A collaborative processing system and method for distributed heterogeneous drone clusters

Abstract

The present invention discloses a collaborative processing system and method for distributed heterogeneous drone clusters, and relates to the technical field of drone cluster control. The system comprises an information collection module for collecting drone data and receiving mission information; a data processing module for assigning a trust index to each drone; assigning tasks to the drones based on the drone's capability vector identification table and trust index; recording mission data during the drone's mission execution, calculating the mission capability factor, and updating the trust index based on the mission capability factor. The present invention quantifies the stability of drone use through the trust index, and updates the trust index when the drone subsequently executes a mission, thereby more accurately representing the stability of the drone. Furthermore, when assigning tasks, tasks with higher importance are preferentially assigned to drones with higher trust indicators, thereby improving the efficiency of task completion and improving the allocation mechanism.

Description

Technical Field

[0001] The present invention relates to the field of drone cluster control technology, and specifically to a collaborative processing system and method for distributed heterogeneous drone clusters. Background Art

[0002] The collaborative processing system of distributed heterogeneous drone clusters integrates drones of different types and functions to achieve efficient information interaction and task collaboration among multiple platforms. Its significance lies in breaking through the limitations of single drones in perception range, computing power and functions, and improving the efficiency and robustness of task execution in complex environments. Its role is reflected in optimizing resource allocation and reducing the risk of single point failures through distributed architecture, and supporting dynamic task reorganization and real-time decision-making, thereby significantly enhancing the system adaptability and combat / service effectiveness in scenarios such as military reconnaissance, disaster relief, and logistics distribution, providing key technical support for intelligent and unmanned cluster operations.

[0003] The mutual cooperation between drones of different models needs to be controlled by a control system, such as a drone cluster collaborative control method and system with patent publication number CN119472471A, which includes the following steps: constructing a drone cluster topology structure; assigning drones to corresponding clusters based on the location of each drone, and determining the main cluster head and secondary cluster head; in the drone cluster, disconnecting the direct communication links between member drones in the cluster and the direct communication links with all drones in other clusters, setting a cluster boundary range centered on the main cluster head, and when executing a task, the task data is transmitted to the main cluster head and the secondary cluster head at the same time. When the main cluster head fails, the secondary cluster head takes over and a new secondary cluster head is selected; using Kalman filtering to predict the drone motion state, and using the isolation forest algorithm to calculate the anomaly score. If the anomaly score exceeds the preset score threshold, it is determined that the cluster has an anomaly and the cluster structure is readjusted to avoid instability of the entire drone cluster due to the failure of individual drones.

[0004] A heterogeneous drone cluster includes drones of various types and functions. The quality of each drone cannot be guaranteed, so the stability of each drone also varies. The above method and the method based on the same principle only consider the drone's own data, such as power, speed, and hardware configuration, when assigning tasks to drones. Only when a failure occurs when the drone is performing a task can it be remedied. Failure prevention cannot be carried out in advance, and the stability of the drone's use cannot be considered within the allowable error of task execution. Therefore, the task assignment of drones is not perfect. Summary of the Invention

[0005] The purpose of the present invention is to provide a collaborative processing system and method for distributed heterogeneous drone clusters to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a collaborative processing system for a distributed heterogeneous UAV cluster, comprising:

[0007] Information collection module: Based on the data records of each drone, a capability vector identification table is established, including the drone's identification code, location, flight time and hardware data;

[0008] Receive task information, including task content and task level;

[0009] Data processing module:

[0010] Preprocess different types of data in the capability vector identification table separately and assign a trust index to each drone;

[0011] According to the UAV capability vector identification table and trust index, the UAV is assigned tasks according to the maximum value allocation method;

[0012] During the UAV's mission execution, mission data is recorded, including mission level, motion trajectory, number of failures, and completion time. The mission capability factor is calculated using the completion calculation method, and the trust index is then updated based on the mission capability factor.

