Complex confrontation environment communication adaptive cluster intelligent streaming task allocation method

Abstract

The invention discloses a complex confrontation environment communication adaptive unmanned cluster intelligent streaming task allocation method. The method comprises the following steps: selecting an object unmanned aerial vehicle B of point-to-point communication; according to the environment information obtained by the current unmanned aerial vehicle A, whether a new to-be-allocated task is generated or not is judged, and a to-be-allocated task list added to the current unmanned aerial vehicle A is judged; the object unmanned aerial vehicle B generates a streaming task message body; according to the sent streaming task message body, updating a to-be-allocated task list, a processing task list and a completion task list of the current unmanned aerial vehicle A; the current unmanned aerial vehicle A offers a quotation for the highest profit, and a temporary distribution result is formed; judging whether a set iteration round is reached or not, and if yes, moving the temporary allocation result from the task list in processing to a completed task list of the current unmanned aerial vehicle A; and if not, returning to continue iteration. According to the method, the adaptability of heterogeneous unmanned cluster dynamic task distributed decision-making to communication conditions among individual unmanned aerial vehicles in the cluster is improved.

Description

Technical Field

[0001] The invention relates to a method for allocating intelligent streaming tasks of adaptive clusters in complex adversarial environment communications, and belongs to the field of group collaboration of artificial intelligence technology. Background Art

[0002] The target allocation problem is widely used when unmanned swarms process multi-task targets. Distributed decision-making capabilities are considered to be one of the manifestations of the intelligence of unmanned swarms and are valued by all parties. However, the decision-making algorithm results of distributed unmanned swarms are easily affected by communication interference. Reducing or even eliminating the impact of communication on distributed decision-making results is of great value.

[0003] Compared with other distributed task methods, the mathematical derivation of the market-like mechanism method is rigorous, the computational complexity can be proven, the theoretical explanation is clear, and there are few restrictions on the network scale and topology structure. In practical applications, cluster individuals only need to calculate the auction or bidding information, which has a small amount of calculation and is suitable for the limited computing power of distributed unmanned cluster platforms. Especially in the future military field, the scale of unmanned clusters is large, the network topology is complex, and the battlefield environment changes instantly. The above characteristics of the market-like mechanism method can better solve the unmanned cluster task planning problem in this scenario. However, the current method requires all drones to process a certain number of tasks at the same time, which cannot meet the streaming processing of tasks. However, in the real environment, tasks are often not generated at the same time, but are initiated by each drone in the cluster. Therefore, it is of great application value to transform the distributed algorithm into a form that supports streaming data input. Summary of the invention

[0004] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art, to propose a complex adversarial environment communication adaptive cluster intelligent streaming task allocation method, and to reduce the impact of adversarial environment communication obstruction on the consistency of cluster task allocation results.

[0005] The technical solution of the present invention is: a method for intelligent streaming task allocation of adaptive cluster communication in complex confrontation environment, comprising:

[0006] 1) According to the communication history between the current drone A and other drones in the entire unmanned cluster, the drone B is selected as the target for this point-to-point communication;

[0007] 2) According to the environmental information obtained by the current UAV A, determine whether to generate a new task to be assigned, and determine whether to add it to the list of tasks to be assigned of the current UAV A;

[0008] 3) The target drone B generates a streaming task message body based on its stored list of pending tasks, list of tasks being processed, and list of completed tasks, and sends it to the current drone A;

[0009] 4) Update the current drone A's own pending task list, in-process task list, and completed task list according to the streaming task message body sent by the target drone B;

[0010] 5) According to the updated list of tasks being processed by the current drone A, the current drone A gives a bid to the one with the highest profit, forms a temporary allocation result, and stores it in the list of tasks being processed;

[0011] 6) Determine whether the tasks in the task list being processed by the current drone A have reached the set iteration round. If so, the temporary allocation result is moved from the task list being processed to the completed task list of the current drone A; if not, return to step 1) to continue iteration.

[0012] In the step 1), based on the communication history between the current drone A and other drones in the entire unmanned cluster, the target drone B for this point-to-point communication is selected, including: using the communication frequency negative power function method to give the probability that other drones in the entire unmanned cluster except the current drone A are selected as the communication objects of the current drone A, and extracting the target drone B based on this probability distribution.

[0013] In the step 3), the streaming task message body generated by the object drone B includes the current list of tasks to be assigned, the list of tasks being processed, and the list of completed tasks. The list of tasks to be assigned stores the task label and task start time of each task; the list of tasks being processed stores the task label, task start time, current quotation, and quotation drone ID of each task.

[0014] The current quotation means that each task in the processing task list will be temporarily given a price for the current execution of the task by a certain drone.

