unit, etc., so as to reduce the number of auxiliary system switching and energy consumption.

[0089] S43. When all constraints are satisfied through the conflict detection and backtracking mechanism, the initial production schedule is output. The initial production schedule includes a task Gantt chart, a resource occupancy matrix, and a risk warning list.

[0090] Furthermore, the constraints described in step S43 include:

[0091] At any given time, the cumulative demand for each resource does not exceed its maximum available capacity;

[0092] Testers belonging to the same mutually exclusive set cannot be used simultaneously at the same time.

[0093] All test conditions within the same work package must be executed continuously without interruption or interleaving with other test tasks.

[0094] In an improved embodiment of the above-mentioned high-altitude platform multi-test mission scheduling method based on clustering algorithms and multi-objective optimization, such as... Figure 1 and Figure 2 As shown, the method further includes:

[0095] S5. Based on user needs, manually modify any one or more of the following: the work package division result, the joint sorting result, and the initial production schedule.

[0096] In an improved embodiment of the above-mentioned high-altitude platform multi-test mission scheduling method based on clustering algorithms and multi-objective optimization, such as... Figure 1 and Figure 2 As shown, the method further includes:

[0097] S6. When equipment failure, temporary order insertion, or schedule delay occurs, the initial production schedule will be partially rescheduled, specifically including:

[0098] S61. Identify the affected work packages and freeze the production schedules of unaffected test tasks;

[0099] S62. For affected work packages, reschedule them within the idle time window released by the freeze operation. The scheduling objective is to minimize the weighted sum of the total task delay time and the adjustment range of the production schedule. The weight coefficient is dynamically set according to the urgency of the task and the system stability requirements.

[0100] Based on the same inventive concept, this invention also provides a high-altitude multi-test mission scheduling system based on clustering algorithms and multi-objective optimization, as described in the following embodiments. Since the principle of the high-altitude multi-test mission scheduling system based on clustering algorithms and multi-objective optimization is similar to the high-altitude multi-test mission scheduling method based on clustering algorithms and multi-objective optimization disclosed in the above embodiments, the implementation of the high-altitude multi-test mission scheduling system based on clustering algorithms and multi-objective optimization can refer to the implementation of the high-altitude multi-test mission scheduling method based on clustering algorithms and multi-objective optimization, and repeated details will not be elaborated further. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0101] The high-altitude test platform multi-test task scheduling system disclosed in this invention is designed around the characteristics of the test tasks, such as clustering algorithms and multi-objective optimization. Figure 2 As shown, the system comprises four core modules: test point sorting and combination, resource matching, automatic scheduling of test tasks, and rescheduling. Interspersed within these modules are information input, manual review, and result output modules. Through the collaborative efforts of these modules, efficient and intelligent scheduling of test tasks is achieved. The following describes each module of the system.

[0102] (1) Information input module, which is used to execute step S1 of the above method:

[0103] This module acquires multiple test tasks and resources and their occupancy status. Each test task has multiple test operating points, and it acquires test parameter data (including altitude, Mach number, inlet temperature, etc.) for each test operating point. At the same time, through the multi-parameter coupled calculation model built into this module, which integrates standard atmospheric models, aerodynamic and thermodynamic formulas, etc., the module automatically calculates the requirements of various test resources based on the type of test task and test parameters.

[0104] Specifically, this module first acquires relevant information about the entire aero-engine testing process through the information input module. Simultaneously, it incorporates a multi-parameter coupled calculation model, integrating standard atmospheric models, aerodynamic and thermodynamic formulas, etc., to automatically calculate various test resource requirements based on test type and parameters. These include:

[0105] Experimental task information includes: task number, specific process requirements, task release time, completion time, priority, risk level, etc.

[0106] Equipment resource information: including equipment numbers, types, work shifts, available time periods, and maintenance cycles for equipment such as test benches and testing equipment;

[0107] Resource occupancy information: includes the special resource number, start and end time of occupancy, and occupancy type (such as exclusive or shared).

[0108] Process resource matching information: Clarify the compatibility between aero-engine test missions and special resources, and the threshold of resource demand, etc.

[0109] Built-in calculation formulas: clearly define the compatibility between aero-engine test missions and special resources, and the threshold of resource requirements. This includes environmental parameter calculations: integrating the Standard Atmospheric Model (ISA) and aerodynamic thermodynamic formulas to achieve one-click temperature / pressure calculations; dynamic flow correction: developing a variable operating condition compensation algorithm to support automatic conversion from engine design airflow to actual airflow; intelligent extraction volume judgment: triggering the dual-flow calculation module through test type tags (such as "extraction test") and automatically adding a 20% redundancy coefficient.