[0013] Information output module: Establish a drone trust index data table and calculate the drone anomaly value based on the fluctuation of the data table. When the anomaly value is greater than the preset threshold, the drone will be grounded and technical personnel will be notified for maintenance.

[0014] Preferably, the completion degree calculation method includes:

[0015] Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time;

[0016] The calculation formula for task completion is:

[0017]

[0018] The calculation formula for task failure is:

[0019]

[0020] Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the estimated moving distance.

[0021] Preferably, the completion degree calculation method includes:

[0022] Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time;

[0023] The calculation formula for task completion is:

[0024] ,

[0025]

[0026] The calculation formula for task failure is:

[0027]

[0028] Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the expected moving distance, f max represents the maximum number of failures observed in history, Δt max Denotes the absolute value of the maximum time difference observed in history, Δd max It represents the absolute value of the maximum distance difference observed in history, and corr represents the correction factor.

[0029] Preferably, the updating of the trust indicator according to the task capability factor specifically includes:

[0030] Record the mission capability factor of the UAV after performing several missions;

[0031] Set a fixed window size, which determines how many recent tasks are considered when calculating the updated confidence metric;

[0032] When the number of completed tasks is less than the window size, the task capability factors of all completed tasks and the average value of the current trust index are calculated, and the trust index is updated;

[0033] When the number of completed tasks reaches or exceeds the window size, each time a new task is completed, the oldest task is removed from the currently considered window and the new task is added to the window. Then, the average of the capability factors and the current trust index of all tasks in the window is calculated to update the trust index.

[0034] Preferably, the updating of the trust indicator according to the task capability factor specifically includes:

[0035] Record the mission capability factor of the UAV after performing several missions;

[0036] Assign a weight to the task capability factor of each task. The closer the task is to the time, the greater the weight of the task capability factor. The sum of the weights is 1.

[0037] Perform weighted summation on the task capability factors to obtain the update index, then calculate the average of the update index and the trust index to update the trust index.

[0038] Preferably, the maximum value allocation method specifically includes:

[0039] Drones are graded based on their trust indicators, with the number of grades matching the number of mission levels.

[0040] Assign tasks from high to low according to their levels;

[0041] When allocating tasks at the target level, the remaining drones from the previous level compete for allocation with the drones at the target level, and the remaining drones are delegated to the next level;

[0042] When assigning tasks at a target level, if the number of drones is insufficient, drones with the highest trust index can be drawn from the next level to supplement the number;

[0043] Then assign the next level of tasks until all tasks or drones are assigned.

[0044] Preferably, the trust index assigned to each drone specifically includes:

[0045] When a new drone joins the cluster, its initial trust index is set to 0.5;

[0046] Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate;

[0047] Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1 and select the smaller value as the trust index of the newly added drone.

[0048] Preferably, the trust index assigned to each drone specifically includes:

[0049] When a newly added drone has the same model as the drones in the cluster, all drones of the same model are first selected and the trust index and number of completed tasks of each drone are recorded;

[0050] The ratio of the number of tasks completed by the target drone to the number of tasks completed by all drones is used as the trust index weight of the target drone. The trust index of each drone is weighted and summed to obtain the initial trust index of the newly added drone.

[0051] Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate;

[0052] Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1 and select the smaller value as the trust index of the newly added drone.

[0053] Preferably, the calculating of the drone abnormal value according to the fluctuation of the data table specifically includes:

[0054] Record the updates of the drone's trust indicators for several times and calculate the average rate of change;

[0055] Calculate the absolute value of the change rate of the most recent trust indicator, and take the ratio of the absolute value of the change rate to the average change rate as the outlier;

[0056] When the outlier value is less than 0.5 or greater than 1.5, the drone will be grounded and the technician will be notified for repair.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] By assigning drones a trust index, the stability of drone use is quantified. When the drone performs subsequent tasks, the trust index is updated to more accurately represent the stability of the drone. When assigning tasks, more important tasks are assigned to drones with higher trust indexes first, thereby improving the efficiency of task completion and making the allocation mechanism more perfect.

