A method for swarm detection situation recognition and task generation
By comprehensively analyzing the time urgency and resource scarcity of aircraft engine detection tasks, dividing multiple scheduling situations and designing scheduling modes, the problem of low scheduling efficiency in multi-engine detection is solved, and the utilization rate of detection resource and task processing efficiency is improved. .
Patent Information
- Application Number
- CN202111571223.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-21
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-12-21
AI Technical Summary
In multi-engine detection, the prior art has problems with low manual detection accuracy and low scheduling efficiency of detection tasks. The traditional scheduling model fails to comprehensively measure the system scheduling situation, resulting in low completion rate and efficiency of detection tasks.
By comprehensively analyzing the time urgency and resource scarcity of aircraft engine fault detection tasks, dividing multiple scheduling situations, and designing different scheduling modes to generate priority for any aircraft group detection tasks, improving the adaptability and efficiency of dynamic scheduling of the system.
It realizes scheduling pattern matching that is more in line with the system situation, improves the scheduling efficiency of detection resources and task processing efficiency, and enhances the system's adaptability and resource utilization.
Smart Images

Figure CN114493100B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of detection task generation for multi-engine detection, and specifically relates to a method for cluster detection scenario recognition and task generation. Background Art
[0002] In multi-engine inspection, aero-engine fault detection has problems such as low manual inspection accuracy and low efficiency of manual inspection task allocation. After the inspection robot technology is applied to aero-engine fault detection, there is a need to schedule multiple inspection tasks. Under limited inspection robot resources, the scheduling model is used to generate a scheduling plan to improve the completion rate and efficiency of the inspection tasks.
[0003] Traditional scheduling model matching methods usually only consider indicators such as time urgency or the degree of matching between demand and service capabilities, and fail to comprehensively measure the system scheduling scenario. Before scheduling model matching, system scheduling scenario identification is introduced, and the system scenario is comprehensively evaluated using time urgency and resource scarcity. Comprehensive considerations can be made from the time dimension and the resource dimension, and a scheduling model that better suits the system situation can be matched. Summary of the invention
[0004] The technical problem to be solved by the present invention is: to overcome the shortcomings of the prior art and provide a method for cluster detection scenario identification and task generation, by comprehensively analyzing the time urgency and resource scarcity of cluster detection tasks, dividing multiple scheduling scenarios, and designing different scheduling modes according to different scenarios, taking into account the realization of multiple scheduling constraints and scheduling goals, and flexibly generating the detection task priority of any cluster, so as to provide guidance for realizing dynamic scheduling of the system and improving its adaptability.
[0005] The present invention solves its technical solution: a method for cluster detection scenario recognition and task generation, including five steps of constructing fault detection resource configuration rules, evaluating the current state of the system, identifying the detection scheduling scenario based on the state evaluation, configuring the detection scheduling mode and scheduling rules based on the scenario category, and generating task priority sorting.
[0006] The specific implementation is as follows:
[0007] Step 1: According to the classification of aircraft engine fault severity, determine the required detection content, detection resources and detection time, and build detection resource configuration rules for aircraft engine fault detection;
[0008] Step 2: Evaluate the fault detection task F n The current state is based on the N tasks to be detected F received within a period of time n , n=1,2,3…,N and the required detection resources determined by the configuration rules in step 1 m=1,2,3…,M,i=1,2,3…,I, determine the task F to be detected n ,n=1,2,3…,N two task states of time urgency and resource scarcity;
[0009] Step 3: Based on the fault detection tasks and fault detection resources in step 2, determine the overall state of the fault detection system, identify the fleet detection task scheduling scenario, and divide the fault detection system detection scenarios into five scenarios: low frequency, medium-low frequency, medium frequency, medium-high frequency, and high frequency according to the two indicators of time urgency and resource scarcity. Based on the state evaluation results, identify the fleet detection task scheduling scenario;
[0010] Step 4: Based on the problem background, comprehensively consider the goals of improving resource utilization and task processing efficiency, and design the detection task scheduling mode and its scheduling rules; then, according to the different characteristics of the five detection scenarios in step 3, determine the scheduling mode and its scheduling rules required for different detection scenarios;
[0011] Step 5: Based on the scheduling mode of step 4, the weights of each scheduling rule of the required scheduling mode are set in a user-defined or recommended manner, and a comprehensive priority ranking for any fleet detection task is generated under the constraints of the corresponding scheduling rule combination. This completes the generation of detection task requirements and comprehensive priorities for the fleet detection scenario.
[0012] The step 1 specifically includes:
[0013] The degree of engine failure is divided into five levels: I, II, III, IV, and V, corresponding to minor consequences, significant consequences, serious consequences, major consequences, and catastrophic consequences, respectively. According to the engine components, the failure modes corresponding to different components of different modules are determined, and corresponded to the engine failure levels. Fault detection contents corresponding to the engine failure levels are established. There are 10 types of failures, including wear, corrosion, cracks, tears, fractures, deformation, looseness, torque transmission surface compression, and blockage of main leakage. The fault types are divided into different levels according to their severity and the functions of detection resources to determine the detection task requirements and detection contents of different fault types. The detection resource configuration rules for the detection time, type and number of detection robots for aircraft engines with different levels of engine failure and corresponding to different fault types are completed.