[0110] (2) Test point sorting and combination module, which is used to perform the above steps S2 and S3:

[0111] Based on the process requirements, calculated test resource requirements, and task priorities of aero-engine test missions, this module uses a clustering algorithm to sort and group all test operating points corresponding to each test mission into several work packages. The clustering features are set around the process requirements, specific resource requirements, test types, risk levels, and test priorities calculated from the aero-engine test operating points, so as to realize the automatic sorting and accurate clustering of test operating points in a single aero-engine test mission.

[0112] Specifically, this module addresses the diverse operating conditions and significant process differences in aero-engine testing missions. Based on the multi-parameter coupled calculation model built into the information input module and integrated standard atmospheric models and aerodynamic thermodynamic formulas, it automatically calculates the resource requirements for various tests based on test type and parameters. It then uses density-based clustering algorithms (such as DBSCAN) or K-means algorithms to accurately group aero-engine test operating point into standardized work packages, laying the foundation for subsequent resource matching and production scheduling optimization.

[0113] When grouping, the clustering features are specifically adapted to aero-engine testing scenarios, including:

[0114] Process requirements vector: covering core process parameters for aero-engine testing, such as gas supply temperature, gas supply pressure, gas supply flow rate, extraction pressure, Mach number, etc.

[0115] Resource demand vector: This refers to the resource requirements for aero-engine testing, including resource types and quantities such as large-scale gas supply, small-scale gas supply, heating, and cooling.

[0116] Priority and risk level: Priority is set based on the research and development stage and completion milestones of the aero-engine test mission, and the risk level during the test process is assessed simultaneously.

[0117] Define the set of aero-engine test missions. for:

[0118] ;

[0119] Experimental Task The test conditions and their sequence are as follows:

[0120] ;

[0121] No. Test operating conditions The feature vectors are:

[0122] ;

[0123] Among them, features This includes core dimensions such as aero-engine testing process requirements (temperature, pressure, flow rate, Mach number, etc.), specific resource requirements (large gas supply, small gas supply, heating, cooling, etc.), priority of test operating points, and risk level. This is represented as the first of the experimental tasks. Each operating condition point This indicates the total number of elements to be considered for sorting, such as This indicates the total gas supply temperature required for the test operating point. This indicates the total air supply pressure required for this test operating point. This indicates the total gas supply flow required at the test operating point. This indicates the total pumping pressure required for the test operating point. This indicates the total pumping flow rate required for this test operating point. This indicates the maximum gas supply flow required for this test operating point. This indicates the minimum gas supply flow rate required for this test operating point. This indicates the required heating gas temperature for this test operating point. This indicates the required heating gas flow rate at the test operating point, etc.

[0124] The Euclidean distance is used as a similarity measure for the test conditions.

[0125] ;

[0126] in, express and Similarity index between them , Indicates the first The, the The first working condition point 1 sorted element This indicates the total number of elements to be considered in the sorting.

[0127] The clustering objective of this algorithm is to minimize the Euclidean distance. After calculating the similarity metrics for all test points based on the above similarity metrics, the order of the test points that minimize the total number of similarity metrics is as follows:

[0128] After minimizing the Euclidean distance, the order of all test condition points in the test task is as follows:

[0129] ;

[0130] in, Still the original number Each test task includes test operating points Compared with the test conditions before clustering They are the same, only the order is different;

[0131] The above The tasks are divided into several work packages according to specific rules, such as total time limits:

[0132] ;

[0133] Each work package A suitable set of aero-engine test resources needs to be matched to ensure that the tasks within the package have similar process or resource requirements. This is used to break down a single test task that is long-duration and multi-condition into multiple work packages, and carry out the test work in sequence, thereby improving the efficiency of resource matching and the rationality of production scheduling.

[0134] (3) Resource matching module, which establishes the constraints for step S4:

[0135] Based on the resource requirements of each work package, this module generates a resource allocation plan for each work package by matching available resources, taking into account factors such as the availability of dedicated resources for aero-engine testing, aero-engine testing shift requirements, aero-engine testing resource conflict avoidance, and aero-engine testing dedicated resource capacity matching.

[0136] Specifically, this module addresses the characteristics of aero-engine testing resources, which are diverse and numerous (such as test benches, test equipment, and gas supply and extraction equipment), by accurately matching available resource combinations based on the resource requirements of the work package and generating a scientific resource allocation plan.