[0059] At the same time, the workload indicator is used to quantify the working status of the current drone cluster. When assigning the drone trust indicator, it is adjusted according to the workload indicator. When there are many tasks and the work progress is tight, the trust indicator of the newly added drone is increased to relieve the task pressure. Conversely, the trust indicator of the newly added drone is reduced to improve the stability of task completion.

[0060] In addition, when drones of the same model join the cluster, introducing historical data of drones of the same model and estimating the initial trust index can improve the accuracy of the initial trust index, thereby increasing the stabilization speed of the trust index and making full use of the working capabilities of the newly added drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the collaborative processing flow of the distributed heterogeneous UAV cluster of the present invention;

[0062] Figure 2 Schematic diagram of the process of the completion degree calculation method of the present invention;

[0063] Figure 3Schematic diagram of the update process of the trust indicator in the first embodiment of the present invention;

[0064] Figure 4 Schematic diagram of the update process of the trust indicator in the third embodiment of the present invention;

[0065] Figure 5 Schematic diagram of the maximum value allocation method in Example 1 of the present invention;

[0066] Figure 6 A schematic diagram of the process of assigning a trust index to a drone in the present invention;

[0067] Figure 7 Schematic diagram of a control panel of the processing system of the present invention. DETAILED DESCRIPTION

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0069] In this application, for ease of understanding, the method steps used do not need to be executed in the order of the steps in this embodiment during actual operation. In other embodiments, these steps may be performed simultaneously or in a different order.

[0070] Example 1:

[0071] A heterogeneous drone cluster includes drones of various types and functions. The quality of each drone cannot be guaranteed, so the operational stability of each drone also varies to a certain extent. When assigning tasks to drones, the operational stability of the drone is taken into account within the allowable error of task execution. Therefore, the task assignment of drones is not perfect.

[0072] like Figure 1-Figure 3 、 Figure 5 and Figure 6 As shown, the present invention provides a technical solution: a collaborative processing system for distributed heterogeneous UAV clusters, comprising:

[0073] Information collection module: Based on the data records of each drone, a capability vector identification table is established, including the drone's identification code, location, flight time and hardware data;

[0074] Receive task information, including task content and task level;

[0075] Data processing module:

[0076] Preprocess different types of data in the capability vector identification table and assign a trust index to each drone;

[0077] According to the UAV capability vector identification table and trust index, the UAV is assigned tasks according to the maximum value allocation method;

[0078] During the UAV's mission execution, mission data is recorded, including mission level, motion trajectory, number of failures, and completion time. The mission capability factor is calculated using the completion calculation method, and the trust index is then updated based on the mission capability factor.

[0079] Information output module: Establish a drone trust index data table and calculate the drone anomaly value based on the fluctuation of the data table. When the anomaly value is greater than the preset threshold, the drone will be grounded and technical personnel will be notified for maintenance.

[0080] It should be noted that drone cluster communications usually adopt a distributed self-organizing network (Ad Hoc Network) architecture, which does not rely on fixed infrastructure and supports dynamic topology and temporary networking. It is an existing technology and will not be described in detail here.

[0081] like Figure 2 As shown, the completion calculation methods include:

[0082] Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time;

[0083] The calculation formula for task completion is:

[0084]

[0085] The calculation formula for task failure is:

[0086]

[0087] Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the estimated moving distance.

[0088] Number of failures (f):

[0089] Direct impact: The greater the number of failures, the lower the mission capability factor.

[0090] Quantification method: The negative impact on the mission capability factor is quantified by directly multiplying the number of failures f by the weight coefficient 0.5.

[0091] The difference between the actual completion time and the expected completion time (Δt):

[0092] Direct impact: The larger the difference, the lower the task capability factor.

[0093] Quantification: The negative impact on the mission capability factor is quantified by multiplying the square of the time difference by a weight coefficient of 0.3. The square term is used to amplify the impact of large differences.

[0094] The difference between the actual moving distance and the expected moving distance (Δd):

[0095] Direct impact: The larger the difference, the lower the task capability factor.