[0014] The step 2 specifically includes:
[0015] For n tasks to be detected F n And testing resources First, the status of the task to be tested is evaluated from the two aspects of time urgency and resource scarcity;
[0016] (1) When there are multiple detection tasks, the deadline is less than the task to be detected Fn The task with the deadline will exist with the task to be detected F n Competition for detection resources, so the number of tasks to be detected at time t is F n The time urgency is recorded as:
[0017]
[0018] Among them, the deadline is less than the task F n The set of tasks with deadlines is denoted as Represents the task to be detected F n The detection time, Represents the task to be detected F n The test preparation time, Represents the task to be detected F n The deadline of , t represents the current time;
[0019] (2) For the task to be detected F n The resource scarcity of the mth detection resource is recorded as:
[0020]
[0021] Among them, the detection resource set is R = {r m ,m=1,2,3…,M}, m represents the type of detection resource, Represents the task to be detected F n The number of the mth detection resources required.
[0022] The step 3 is specifically implemented as follows:
[0023] (1) The overall time urgency U is used to detect the task F n The time urgency is expressed in the form of a collection:
[0024]
[0025] Overall resource scarcity S m , if S m If it is greater than 1, it means that the current m-th detection resource cannot meet the needs of the detection task. m Denoted as:
[0026]
[0027] (2) The time urgency of the detection task and the scarcity of detection resources are used as the classification criteria for the detection scheduling scenarios. By solving the time urgency of the detection task and the scarcity of detection resources, the current detection scheduling scenario is identified; the i-th detection scheduling scenario is represented by DC i Indicates that it is divided into five categories: low-frequency situational DC1 , medium and low frequency DC 2 , medium frequency scenario DC 3 , medium and high frequency scenario DC 4 , high frequency situational DC 5 ;
[0028] The classification criteria for the detection scheduling scenario are: when the time urgency of each detection task is less than the threshold u * , and resource scarcity S m are all less than 1, at this time, the system resources are sufficient, the task can be executed in sufficient time, and the detection scheduling situation is judged as a low-frequency situation; when the proportion of detection tasks with time urgency greater than the threshold u* is greater than 10%, and the resource scarcity S t When both are less than 1, the system resources are sufficient and the task execution time is relatively ample, and the detection scheduling situation is judged as a medium-low frequency situation. When the proportion of detection tasks with time urgency greater than the threshold u* is greater than 30%, and the maximum value of resource scarcity is 1, the system resources are relatively sufficient and the task execution time is relatively ample, and the detection scheduling situation is judged as a medium frequency situation; when the proportion of detection tasks with time urgency greater than the threshold u* is greater than 50%, and there are resource scarcity greater than 1, the situation is judged as a medium-high frequency situation; when the proportion of detection tasks with time urgency greater than the threshold u* is greater than 50%, and the resource scarcity is greater than 1, the situation is judged as a high frequency situation.
[0029] The step 4 specifically includes:
[0030] Based on five detection scenarios DC i Different scheduling modes are configured for different characteristics. The detection scenarios are divided into two aspects: time urgency and resource scarcity. The scheduling mode is configured from two levels: improving resource utilization and task processing efficiency. The scheduling mode is expressed as SM i ,i=1,2…,6, scheduling mode SM i The corresponding scheduling rule is expressed as j=1,2…,m(i), m(i) is the scheduling mode SM i The number of corresponding dispatch rules;
[0031] (1) Consider resource utilization scheduling mode:
[0032] Scheduling mode SM 1 : Reduce costs
[0033] The total cost of using the inspection robot is the sum of the depreciation and electricity cost of each robot during use and the standby cost during the non-use period:
[0034]
[0035]
[0036] Among them, a h represents the unit cost of the h-type detection robot during its use, b h represents the unit standby cost of the h-type detection robot during the unused period, and then sums the usage cost of each detection robot. The scheduling mode SM 1 The corresponding scheduling rule is that the higher the detection cost, the higher the priority As shown in formula (6);
[0037] Scheduling mode SM 2 : Reduce the average waiting time
[0038] The average waiting time is the average time each detection task waits in the detection system from the time it is received by the system to the time it starts to execute the detection:
[0039]
[0040]
[0041] in, Representative Task F n The start detection time, Representative Task F n The arrival time of the dispatch mode SM 2 The corresponding dispatching rule is that the earlier you arrive, the higher your priority. As shown in formula (8);
[0042] Scheduling mode SM 3 :Improve the utilization rate of detection resources
[0043] The detection resource utilization rate is the ratio of the average usage time of each detection tool to the total detection time:
[0044]
[0045]
[0046]
[0047] in, Represents the detection resource of type h Use time, Represents the detection task F n The end detection time, The maximum value of indicates the end time of the entire detection process. Represents the detection task F n The start detection time, The minimum value of indicates the start time of the entire detection process. Scheduling mode SMi 3The corresponding scheduling rule is that the longer the detection time, the higher the priority Tasks with earlier deadlines have higher priority As shown in formula (10), (11);
[0048] (2) Scheduling mode considering task processing efficiency:
[0049] Scheduling mode SM 4 : Reduce the total completion time
[0050] The total completion time is the total time to complete all inspection tasks:
[0051]
[0052]
[0053] in, Represents the detection task F n The end detection time, Represents the detection task F n Start detection time, scheduling mode SM 4 The corresponding scheduling rule is that the shorter the detection time, the higher the priority As shown in formula (13);
[0054] Scheduling mode SM 5 : Reduce the proportion of delayed tasks
[0055]
[0056]
[0057]
[0058] Among them, N tardiness Represents the number of delayed tasks, N represents the total number of detection tasks, and the scheduling mode is SM 5 The corresponding scheduling rule is that the fewer detection resources are needed, the higher the priority The shorter the detection time, the higher the priority As shown in formulas (15) and (16);
[0059] Scheduling mode SM 6 :Reducing the average delay time:
[0060]
[0061]
[0062]
[0063] in, Represents the detection task F n The end detection time, Represents the detection task F n Task deadline, scheduling mode SM 6 The corresponding scheduling rule is that the shorter the detection time, the higher the priority The more time-sensitive the situation, the higher the priority. As shown in formulas (18) and (19);
[0064] (3) The scheduling mode is configured according to the characteristics of the detection scenario. When configuring the scheduling mode, since the detection pressure in low-frequency and medium-low-frequency scenarios is small, it is considered to improve the utilization rate of detection resources. As the scenario improves, the detection pressure increases, and the focus of scheduling mode selection is transferred from improving the utilization rate of detection resources to improving the processing efficiency of detection tasks. The scheduling rules corresponding to the low-frequency detection scenario are: The dispatch rules corresponding to the low-medium frequency detection scenario are: The dispatch rules corresponding to the medium frequency detection scenario are: The dispatch rules corresponding to the medium and high frequency detection scenarios are: The scheduling rules corresponding to the high-frequency detection scenario are: The higher the detection cost, the higher the priority. The first to arrive will have priority. The longer the detection time, the higher the priority. The earlier the deadline, the higher the priority. The shorter the detection time, the higher the priority. The fewer resources required for testing, the higher the priority. The shorter the detection time, the higher the priority. The shorter the detection time, the higher the priority. The greater the time urgency, the higher the priority, and then generate the corresponding relationship matrix RM between the detection scenario and the scheduling rule.
[0065] The step 5 specifically includes:
[0066] (1) Determine the weight of each dispatch rule. After matching the dispatch mode and dispatch rules, it is also necessary to determine the priority of each detection task. Since each detection scenario corresponds to multiple dispatch rules, when the dispatch rules are combined, it is necessary to set the importance of each dispatch rule, that is, their weight coefficient. The weight of each dispatch rule is set by user definition or system recommendation. The user definition method also needs to adopt corresponding constraints;
[0067] (2) Based on the dispatch rule weights determined in step (1), combined with multiple dispatch rules under different scenarios, the fleet detection tasks are comprehensively prioritized.
[0068] The advantages of the present invention compared with the prior art are:
[0069] (1) Based on different aircraft engine fault levels, the aircraft engine detection contents are divided and the resource allocation rules for aircraft engine fault detection are constructed, which is more conducive to the standardized generation of detection tasks and is conducive to improving the efficiency of detection resource scheduling.
[0070] (2) When evaluating the status of the detection system, not only the urgency of the task to be detected is taken into account, but also the resource scarcity of the current system is comprehensively considered, which can comprehensively evaluate the current status of the detection system.
[0071] (3) The system context recognition method is applied to the detection task generation process. Different detection mode configurations are designed for different system detection contexts to improve the adaptability between task priority and system detection context. BRIEF DESCRIPTION OF THE DRAWINGS
[0072] Figure 1 is a flow chart of the present invention;
[0073] Figure 2 It is a diagram illustrating the detection rules of a faulty compressor module of an aircraft engine;
[0074] Figure 3 It is a diagram illustrating the detection rules of the combustion module of an aircraft engine failure;
[0075] Figure 4 It is a diagram illustrating the detection rules of a faulty turbine module of an aircraft engine;
[0076] Figure 5 It is a diagram illustrating the detection rules of the faulty accessory drive module of an aircraft engine;
[0077] Figure 6 It is a standard diagram for dividing detection and scheduling scenarios;
[0078] Figure 7 It is a correspondence diagram between detection scenarios and scheduling modes. DETAILED DESCRIPTION
[0079] The present invention is described in further detail below in conjunction with the accompanying drawings.
[0080] The present invention discloses a method for identifying a cluster detection scenario and generating a task, which includes five steps: constructing fault detection resource configuration rules, evaluating the current state of the system, identifying the detection scheduling scenario based on the state evaluation, configuring the detection scheduling mode and scheduling rules based on the scenario category, and generating task priority sorting. The present invention divides a plurality of scheduling scenarios by comprehensively analyzing the time urgency and resource scarcity of the cluster detection task, and designs different scheduling modes according to different scenarios, taking into account the realization of multiple scheduling constraints and scheduling goals, and flexibly generates the detection task priority of any cluster, providing guidance for realizing the dynamic scheduling of the system and improving its adaptability.