[0137] Resource combination The following requirements must be met for aero-engine testing: resource availability, shift work constraints, and capacity limitations.

[0138] ;

[0139] The objective function of resource matching is to minimize resource conflicts and idle time, thereby improving the utilization rate of aero-engine testing resources.

[0140] ;

[0141] in, , All are weighting coefficients, which can be dynamically adjusted according to the importance of aero-engine test resources; Conflict represents the number of resource conflicts; IdleTime represents the resource idle time.

[0142] (4) Automatic production scheduling module for test tasks. This module is used to execute step S4:

[0143] This module first sorts all test tasks based on set rules, including the priority of aero-engine test tasks, total test duration, test risk level, equipment utilization rate, aero-engine test task completion rate, aero-engine test task delay rate, and robustness of aero-engine test production schedule. Then, it uses a heuristic algorithm to jointly schedule test tasks involving multiple testers, multiple test tasks, and multiple test resources.

[0144] Specifically, this module addresses the Resource Constrained Project Scheduling Problem (RCPSP) in aero-engine testing scenarios. It combines test module mutual exclusion constraints, special resource upper limit control, and task sequence dependencies, and employs a priority rule-based heuristic algorithm to automatically schedule test tasks.

[0145] Among them, the following core objectives for adapting to aero-engine testing scenarios must be met:

[0146] Sequence constraints: Task packages are attempted to be scheduled in the order of input, but can be flexibly adjusted according to the urgency of the aero-engine test, and the original order is not mandatory;

[0147] Resource constraints: Strictly meet the upper limit of the utilization rate of aero-engine test equipment and special resources set by the chief test engineer to avoid resource overload;

[0148] Mutual exclusion constraint: The lowest level constraint, mutual exclusion test chambers cannot operate in parallel to prevent mutual interference between tests;

[0149] Time constraints: Under the premise of meeting all constraints, find the earliest feasible start time for each test task to ensure delivery milestones.

[0150] The core principle of this module is shown in Table 1 below:

[0151] Table 1: Constraints in the Automatic Scheduling Module for Experimental Tasks

[0152]

[0153] The constraints mainly include resource capacity constraints, test module mutual exclusion constraints, and mission continuity constraints:

[0154] 4.1 Resource capacity constraints:

[0155] ;

[0156] in, Indicates time A set of ongoing experimental tasks.

[0157] 4.2 Mutual Exclusion Constraints of the Test Chamber:

[0158] .

[0159] 4.3 Task Continuity Constraints:

[0160] .

[0161] Among them, see Figure 3 The diagram shows a flowchart of window scanning using a heuristic algorithm. The specific steps of the heuristic algorithm are as follows:

[0162] a) Initialize the time axis: Set the time range for aero-engine test production and initialize the time axis parameters synchronously;

[0163] b) Create a resource time capacity matrix Real-time mapping of available capacity of dedicated resources at each time point;

[0164] c) Initialize the mutual exclusion test chamber status table: record the availability status and occupation time of each test chamber;

[0165] d) Construct a task priority queue: Based on the input order and the priority of the aero-engine test tasks, create a task queue Q={J1,J2,...,Jn} to be scheduled;

[0166] e) Dynamic time window scanning: Real-time monitoring of resource status and task progress via a sliding time window;

[0167] f) Conflict resolution mechanism: When encountering mutually exclusive conflicts, the task schedule is adjusted using the "earliest available time first" strategy; in the case of insufficient resources, the "time window sliding + resource pre-allocation" strategy is implemented to ensure that core tasks are carried out in priority.

[0168] In practical applications, this heuristic algorithm can control the time complexity to O(n²) through a sliding time window and state caching mechanism, making it suitable for medium-scale (<1000 tasks) production scheduling scenarios in aero-engine testing. Finally, the structured output module generates visual results such as Gantt charts and resource load curves, intuitively presenting the production scheduling plan.

[0169] (5) Manual review module, which is used to perform step S5 above:

[0170] This module can support multiple stages in the joint scheduling process of aero-engine test missions. For example: the first manual review checks whether the sorting results of test operating points meet expectations and habits. If adjustments are needed, the sorting results can be modified manually and then saved to the program. The second manual review checks whether the allocation results of resources and equipment meet expectations and habits. If adjustments are needed, the sorting results can be modified manually and then saved to the program. The third manual review checks whether the scheduling results of the test mission meet expectations and habits. If adjustments are needed, the adjusted test scheduling sheet is output and issued, and the software scheduling ends.