[0096] Quantification: The negative impact on the task capability factor is quantified by multiplying the square of the distance difference by a weight factor of 0.2. A square term is also used to amplify the impact of larger distance differences.

[0097] Mission level (g):

[0098] Indirect impact: The higher the task level, the greater the positive impact on the task capability factor (under the same conditions, the task capability factor is higher).

[0099] Quantification: The positive impact of mission level on mission capability factor is quantified using the mission level adjustment factor (1 + 0.4(g-1)). Here, 0.4 is the mission level adjustment factor, and (g-1) converts the mission level from 1 to the adjustment factor. When g = 1, the adjustment factor is 1, which has no effect on the mission capability factor. When g > 1, the adjustment factor is greater than 1, which increases the mission capability factor.

[0100] It should be noted that for ease of understanding, simulated data is used as follows:

[0101] When a drone performs a certain mission, the actual record is as follows:

[0102] Number of failures (f): 1

[0103] Actual completion time (T r ): 120 minutes

[0104] Estimated completion time (T e ): 100 minutes

[0105] Mission level (g): 3 levels (1 is the lowest level) (set by technical personnel, no specific restrictions)

[0106] Actual moving distance (D r ): 150 km

[0107] Estimated moving distance (D e ): 140 km

[0108] Calculate Δt and Δd =:

[0109] Δt=T r −T e =120−100=20 minutes

[0110] Δd=D r −D e =150−140=10 km

[0111] Substituting this value into the formula yields Tcf ≈ 0.719. Tcf ranges from 0 to 1, with values ​​closer to 1 indicating better performance on the target mission and a more significant improvement in the trust index.

[0112] like Figure 3 As shown, the trust indicator is updated according to the task capability factor, specifically including:

[0113] Record the mission capability factor of the UAV after performing several missions;

[0114] Set a fixed window size, which determines how many recent tasks are considered when calculating the updated confidence metric;

[0115] When the number of completed tasks is less than the window size, the task capability factors of all completed tasks and the average value of the current trust index are calculated, and the trust index is updated;

[0116] When the number of completed tasks reaches or exceeds the window size, each time a new task is completed, the oldest task is removed from the currently considered window and the new task is added to the window. Then, the average of the capability factors and the current trust index of all tasks in the window is calculated to update the trust index.

[0117] It should be noted that for ease of understanding, simulated data is used as follows:

[0118] Assume that drone 001's original trust index T0 = 0.8. It has completed five missions, with mission capability factors Tcf1 = 0.7, Tcf2 = 0.85, Tcf3 = 0.9, Tcf4 = 0.75, and Tcf5 = 0.95. Set the window size k = 3 (selecting the mission capability factors after completing three missions). (The specific window size is not limited.)

[0119] Trust Indicator (T) Update:

[0120] When n=3 (complete the first 3 tasks):

[0121] T3=0.25×(0.7+0.85+0.9+0.8)=0.8125

[0122] When n=4 (4th mission completed, Tcf1 removed, Tcf4 added):

[0123] T4=0.25×(0.85+0.9+0.75+0.8)=0.825

[0124] When n=5 (the fifth mission is completed, Tcf2 is removed and Tcf5 is added):

[0125] T5=0.25×(0.9+0.75+0.95|+0.8)=0.85

[0126] The mission capability factor will be updated after each mission is completed, and the trust index (T) will be updated with different sensitivity levels according to the selected window size to achieve continuous inspection of the stability of the UAV, thereby ensuring the rationality of task allocation.

[0127] refer to Figure 5 , the maximum value allocation method specifically includes:

[0128] Drones are graded based on their trust indicators, with the number of grades matching the number of mission levels.

[0129] Assign tasks from high to low according to their levels;

[0130] When allocating tasks at the target level, the remaining drones from the previous level compete for allocation with the drones at the target level, and the remaining drones are delegated to the next level;

[0131] When assigning tasks at a target level, if the number of drones is insufficient, drones with the highest trust index can be drawn from the next level to supplement the number;

[0132] Then assign the next level of tasks until all tasks or drones are assigned.