[0081] like Figure 1 As shown, the specific implementation of the present invention is as follows:
[0082] Step 1: Fault detection resource configuration rules, the specific implementation is as follows:
[0083] (1) The engine failure degree is divided into I~V:
[0084] Level I, minor consequences, the shell is damaged, including scratches, dents, cracks, etc., but there are no other safety hazards;
[0085] Level II, significant consequences; connecting components such as crankshaft connecting rods are slightly damaged, including twisting and separation;
[0086] Level III, severe consequences, pipeline blockage causing small-scale fire and other damage, minor fuel leakage;
[0087] Level IV: Major consequences: internal parts are seriously damaged by penetration, cracks, etc., affecting the normal operation of the engine;
[0088] Level V, catastrophic consequences, causing substantial damage to the aircraft and engine systems, permanent loss of thrust; important components falling off or unable to operate normally.
[0089] The engine failure level is represented by Level, Level = {I, II, III, IV, V}.
[0090] (2) Sorting out engine components and fault forms: A certain type of turbine engine can be divided into four modules, and the fault types can be divided into 10 types: wear, corrosion, cracks, tears, fractures, deformation, looseness, torque transmission surface damage, and blockage of main leakage. According to the above division, the fault forms corresponding to different components in different modules are summarized.
[0091] Engine module M = {A, B, C, D}, where A is the compressor module, B is the combustion module, C is the turbine module, and D is the accessory drive module. The specific components of the engine are represented by M ij, where i represents the serial number of the component to which the component belongs, and j represents the serial number of the component. For example, the intake casing component belonging to the compressor module and supercharger assembly is represented by A 11 .
[0092] The fault types F of a certain type of engine include: wear, corrosion, crack, tear, break, deformation, looseness, torque transmission surface damage, blockage, leakage. For specific components, the fault type is expressed as Where M represents the module to which it belongs, i represents the serial number of the component to which the component belongs, j represents the serial number of the component in the component, and f represents the serial number of the fault type. Figure 2-Figure 5 , Figure 2-Figure 5 The components and detailed fault detection rules of compressor module A, combustion module B, turbine module C, and accessory drive module D are described in detail. The module components and component names of different modules are introduced in detail. The related fault types that exist are listed and matched with different engine fault levels, and the fault detection content required for different engine fault levels is explained. When the corresponding fault form exists for the fault level, fill in the corresponding space in the table with "√", and fill in the table with "\" when the fault form does not exist.
[0093] (3) Configure the corresponding rules between engine fault forms and detection resources. For the construction of detection resource configuration rules, the fault types are divided into different types according to their severity and the functions of the detection resources, and the number of detection resources required is determined to complete the construction of the detection resource configuration rules.
[0094] There are three main types of detection resources R = {X, Y, Z}: detection resource X is responsible for scanning the narrow and hidden places in the engine with the camera carried. These images are then transmitted back to the detection center through network communication. The technicians of the detection center judge the engine failure according to the impact; detection resource Y should be flexible enough to pass through the narrow aperture and scan and inspect the pipelines or seals, mainly for valves, seals and pipelines; detection resource Z is used to carry the rest of the detection resources into the engine. After completing the scanning and detection task, the crawler robot is recovered and taken out of the engine.
[0095] The fault form corresponding to the fault level is as follows Figure 2-Figure 5 As shown. The spaces filled with "\" indicate that there is no such type of fault for this level of fault, and the remaining spaces filled with "√" indicate that there is such type of fault for this level of fault, and the number of different types of detection robots that match it can be calculated by Indicates, where R indicates the detection resource type, Indicates the fault type, and L indicates the fault level. For example, Figure 2The amount of resource X required to wear the intake casing in the compressor module is used Indicates that the quantity of resource Y can be expressed as Indicates that the quantity of resource Z is expressed as express.
[0096] Step 2: Evaluate the current state of the system. The specific implementation is as follows:
[0097] For n tasks to be detected F n ,n=1,2,3…,N and required detection resources m=1,2,3…,M,i=1,2,3…,I,Firstly, the current load status of the detection system is evaluated from the aspects of time urgency and resource scarcity.
[0098] (1) In the absence of other tasks, the task to be detected F n The larger the ratio of the detection time to the available time, the more urgent the task; in the case of multiple detection tasks, the deadline is less than the task F. n The task with the deadline will exist with F n Competition for detection resources, so the number of tasks to be detected at time t is F n The time urgency is recorded as:
[0099]
[0100] Among them, the deadline is less than the task F n The set of tasks with deadlines is denoted as Represents the task to be detected F n The detection time, Represents the task to be detected F n The test preparation time, Represents the task to be detected F n The deadline of , t represents the current time.
[0101] (2) For the task to be detected F n The resource scarcity of the mth detection resource is recorded as:
[0102]
[0103] Among them, the detection resource set is R = {r m ,m=1,2,3…,M}, m represents the type of detection resource, Represents the task to be detected F n The number of the mth detection resources required.