[0171] Specifically, this module supports multiple stages in the aircraft engine test and production scheduling process, including the following steps:

[0172] First manual review: Review whether the test condition point sorting results meet expectations and habits. If adjustments are needed, the sorting results can be manually modified directly and then saved to the program.

[0173] Second manual review: Review whether the resource and equipment allocation results meet expectations and habits. If adjustments are needed, the sorting results can be modified manually and then saved to the program.

[0174] The third manual review: review whether the production scheduling results of the test task meet expectations and habits. If adjustments are needed, output the adjusted test scheduling sheet and issue it, and end the software scheduling.

[0175] (6) Rescheduling module: This module is used to execute the above step S6 and supports dynamic rescheduling under trigger conditions such as temporary order insertion, mission delay, and manual adjustment during the aero-engine test process.

[0176] Specifically, this module addresses the challenges of frequent and widespread unforeseen events in aero-engine testing missions. In the following scenarios, planners can trigger a production rescheduling to ensure the flexibility and adaptability of the production plan:

[0177] a) Temporary order insertion and rescheduling: After updating the task list, the input information of the scheduling algorithm is affected, and rescheduling yields improved results. When selecting rescheduling targets, the similarity between the rescheduling and the original scheduling results can be ensured by marking tasks that will not be included in this rescheduling.

[0178] b) Manual adjustment of rescheduling: By adjusting the priority of tasks, the input information of the scheduling algorithm is affected, and the rescheduling results are improved. When selecting rescheduling targets, tasks that are not included in the current rescheduling can be marked to ensure the similarity between the rescheduling and the original scheduling results.

[0179] c) Trial Delay Rescheduling: Input the extended occupancy time into the equipment occupancy table to affect the input information of the scheduling algorithm, and rescheduling will yield improved results. When selecting rescheduling targets, the similarity between the rescheduling and the original scheduling results can be ensured by marking tasks that will not be included in this rescheduling.

[0180] d) Equipment failure rescheduling: If equipment failure prevents subsequent tasks from being performed, delete the tasks scheduled for that tester, update the task list, and affect the input information of the scheduling algorithm. Rescheduling will then yield improved results.

[0181] The rescheduling strategy of this module adopts a partial rescheduling + retention mechanism. Its core logic is: retain the schedule of unaffected aero-engine test missions; only reschedule the affected missions; and use heuristic rules (such as shortest test time priority and earliest delivery priority) to quickly generate new solutions, balancing efficiency and optimization effect.

[0182] Specifically, let the set of affected tasks be... The rearrangement production target is:

[0183] ;

[0184] in, , These are the weighting coefficients for mission delay and resource status, which can be dynamically adjusted according to the importance of the aero-engine test mission; Delay represents the mission delay time; Cost represents the resource usage or amount of modification.

[0185] The production rescheduling logic process is as follows: Figure 4 As shown, the specific process is as follows:

[0186] a) Update the task list: Update the aero-engine test task list and related resource information synchronously according to emergencies (order insertion, delay, equipment failure, etc.);

[0187] b) Identify unaffected tasks: The system automatically analyzes the impact of emergencies, identifies unaffected aero-engine test tasks, and retains their original schedules;

[0188] c) Rescheduling affected tasks: For the set of affected tasks, heuristic rules are used to quickly generate new scheduling plans to ensure that multi-objective optimization requirements are met;

[0189] d) Output new production schedule: Integrate the retained original schedule with the newly generated affected task schedule, output a complete new production schedule, and update the visualization results simultaneously.

[0190] (7) Result Output Module: This module generates and outputs the following structured results, including an overview of aero-engine test missions, a dedicated resource allocation table for aero-engine tests, and a production schedule for aero-engine test equipment:

[0191] a) Summary table of test tasks: including task number, test start time, test end time, priority, risk level, etc.;

[0192] b) Experimental Task Resource Table: including resource allocation, resource occupancy time, resource utilization rate, etc.;

[0193] c) Equipment work schedule: including test equipment number, test task number, occupation time, equipment utilization rate, etc.

[0194] Production scheduling results can be as follows Figure 5 The data is displayed in the format shown and can be exported to Excel, PDF, and other formats for easy viewing and analysis by users.