[0133] The bidding allocation in the UAV cluster task allocation simulates the market mechanism. Usually, the tenderer (such as the ground control station or the node with task allocation capability in the cluster) publishes the task information. The bidder (the UAV node in the cluster) evaluates the task based on its own information and the resources it carries, participates in the bidding and sends the task request to the tenderer. After the evaluation, the tenderer selects the winner and assigns the task. The specific bidding allocation process and principle are existing technologies and will not be elaborated here.

[0134] It should be noted that for ease of understanding, simulated data is used as follows:

[0135] Assume that the task level is four (set by technical personnel based on actual task requirements. The more levels there are, the more detailed the assignment will be. There is no limit on the specific number);

[0136] According to the trust index of drones, drones are also divided into four levels, including:

[0137] Level 1 interval (0.75,1]): has 13 drones;

[0138] Secondary interval (0.5, 0.75]: with 8 drones;

[0139] Level 3 interval (0.25, 0.5]: has 11 drones;

[0140] Level 4 interval [0,0.25]: has 4 drones;

[0141] There are 6 tasks that need to be executed, including 4 first-level tasks, 1 second-level task and 1 first-level task. First, the first-level drones bid for the 4 first-level tasks. Assume that 3 of the tasks are taken away, and the other tasks require infrared detectors, which the first-level drones do not have. From the second-level drones, they are extracted from high to low according to the trust index of the drones until a drone with an infrared detector is extracted and the task is assigned. Then, the 10 drones without tasks in the first-level drones are transferred to the second level, and the above task allocation method is repeated again to bid for the second-level tasks until all tasks are assigned or there are no drones available.

[0142] refer to Figure 6 , the trust indicators given to each drone include:

[0143] When a new drone joins the cluster, its initial trust index is set to 0.5;

[0144] Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate;

[0145] Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1 and select the smaller value as the trust index of the newly added drone.

[0146] It should be noted that for ease of understanding, simulated data is used as follows:

[0147] Assume that a new drone joins the cluster. Its model is "SkyLink X9-Pro". There is no other drone with the same model in the existing cluster.

[0148] In addition, the total number of existing drones is 24, 22 of which are on missions, and the number of idle drones is 24-22=2. The workload index is 2÷24≈0.083, and the difference between the initial trust index and the workload index is 0.417, which is greater than 0 and less than 1. Therefore, the trust index of the newly added drone is 0.417.

[0149] The workload indicator is used to quantify the working status of the current drone cluster. When assigning the drone trust indicator, it is adjusted according to the workload indicator. When there are many tasks and the work progress is tight, the trust indicator of the newly added drone is increased to relieve the task pressure. Conversely, the trust indicator of the newly added drone is reduced to improve the stability of task completion.

[0150] Calculate drone outliers based on data table fluctuations, including:

[0151] Record the updates of the drone's trust indicators for several times and calculate the average rate of change;

[0152] Calculate the absolute value of the change rate of the most recent trust indicator, and take the ratio of the absolute value of the change rate to the average change rate as the outlier;

[0153] When the outlier value is less than 0.5 or greater than 1.5, the drone will be grounded and the technician will be notified for repair.

[0154] It should be noted that for ease of understanding, simulated data is used as follows:

[0155] Assume that after drone 002 has performed multiple missions, the trust index data is as follows: 0.616, 0.623, 0.632, 0.644, 0.715;

[0157] The average rate of change is 0.0388 (the calculation process is based on existing technology and will not be demonstrated again);

[0158] The absolute value of the most recent rate of change is 0.1102;

[0159] The abnormal value is 0.0388÷0.1102≈0.35<0.5. Therefore, the last trust index change is abnormal. The drone flies to the designated parking location and notifies the technicians for maintenance, thereby eliminating possible subsequent faults in advance and ensuring the smooth completion of the mission.