[0104] Step 3: Identify the detection scheduling scenario based on the status evaluation, which is implemented as follows:
[0105] The detection scheduling scenario is represented by DC and is divided into 5 categories: low-frequency scenario DC 1 , medium and low frequency DC 2 , medium frequency scenario DC 3 , medium and high frequency scenario DC 4 , high frequency situational DC 5 The urgency of the detection task and the scarcity of detection resources are used as the criteria for dividing the detection scheduling scenarios. The current detection scheduling scenarios are identified by solving the time urgency of the detection task and the scarcity of detection resources.
[0106] The overall time urgency U is used to detect the task F n The time urgency is expressed in the form of a collection:
[0107]
[0108] Overall resource scarcity S m , if S m If it is greater than 1, it means that the current m-th detection resource cannot meet the needs of the detection task. m Denoted as:
[0109]
[0110] The classification criteria for the detection scheduling scenario are: when the time urgency of each detection task is less than the threshold u*, and the resource scarcity S m are all less than 1, at this time, the system resources are sufficient, the task can be executed in sufficient time, and the detection scheduling situation is judged as a low-frequency situation. When the proportion of detection tasks with time urgency greater than the threshold u* is greater than 10%, and the resource scarcity S t Both are less than 1. At this time, the system resources are sufficient and the task execution time is relatively ample, and the detection scheduling situation is judged as a medium-low frequency situation. When the proportion of detection tasks with time urgency greater than the threshold u* is greater than 30%, and the maximum resource scarcity is 1, the system resources are relatively sufficient and the task execution time is relatively ample, and the detection scheduling situation is judged as a medium frequency. When the proportion of detection tasks with time urgency greater than the threshold u* is greater than 50%, and there are resource scarcity greater than 1, the situation is judged as medium-high frequency. When the proportion of detection tasks with time urgency greater than the threshold u* is greater than 50%, and the resource scarcity is greater than 1, the situation is judged as high frequency. The value of the threshold u* is determined by the actual operator, and the reference value is 1. The scheduling situation division criteria are as follows: Figure 6 As shown in the figure, the detection scheduling scenarios are divided into five different scenarios based on task urgency and resource scarcity: low frequency, medium-low frequency, medium frequency, medium-high frequency, and high frequency. The requirement of the overall task urgency in the low frequency scenario is the time urgency of each detection task. are all less than the threshold u*. The requirement for resource scarcity is the sum of the ratio of the demand for the mth resource of each detection task to the existing quantity S m ,m=1...M are all less than 1. The requirements for overall task urgency in medium and low frequency situations are The number of tasks greater than the threshold u* accounts for more than 10% of the total number of tasks. The requirement for resource scarcity is the sum of the ratio of the mth resource demand of each detection task to the existing quantity S m ,m=1...M are all less than 1. The requirement of the medium-frequency scenario for the overall task urgency is The number of tasks greater than the threshold u* accounts for more than 30% of the total number of tasks. The requirement for resource scarcity is the sum of the ratio of the mth resource demand of each detection task to the existing quantity S m The maximum value of is equal to 1. The requirement for overall task urgency in medium and high frequency scenarios is The number of tasks greater than the threshold u* accounts for more than 50% of the total number of tasks. The requirement for resource scarcity is the sum of the ratio of the mth resource demand of each detection task to the existing quantity S m ,m=1...M exists in the case of being greater than 1. The requirement of high-frequency situations for the overall task urgency is time urgency The number of tasks greater than the threshold u* accounts for more than 50% of the total number of tasks. The requirement for resource scarcity is the sum of the ratio of the mth resource demand of each detection task to the existing quantity S m ,m=1...M are all greater than 1.
[0111] Step 4: Configure the detection scheduling mode and its scheduling rules according to the situation category. The specific implementation is as follows:
[0112] Based on five detection scenarios DC i Different scheduling modes are configured for different characteristics. Since the detection scenarios are divided from two aspects: time urgency and resource scarcity, the present invention considers configuring the scheduling mode SM from two aspects: improving resource utilization and task processing efficiency. i ,i=1,2…,6. Scheduling rules corresponding to the scheduling mode j=1,2…,m(i). m(i) is the scheduling mode SM i The corresponding number of dispatch rules, such as Figure 7 As shown, Figure 7The scheduling mode is set from the two aspects of improving resource utilization and task processing efficiency, and the scheduling mode corresponds to the corresponding scheduling rules. Consider the characteristics of the detection scenario to match the scheduling rules. The task urgency and resource scarcity of the low-frequency scenario are low, and the scheduling rules that can improve resource utilization are mainly considered. The corresponding blanks in the table are filled with "√"; the task urgency of the medium-low frequency scenario is higher than that of the low frequency. When the scheduling rules that improve resource utilization are mainly considered, some scheduling rules that improve task processing efficiency are also adopted; the task urgency and resource scarcity of the medium frequency are both improved. When the scheduling rules that improve resource utilization are considered, the scheduling rules that improve task processing efficiency are also adopted; the task urgency of the medium-high frequency scenario is higher than that of the medium frequency. When the scheduling rules that improve resource utilization are considered, the scheduling rules that improve task processing efficiency are also adopted; the task urgency of the high-frequency scenario is higher than that of the medium frequency. When the scheduling rules that improve resource utilization are considered, the scheduling rules that improve task processing efficiency are also adopted; the task urgency of the high-frequency scenario is high, and the scheduling rules that improve task processing efficiency are mainly considered. Fill in the corresponding blanks in the table with "√".