[0195] Verified by actual test data, in a real-world application scenario at an aero-engine test base, the intelligent scheduling algorithm of this invention significantly optimized all core indicators, with specific results as follows: single test planning efficiency increased by approximately 90%, single test cycle shortened by approximately 20%, and single test task cost reduced by approximately 20%; overall scheduling efficiency increased by approximately 90%, monthly effective test hours increased by approximately 15%, and overall test cost reduced by approximately 20%; the degree of manual intervention in the entire scheduling process was controlled within 20%. These achievements bring significant economic and management benefits, effectively solving the technical limitations of low efficiency, high cost, and high reliance on manual labor under the traditional scheduling model.

[0196] This invention proposes a method that establishes a two-layer intelligent scheduling mechanism combining intra-task clustering and cross-task multi-objective collaborative scheduling. First, within each aero-engine test mission, based on the process attributes and resource requirements of the operating point, a clustering algorithm automatically generates highly cohesive work packages to ensure the continuity and integrity of mission execution. Second, at the global level, all work packages are used as scheduling units to construct a multi-objective optimization model that integrates task priority, equipment utilization, completion rate, and plan robustness. The similarity of resource requirements between work packages is explicitly introduced as an optimization factor to achieve efficient resource reuse and proximity arrangement across missions. Simultaneously, the system supports manual intervention and local rescheduling to ensure rapid recovery to stable operation under dynamic disturbances.

[0197] Compared with the prior art, the method of the present invention has the following beneficial effects:

[0198] 1. Significantly improves production scheduling efficiency: The clustering algorithm automatically groups test conditions and generates work packages, greatly reducing the workload of manual task decomposition and scheduling design; according to actual tests at an aero-engine test base, the human participation rate is less than 20%, and the overall production scheduling efficiency is improved by more than 90%.

[0199] 2. Significantly improves resource utilization: Under a multi-objective optimization framework, combining a resource matching model with a similarity-guided proximity arrangement strategy, it achieves efficient allocation and reuse of key resources such as gas supply, extraction, fuel oil, and cooling water; actual test data shows that the overall resource utilization rate is improved by more than 10%, and the number of auxiliary system start-ups and shutdowns is reduced by 30%.

[0200] 3. Enhance the robustness of production scheduling: Explicitly introduce "planning robustness" indicators (such as sensitivity to disturbances, buffer time distribution, etc.) into the optimization objectives to make the generated production scheduling plan more resistant to interference and reduce the risk of cascading delays caused by minor disturbances.

[0201] 4. The system has multiple built-in rescheduling trigger mechanisms to respond to scenarios such as temporary order insertion, equipment failure, task delays and manual adjustments. It adopts a freeze-rescheduling strategy to reschedule only the affected tasks locally, which improves the rescheduling speed by more than 50% and ensures the stability of the overall plan.

[0202] 5. This solution fully considers the characteristics of strong coupling of multiple resources, complex process logic, and diverse task priorities. It has been successfully deployed at a national-level aero-engine test base and can support the collaborative production of multiple medium and large test vehicles. It has good engineering scalability and industry universality.

[0203] 6. The method of the present invention is not only applicable to high-altitude simulation test benches for aero-engines, but can also be extended to high-complexity multi-mission test scheduling scenarios for gas turbines, aerospace propulsion systems, or new energy power plants.

[0204] In this embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned high-altitude multi-test task scheduling methods based on clustering algorithms and multi-objective optimization.

[0205] Specifically, the computer device can be a computer terminal, a server, or a similar computing device.

[0206] In this embodiment, a computer-readable storage medium is provided, which stores a computer program that executes any of the above-described high-altitude platform multi-test task scheduling methods based on clustering algorithms and multi-objective optimization.

[0207] Specifically, computer-readable storage media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable storage media does not include transient media, such as modulated data signals and carrier waves.

[0208] Obviously, those skilled in the art should understand that the modules or steps of the above-described embodiments of the present invention can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of the present invention are not limited to any particular hardware and software combination.