[0160] Example 2:

[0161] During the UAV's mission execution process, the four parameters of mission level, moving distance, number of failures, and completion time will affect each other. A longer proportional moving distance may result in a slower completion time. When calculating the mission capability factor in Example 1, the correlation possibility between the four parameters was not considered, which may result in the calculated mission capability factor being inaccurate. Based on this, this embodiment provides another completion calculation method to calculate the mission capability factor.

[0162] Completion calculation methods include:

[0163] Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time;

[0164] The calculation formula for task completion is:

[0165] ,

[0166]

[0167] The calculation formula for task failure is:

[0168]

[0169] Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the expected moving distance, f max represents the maximum number of failures observed in history, Δt max Denotes the absolute value of the maximum time difference observed in history, Δd max It represents the absolute value of the maximum distance difference observed in history, and corr represents the correction factor.

[0170] Normalize each parameter to the range [0, 1] (or close to this range) to facilitate comparison and weighting.

[0171] The cubic term is used to further amplify the impact of larger differences, because in practical applications, larger time differences or distance differences may have a more significant impact on reliability.

[0172] It should be noted that for ease of understanding, simulated data is used as follows:

[0173] Number of failures (f): 1;

[0174] Actual completion time (T r ): 120 minutes;

[0175] Estimated completion time (T e ): 100 minutes;

[0176] Mission level (g): Level 3 (assuming level 5 is the highest level);

[0177] Actual moving distance (D r ): 150 km;

[0178] Estimated moving distance (D e ): 140 km;

[0179] The maximum number of failures observed in history (fmax): 5 times;

[0180] The largest time difference (Δt max ): 50 minutes (absolute value);

[0181] Expected maximum distance difference (Δd max ): 30 km (absolute value);

[0182] Calculate the time difference (Δt) and distance difference (Δd):

[0183] Δt=T r −T e =120−100=20 minutes;

[0184] Δd=D r −D e =150−140=10 kilometers.

[0185] Substituting the data into the formula, we can calculate that corr≈1.06825 and Tcf≈0.767.

[0186] The mutual influence between the four parameters is quantified by the correction factor corr, and the task capability factor is corrected to make the calculation result more accurate.

[0187] Example 3:

[0188] When updating the trust indicator in Example 1, the time series is not taken into consideration. The closer the time, the greater the impact of the task capability factor on the trust indicator. Based on this, this example provides another way to update the trust indicator to optimize the impact of the task capability factor on the trust indicator.

[0189] refer to Figure 4 , the trust index is updated according to the task capability factor, including:

[0190] Record the mission capability factor of the UAV after performing several missions;

[0191] Assign a weight to the task capability factor of each task. The closer the task is to the time, the greater the weight of the task capability factor. The sum of the weights is 1.

[0192] Perform weighted summation on the task capability factors to obtain the update index, then calculate the average of the update index and the trust index to update the trust index.

[0193] It should be noted that for ease of understanding, simulated data is used as follows:

[0194] The closer the task capability factors are in time, the greater their influence will be. Therefore, the task capability factors are weighted by exponential decay. The specific formula is:

[0195]

[0196] where w i represents the weight of the i-th task capability factor, 0.7 is the decay factor, n−i represents the “distance” from task capability factor (n) to the current task capability factor (i), n represents the total number of task capability factors, and the denominator is a normalization factor to ensure that the sum of all weights is 1.

[0197] Assume that drone 001's original trust index T0 = 0.8. It has completed five missions, with mission capability factors Tcf1 = 0.7, Tcf2 = 0.85, Tcf3 = 0.9, Tcf4 = 0.75, and Tcf5 = 0.95. w1 ≈ 0.0866, w2 ≈ 0.1237, w3 ≈ 0.1767, w4 ≈ 0.2524, and w5 ≈ 0.3606.

[0198] The updated index is 0.0866×0.7+0.1237×0.85+0.1767×0.9+0.2524×0.75+0.3606×0.95≈0.8567.

[0199] The trust index is then calculated as 0.5×(0.8567+0.8)≈0.828.