[0113] (1) Scheduling mode considering resource utilization:
[0114] Scheduling mode SM 1 : Reduce costs
[0115] The total cost of using the inspection robot is the sum of the depreciation and electricity costs of each robot during use and the standby cost during the period of non-use.
[0116]
[0117] Among them, a h represents the unit cost of the h-type detection robot during its use, b h represents the unit standby cost of h type detection robots during the period of non-use, and then the usage cost of each detection robot is summed up.
[0118] The corresponding scheduling rule is that the higher the detection cost, the higher the priority Denoted as:
[0119]
[0120] Scheduling mode SM 2 : Reduce the average waiting time.
[0121] The average waiting time is the average time each detection task waits in the detection system from the time it is received by the system and the detection is started.
[0122]
[0123] in, Representative Task F n The start detection time, n=1,2,...N represents task Fn The arrival time of N is the total number of tasks to be detected.
[0124] The corresponding dispatching rule is that the earlier you arrive, the higher your priority. Denoted as:
[0125]
[0126] Scheduling mode SM 3 :Improve the utilization rate of detection resources
[0127] The detection resource utilization rate is the ratio of the average usage time of each detection tool to the total detection time:
[0128]
[0129] Where L is the total number of detection resources, Represents the detection resource of type h Use time, Represents the detection task F n The end detection time, The maximum value of indicates the end time of the entire detection process. Represents the detection task F n The start detection time, The minimum value of indicates the starting time of the entire detection process.
[0130] The corresponding scheduling rule is that the longer the detection time, the higher the priority Tasks with earlier deadlines have higher priority Denoted as:
[0131]
[0132]
[0133] (2) Scheduling mode considering task processing efficiency:
[0134] Scheduling mode SM 4 : Reduce the total completion time.
[0135] The total completion time is the total time to complete all inspection tasks:
[0136]
[0137] in, Represents the detection task F n The end detection time, Represents the detection task F n The start detection time.
[0138] The corresponding scheduling rule is that the shorter the detection time, the higher the priority Denoted as:
[0139]
[0140] in, Represents the detection task F n The time required for detection, Represents the detection task F n The time required for preparation.
[0141] Scheduling mode SM 5 : Reduce the proportion of delayed tasks
[0142]
[0143] Among them, N tardiness represents the number of delayed tasks, and N represents the total number of detection tasks.
[0144] The corresponding scheduling rule is that the fewer detection resources are needed, the higher the priority The shorter the detection time, the higher the priority Denoted as:
[0145]
[0146]
[0147] in, Represents the detection task F n The time required for detection, Represents the detection task F n The number of detection resources of type h required.
[0148] Scheduling mode SM 6 :Reducing the average delay time:
[0149]
[0150] in, Represents the detection task F n The end detection time, Represents the detection task F n deadline of the task.
[0151] The corresponding scheduling rule is that the shorter the detection time, the higher the priority The more time-sensitive the situation, the higher the priority. Denoted as:
[0152]
[0153]
[0154] in, Represents the detection task F n The time required for detection, Represents the detection task F n The time required for preparation, Represents the detection task F n The task deadline, Represents the detection task F n The arrival time of the task.
[0155] (3) Configure the scheduling mode according to the characteristics of the detection scenario; consider the classification criteria of the detection scenario. The system load in the low-frequency scenario is low and the service capacity is sufficient. As the scenario increases, the system load becomes heavier and the service capacity becomes scarcer. When configuring the scheduling mode, since the detection pressure in low-frequency and medium-low-frequency scenarios is small, the main consideration is to improve the utilization rate of detection resources. As the scenario increases, the detection pressure increases. Therefore, the focus of scheduling mode selection is transferred from improving the utilization rate of detection resources to improving the processing efficiency of detection tasks. The corresponding relationship between each level of detection scenario and scheduling mode is as follows: Figure 2 As shown, the detection scenario and dispatch rule correspondence matrix RM is generated again, and the RM generation formula is as follows:
[0156]
[0157]
[0158] Where p is the number of detection scheduling scenarios, m(i) is the scheduling mode SM i The corresponding number of scheduling rules.
[0159] Step 5: Generate task priority ranking, which is implemented as follows:
[0160] (1) Determine the weight of each dispatch rule
[0161] After matching the scheduling mode and scheduling rules, you also need to determine the priority of each detection task. Since each detection scenario corresponds to multiple scheduling rules, when the scheduling rules are combined, you need to set the importance of each scheduling rule, that is, their weight coefficient. You can set the weight of each scheduling rule by user definition or system recommendation.