[0209] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-altitude test mission scheduling method based on clustering algorithm and multi-objective optimization, characterized in that, The method includes: Multiple aero-engine test missions are acquired, each of which contains multiple test conditions. Based on a preset multi-parameter coupled calculation model, the exclusive resource requirements for each test condition are automatically calculated. For each test task, based on the multidimensional features of the test conditions, a clustering algorithm is used to divide it into several work packages. These multidimensional features include specific resource requirements. Specifically, for each test task, a feature vector is constructed for its test conditions, including process requirement type, specific resource requirements, test type, risk level, and task priority. Within the test task, a clustering algorithm is used to measure the similarity of all test conditions, aiming to minimize the Euclidean distance, and all test conditions of the test task are divided into several work packages. For the test conditions within each work package, the execution order is determined according to process logic or resource continuity logic, ensuring that the total execution time of the work package does not exceed the preset upper limit of a single continuous test time. A multi-objective optimization model is constructed, using task priority, total trial duration, equipment utilization, task completion rate, and scheduling robustness as optimization objectives. The model globally performs joint sorting of all trial task packages. Equipment utilization optimization is achieved by arranging work packages with similar resource demand patterns in temporal or spatial proximity. Specifically, this includes: using work packages as basic scheduling units to form a set of tasks to be scheduled; defining multiple optimization objectives, including minimizing total trial duration, maximizing equipment utilization, minimizing risk, minimizing task delay probability, maximizing task completion rate, and improving the robustness of the scheduling plan to disturbances; explicitly incorporating the similarity of resource demand between work packages as an optimization factor into the equipment utilization objective; constructing a weighted comprehensive objective function based on multiple optimization objectives, normalizing each objective, and dynamically adjusting the weights of each objective according to any one or more of real-time task priority, resource stress, or historical disturbance frequency; and using a multi-objective evolutionary algorithm to globally jointly sort all work packages and solve the weighted comprehensive objective function to generate a Pareto optimal solution set. Under the constraints of resource capacity, mutual exclusion of test chambers, and task continuity, an available tester and a continuous time window that meets the execution duration requirements are allocated to each work package based on the joint sorting results, and an initial production schedule is generated.

2. The high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, Based on a pre-defined multi-parameter coupled calculation model, the specific resource requirements for each test condition point are automatically calculated, including: Obtain the parameters of the test conditions included in each test task, including Mach number, altitude, intake air flow rate, intake air temperature, exhaust air temperature, exhaust pressure, and fuel flow rate; A multi-parameter coupled calculation model is established by integrating the standard atmospheric model, gas dynamics function, flow conservation equation and heat balance formula. Based on the multi-parameter coupled calculation model, the required gas supply, gas extraction, fuel consumption and cooling water consumption for each test operating point are automatically derived according to the parameters to form a specific resource requirement.

3. The high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, The clustering algorithm is either the K-means algorithm or the DBSCAN algorithm, and the similarity metric is calculated using weighted Euclidean distance. The weights are dynamically configured based on the scarcity of resources or scheduling sensitivity in the high-altitude station system when calculating the similarity metric.

4. The high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, The multi-objective evolutionary algorithm is the NSGA-II algorithm, the MOEA / D algorithm, or a combination of both. The weight coefficients of each objective are adjusted online according to the real-time task priority, resource stress, or historical perturbation frequency.

5. The high-altitude test mission scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, Generate an initial production schedule, including: Establish constraint models for resource capacity constraints, mutual exclusion constraints of experimental modules, and mission continuity constraints; Based on the heuristic scheduling algorithm, each work package is sequentially assigned to an available test unit according to the joint sorting result. During the assignment process, if the current work package is highly similar to other assigned test task work packages in terms of air supply, air extraction, fuel consumption, and cooling water consumption, they are given priority to be arranged in adjacent time periods of the same test unit to reuse the same air supply unit, air extraction unit, fuel pump, and cooling water unit, thereby reducing the number of auxiliary system switching and energy consumption. When all constraints are satisfied through conflict detection and backtracking mechanisms, an initial production schedule is output, which includes a task Gantt chart, a resource occupancy matrix, and a risk warning list.

6. The high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to claim 5, characterized in that, The constraints include: At any given time, the cumulative demand for each resource does not exceed its maximum available capacity; Testers belonging to the same mutually exclusive set cannot be used simultaneously at the same time. All test conditions within the same work package must be executed continuously without interruption or interleaving with other test tasks.

7. The high-altitude test mission scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, The method further includes: Based on user needs, any one or more of the following can be manually modified: the work package division result, the joint sorting result, and the initial production schedule.

8. The high-altitude platform multi-test task scheduling method based on clustering algorithm and multi-objective optimization according to claim 1, characterized in that, The method further includes: When equipment failure, unexpected order insertion, or schedule delays occur, the initial production schedule will be partially rescheduled, specifically including: Identify the affected work packages and freeze the scheduling of unaffected test tasks; For affected work packages, they are rescheduled within the idle time window released by the freeze operation. The scheduling objective is to minimize the weighted sum of the total task delay time and the adjustment of the production schedule. The weighting coefficients are dynamically set according to the urgency of the task and the system stability requirements.

Images