[0200] Compared with the first embodiment, a greater weight is assigned to the capability factors of tasks that are close in time, which reflects the impact of time series on the trust index and makes the calculation more accurate.

[0201] Example 4:

[0202] In the first embodiment, when assigning a trust index to a drone, the initial trust index is 0.5, which is corrected according to the workload index. However, when the newly added drone is of the same model as the drones in the cluster, the initial trust index can be estimated based on the historical performance of the drone of that model to improve the stabilization speed of the trust index. Based on this, this example provides another method for assigning a trust index to each drone.

[0203] The trust indicators assigned to each drone include:

[0204] When a newly added drone has the same model as the drones in the cluster, all drones of the same model are first selected and the trust index and number of completed tasks of each drone are recorded;

[0205] The ratio of the number of tasks completed by the target drone to the number of tasks completed by all drones is used as the trust index weight of the target drone. The trust index of each drone is weighted and summed to obtain the initial trust index of the newly added drone.

[0206] Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate;

[0207] Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1, select the smaller value as the trust index of the newly added drone

[0208] It should be noted that for ease of understanding, simulated data is used as follows:

[0209] Assume that a new drone joins the swarm, and its model is "SkyLink X9-Pro". In the existing drone swarm, there are no three drones of the same model.

[0210] The current trust indices of these three drones are 0.625, 0.598, and 0.611 respectively; the number of completed missions are 5, 6, and 8 respectively;

[0211] In addition, the total number of existing drones is 24, 22 of which are currently on missions, and the number of idle drones is 24-22=2. The workload index is 2÷24≈0.083;

[0212] Calculate the initial trust index: 5×0.625÷(5+6+8)+6×0.598÷(5+6+8)+8×0.611÷(5+6+8)≈0.6106;

[0213] The difference between the initial trust index and the workload index is 0.528, which is greater than 0 and less than 1. Therefore, the trust index of the newly added drone is 0.528.

[0214] When drones of the same model join the cluster, introducing historical data of drones of the same model and estimating the initial trust index can improve the accuracy of the initial trust index, thereby increasing the stabilization speed of the trust index and making full use of the working capabilities of the newly added drones.

[0215] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is limited by the accompanying embodiments and their equivalents.

Claims

1. A collaborative processing system for distributed heterogeneous drone clusters, characterized by: include: Information collection module: Based on the data records of each drone, a capability vector identification table is established, including the drone's identification code, location, flight time and hardware data; Receive task information, including task content and task level; Data processing module: Preprocess different types of data in the capability vector identification table and assign a trust index to each drone; According to the UAV capability vector identification table and trust index, the UAV is assigned tasks according to the maximum value allocation method; During the UAV's mission execution, mission data is recorded, including mission level, motion trajectory, number of failures, and completion time. The mission capability factor is calculated using the completion calculation method, and the trust index is then updated based on the mission capability factor. Information output module: Creates a drone trust index data table and calculates drone anomalies based on fluctuations in the data table. When the anomaly exceeds a preset threshold, the drone is grounded and technicians are notified for repair. The completion degree calculation method includes: Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time; The calculation formula for task completion is: ； The calculation formula for task failure is: ; Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the estimated moving distance; The updating of the trust indicator according to the task capability factor specifically includes: Record the mission capability factor of the UAV after performing several missions; Set a fixed window size, which determines how many recent tasks are considered when calculating the updated confidence metric; When the number of completed tasks is less than the window size, the task capability factors of all completed tasks and the average value of the current trust index are calculated, and the trust index is updated; When the number of completed tasks reaches or exceeds the window size, each time a new task is completed, the earliest task is removed from the currently considered window and the new task is added to the window. Then, the average of the capability factors of all tasks in the window and the current trust index is calculated to update the trust index. The maximum value allocation method specifically includes: Drones are graded based on their trust indicators, with the number of grades matching the number of mission levels. Assign tasks from high to low according to their levels; When allocating tasks at the target level, the remaining drones from the previous level compete for allocation with the drones at the target level, and the remaining drones are delegated to the next level; When assigning tasks at a target level, if the number of drones is insufficient, drones with the highest trust index can be drawn from the next level to supplement the number; Then assign the next level of tasks until all tasks or drones are assigned.