[0162] The recommended method of the system is to balance the importance of each scheduling rule and obtain the weight matrix W b :
[0163]
[0164]
[0165] Among them, m is the total number of scheduling rules, i is the current system scheduling context, and the user-defined method is determined by the user according to actual needs. For the value of n=1,2,…K, the constraints that need to be noted are:
[0166]
[0167] (2) Generate a comprehensive priority ranking for cluster detection tasks
[0168] Generate a comprehensive priority ranking for fleet detection tasks based on a given method. Based on the dispatch rule weights determined in the previous step, combined with multiple dispatch rules under different scenarios, use existing TOPSIS approximate ideal solution sorting method, fuzzy comprehensive evaluation method, RSR rank sum ratio comprehensive evaluation method and other methods to generate a comprehensive priority ranking for fleet detection tasks. Taking the TOPSIS method as an example, if the system is selected to assign dispatch target weights in the previous step, the standard TOPSIS method is used for priority sorting; if the user-defined dispatch target weights are selected, the weighted TOPSIS method can be used for priority sorting.
[0169] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.
[0170] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for swarm detection situation recognition and task generation, characterized in that The steps include: Step 1: According to the classification of aircraft engine fault severity, determine the required detection content, detection resources and detection time, and build detection resource configuration rules for aircraft engine fault detection; Step 2: Evaluate the fault-to-be-detected tasks The current state, based on the received Tasks to be tested And the required detection resources determined by the configuration rules in step 1 , determine the tasks to be detected The two task states are time urgency and resource scarcity; Step 3: Based on the fault detection tasks and fault detection resources in step 2, determine the overall state of the fault detection system, identify the fleet detection task scheduling scenario, and divide the fault detection system detection scenarios into five scenarios: low frequency, medium-low frequency, medium frequency, medium-high frequency, and high frequency according to the two indicators of time urgency and resource scarcity. Based on the state evaluation results, identify the fleet detection task scheduling scenario; Step 4: Based on the problem background, comprehensively consider the goals of improving resource utilization and task processing efficiency, and design the detection task scheduling mode and its scheduling rules; then, according to the different characteristics of the five detection scenarios in step 3, determine the scheduling mode and its scheduling rules required for different detection scenarios; Step 5: Based on the scheduling mode in step 4, the weights of each scheduling rule of the required scheduling mode are set in a user-defined or recommended manner, and a comprehensive priority ranking for any fleet detection task under the constraints of the corresponding scheduling rule combination is generated, that is, the detection task requirement content and comprehensive priority generation for the fleet detection scenario are completed; The step 1 specifically includes: The engine failure degree is divided into five levels: I, II, III, IV, and V, corresponding to minor consequences, significant consequences, serious consequences, major consequences, and catastrophic consequences, respectively. According to the engine component modules, the failure forms corresponding to different components of different modules are determined, and corresponded to the engine failure level. The fault detection content corresponding to the engine failure level is established. The fault types are wear, corrosion, cracks, tears, fractures, deformation, looseness, torque transmission surface compression, and blockage of main leakage. According to the severity of the fault type and the function of the detection resources, the detection task requirements and detection contents of different fault types are determined. The detection resource configuration rules for the detection time, type and number of detection robots of aircraft engines corresponding to different levels of engine failure degree and different fault types are completed. The step 2 specifically includes: For n tasks to be detected And testing resources ,First, the status of the task to be detected is evaluated from two aspects of time urgency and resource scarcity; (1) When there are multiple detection tasks, the deadline is less than the number of tasks to be detected. The tasks with deadlines will exist with Competition for detection resources, so the number of tasks to be detected at time t The time urgency is recorded as: Among them, the deadline is less than the task The set of tasks with deadlines is denoted as , Indicates the task to be detected The detection time, Indicates the task to be detected The test preparation time, Indicates the task to be detected The deadline of , t represents the current time; (2) For the task to be detected The resource scarcity of the mth detection resource is recorded as: Among them, the detection resource set is , m represents the type of detection resource, Indicates the task to be detected The number of the mth detection resources required; The step 3 is specifically implemented as follows: (1) Overall time urgency U for detection tasks The time urgency is expressed in the form of a collection: Overall resource scarcity ,like If it is greater than 1, it means that the current m-th detection resource cannot meet the needs of the detection task. Denoted as: (2) The time urgency of the detection task and the scarcity of detection resources are used as the criteria for dividing the detection scheduling scenarios. By solving the time urgency of the detection task and the scarcity of detection resources, the current detection scheduling scenario is identified; the i-th detection scheduling scenario is used Indicates that it is divided into five categories: low-frequency situations , medium and low frequency scenarios , Medium Frequency Scenario , medium and high frequency scenarios , high frequency scenarios ; The criteria for dividing the detection scheduling scenarios are: when the time urgency of each detection task is less than the threshold , and resource scarcity are all less than 1, at this time, the system resources are sufficient, the task can be executed in sufficient time, and the detection scheduling situation is judged as a low-frequency situation; when the time urgency is greater than the threshold The proportion of detection tasks is greater than 10%, and the scarcity of resources are all less than 1. At this time, the system resources are sufficient and the task execution time is relatively ample. The detection and scheduling situation is judged as a medium-low frequency situation. When the time urgency is greater than the threshold The proportion of detection tasks is greater than 30%, and the maximum resource scarcity is 1. At this time, the system resources are sufficient and the task execution time is sufficient. The detection scheduling situation is judged as medium frequency. When the time urgency is greater than the threshold If the proportion of detection tasks is greater than 50% and there is a resource scarcity greater than 1, the situation is judged as medium-high frequency; when the time urgency is greater than the threshold If the proportion of detection tasks is greater than 50% and the resource scarcity is greater than 1, the situation is judged as high frequency; The step 4 specifically includes: Based on five detection scenarios Different scheduling modes are configured for different characteristics. The detection scenarios are divided into two aspects: time urgency and resource scarcity. The scheduling mode is configured from two levels: improving resource utilization and task processing efficiency. The scheduling mode is expressed as , scheduling mode The corresponding scheduling rule is expressed as , For scheduling mode The number of corresponding dispatch rules; (1) Consider resource utilization scheduling mode: Scheduling Mode : Reduce costs The total cost of using the inspection robot is the sum of the depreciation and electricity cost of each robot during use and the standby cost during the non-use period: in, represents the unit cost of the h-type detection robot during its use, represents the unit standby cost of h type detection robots during the unused period, and then sums the usage cost of each detection robot, the scheduling mode The corresponding scheduling rule is that the higher the detection cost, the higher the priority , as shown in formula (6); Scheduling Mode : Reduce the average waiting time The average waiting time is the average time each detection task waits in the detection system from the time it is received by the system to the time it starts to execute the detection: in, Representative tasks The start detection time, Representative tasks Arrival time, scheduling mode The corresponding dispatching rule is that the earlier you arrive, the higher your priority , as shown in formula (8); Scheduling Mode :Improve the utilization rate of detection resources The detection resource utilization rate is the ratio of the average usage time of each detection tool to the total detection time: in, represent Type of detection resource Use time, Representative detection tasks The end detection time, The maximum value of indicates the end time of the entire detection process. Representative detection tasks The start detection time, The minimum value of indicates the start time of the entire detection process, the scheduling mode The corresponding scheduling rule is that the longer the detection time, the higher the priority , The earlier the deadline, the higher the priority , as shown in formula (10), (11); (2) Scheduling mode considering task processing efficiency: Scheduling Mode :Reduce the total completion time The total completion time is the total time to complete all inspection tasks: in, Representative detection tasks The end detection time, Representative detection tasks Start detection time, scheduling mode The corresponding scheduling rule is that the shorter the detection time, the higher the priority , as shown in formula (13); Scheduling Mode : Reduce the proportion of delayed tasks: in, Represents the number of delayed tasks. Represents the total number of detection tasks and the scheduling mode The corresponding scheduling rule is that the fewer detection resources are needed, the higher the priority , the shorter the detection time, the higher the priority , as shown in formula (15), (16); Scheduling Mode :Reducing the average delay time: in, Representative detection tasks The end detection time, Representative detection tasks Task deadline, scheduling mode The corresponding scheduling rule is that the shorter the detection time, the higher the priority , the greater the time urgency, the higher the priority , as shown in formula (18), (19); (3) Configure the scheduling mode according to the characteristics of the detection scenario. When configuring the scheduling mode, since the detection pressure in low-frequency and medium-low-frequency scenarios is small, consider improving the utilization rate of detection resources. As the scenario improves, the detection pressure increases, and the focus of scheduling mode selection shifts from improving the utilization rate of detection resources to improving the processing efficiency of detection tasks. The scheduling rules corresponding to the low-frequency detection scenario are: , , , , the dispatching rules corresponding to the low-medium frequency detection scenario are: , , , ,The scheduling rules corresponding to the intermediate frequency detection scenario are: , , , , , ,The scheduling rules corresponding to the medium and high frequency detection scenarios are: , , , , ,The scheduling rules corresponding to the high-frequency detection scenario are: , , , , The higher the detection cost, the higher the priority. The first to arrive will have priority. The longer the detection time, the higher the priority. The earlier the deadline, the higher the priority. The shorter the detection time, the higher the priority. The fewer resources required for testing, the higher the priority. The shorter the detection time, the higher the priority. The shorter the detection time, the higher the priority. ,The greater the time urgency is, the higher the priority is, and the corresponding relationship matrix RM between the detection scenarios and the scheduling rules is generated.
2. The method for swarm detection situation identification and task generation according to claim 1, characterized in that: The step 5 specifically includes: (1) Determine the weight of each dispatch rule. After matching the dispatch mode and dispatch rules, it is also necessary to determine the priority of each detection task. Since each detection scenario corresponds to multiple dispatch rules, when the dispatch rules are combined, it is necessary to set the importance of each dispatch rule, that is, the weight coefficient. The weight of each dispatch rule is set by user definition or system recommendation. The user definition method also needs to adopt corresponding constraints; (2) Based on the dispatch rule weights determined in step (1), combined with multiple dispatch rules under different scenarios, the fleet detection tasks are comprehensively prioritized.
Citation Information
Patent Citations
Abnormal mainframe group detection method based on dynamic graph
CN107528734A
Manufacturing service adaptive scheduling method based on supply-demand relationship
CN113240287A
Cluster detection equalization scheduling method considering state transition
CN113743728A
Methods and systems for providing a global view of airline operations
US20080010107A1
Planning and scheduling of maintenance, repair, and overhaul services
US7937280B1