2. The collaborative processing system for distributed heterogeneous UAV clusters according to claim 1, characterized in that: The completion degree calculation method includes: Record the data of the drone when performing the mission, including mission level, movement distance, number of failures and completion time; The calculation formula for task completion is: ， ； The calculation formula for task failure is: ; Where Tcf represents the task capability factor, f represents the number of failures, Δt represents the difference between the actual completion time and the expected completion time, g represents the task level, Δd represents the difference between the actual moving distance and the expected moving distance, and T e Indicates the estimated completion time, D e Indicates the expected moving distance, f max represents the maximum number of failures observed in history, Δt max Denotes the absolute value of the maximum time difference observed in history, Δd max It represents the absolute value of the maximum distance difference observed in history, and corr represents the correction factor.

3. The collaborative processing system for distributed heterogeneous UAV clusters according to claim 1, characterized in that: The updating of the trust indicator according to the task capability factor specifically includes: Record the mission capability factor of the UAV after performing several missions; Assign a weight to the task capability factor of each task. The closer the task is to the time, the greater the weight of the task capability factor. The sum of the weights is 1. Perform weighted summation on the task capability factors to obtain the update index, then calculate the average of the update index and the trust index to update the trust index.

4. The collaborative processing system for a distributed heterogeneous UAV cluster according to claim 1, characterized in that: The trust indicators assigned to each drone specifically include: When a new drone joins the cluster, its initial trust index is set to 0.5; Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate; Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1 and select the smaller value as the trust index of the newly added drone.

5. The collaborative processing system for distributed heterogeneous UAV clusters according to claim 1, characterized in that: The trust indicators assigned to each drone specifically include: When a newly added drone has the same model as the drones in the cluster, all drones of the same model are first selected and the trust index and number of completed tasks of each drone are recorded; The ratio of the number of tasks completed by the target drone to the number of tasks completed by all drones is used as the trust index weight of the target drone. The trust index of each drone is weighted and summed to obtain the initial trust index of the newly added drone. Get the ratio of the current number of idle drones to the current number of all drones, obtain the workload index, and then calculate; Calculate the difference between the initial trust index and the workload index, compare it with 0, select the larger value, and then compare it with 1 and select the smaller value as the trust index of the newly added drone.

6. The collaborative processing system for distributed heterogeneous UAV clusters according to claim 1, characterized in that: The calculation of drone abnormal values ​​based on data table fluctuations specifically includes: Record the updates of the drone's trust indicators for several times and calculate the average rate of change; Calculate the absolute value of the change rate of the most recent trust indicator, and take the ratio of the absolute value of the change rate to the average change rate as the outlier; When the outlier value is less than 0.5 or greater than 1.5, the drone will be grounded and the technician will be notified for repair.

7. A collaborative processing method for distributed heterogeneous UAV clusters, characterized in that: include: Information collection: Based on the data records of each drone, a capability vector identification table is established, including the drone's identification code, location, flight time, and hardware data; Receive task information, including task content and task level; Assigning trust indicators: Preprocess different types of data in the capability vector identification table separately and assign a trust indicator to each drone; Task allocation: Based on the UAV capability vector identification table and trust index, the UAV is assigned tasks according to the maximum value allocation method; Update trust indicators: During the UAV's mission execution, mission data is recorded, including mission level, motion trajectory, number of failures, and completion time. The mission capability factor is calculated using the completion calculation method, and the trust indicator is then updated based on the mission capability factor. Trust indicator abnormality feedback: Establish a drone trust indicator data table and calculate the drone abnormal value based on the fluctuation of the data table. When the abnormal value is greater than the preset threshold, the drone will be grounded and technical personnel will be notified for maintenance.

Images