[0015] The completed task list includes the task tag, task start time, current quotation and quotation drone ID; the task tag includes information for calculating the profit of the current drone A performing this task.

[0016] In step 4), the task list to be assigned, the task list in progress and the task list completed of the current drone A are updated, including:

[0017] Update the list of tasks to be assigned: remove duplicate tasks from the list of tasks to be assigned, set the quotation drone of each task in the list of tasks to be assigned to the current drone A and set the current quotation to 0, and then move the task to the list of tasks being processed;

[0018] Update the list of tasks in progress: remove duplicate tasks from the list of tasks in progress, and merge the quotations in the list of tasks in progress of the current drone A with the quotations in the list of tasks in progress of the target drone B;

[0019] Update the completed task list: remove duplicate tasks from the completed task list, and merge the quotations and quotation drone IDs of the completed task list.

[0020] When completing the elimination of duplicate tasks in the list of tasks to be assigned, completing the elimination of duplicate tasks in the list of tasks being processed, and completing the elimination of duplicate tasks in the list of completed tasks, the judgment criteria for the duplicate tasks are: based on the description in the task label, determine whether the tasks from the current drone A and the target drone B are the same task.

[0021] The merging of the quotations in the processing task list of the current drone A and the quotations in the processing task list of the target drone B includes: for each task in the processing task list of the target drone B, if it also exists in the current drone A, the current quotation in the processing task list of the current drone A is updated to the higher quotation and the quoting drone ID in the processing task list of the current drone A is updated; if it is not in the processing task list of the current drone A, the task is added to the processing task list of the current drone A, and the quotation and the quoting drone are the same as the data in the target drone B.

[0022] The merging of the quotations and quoting drone IDs of the completed task list includes: for each task in the completed task list of the object drone B, if it also exists in the current drone A, the current quotation of the completed task list of the current drone A is updated to the higher quotation and the quoting drone ID in the completed task list of the current drone A is updated; if it is not in the completed task list of the current drone A, the task is added to the completed task list of the current drone A, and the quotation and quoting drone are the same as the data in the object drone B.

[0023] In the step 5), according to the updated list of tasks being processed by the current drone A, the current drone A makes a bid for the one with the highest profit to form a temporary allocation result, including: the current drone A calculates the income of each task through the task tags in the list of tasks being processed, solves the net income of each current task according to the difference between the income of each task and the current bid, selects the task with the largest net income of the current drone A, bids for it, and modifies the current bid and the bidding drone ID of this task.

[0024] The beneficial effects of the present invention are:

[0025] 1. Existing distributed task allocation methods usually only consider allocating a certain number of tasks to a certain number of drones, while the present invention adopts a streaming task message body, which accommodates more task information and can support distributed task generation and streaming input.

[0026] 2. The existing distributed task allocation method adopts the fixed communication assumption, while the present invention proposes a communication object selection method based on actual application conditions. Probabilistically extracting objects with fewer communications can greatly increase communication efficiency, while avoiding many application problems caused by the original method of communicating in a fixed order.

[0027] 3. The premise of the existing distributed task allocation method to prove convergence from the theoretical and experimental level is the connectivity of the unmanned cluster communication network, which requires the stability of the communication network structure, but the communication network cannot remain stable in real applications. The probabilistic communication method used in the present invention relaxes the strict requirements for the communication network structure and can better adapt to real application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0029] like Figure 1 As shown, the present invention proposes an intelligent streaming task allocation method for an adaptive unmanned cluster in a complex confrontation environment communication, the steps comprising: (distributed deployment and operation on each drone in the unmanned cluster)

[0030] The first step is to use probability P j Extract the drone j that can communicate with the current drone A as the object of this communication:

[0031]

[0032] Among them, n is the total number of drones that have communicated with the current drone, μj is the number of communications between the current drone and drone j, μk is the number of communications between the current drone and drone k, and the extracted drone is called the object drone B.

[0033] The second step is to initialize the allocation results saved in all drones, and determine whether a new task to be allocated is generated based on the actual task constraints. If so, the newly generated task to be allocated is assigned to the drone that proposed the task, and the drone bids 0 for it.

[0034] In the third step, the selected object drone B generates a streaming task message body with its stored list of tasks to be assigned, list of tasks being processed, and list of completed tasks, and sends it to the current drone A.

[0035] Step 4: Perform a quotation merge operation for the current drone A and the target drone B. That is, for tasks that exist in both quotations, the merged quotation is the larger of the two original quotations. For tasks that only exist in one of them, it is added to the other's quotation and the quotation remains unchanged.

[0036] Step 5: Based on the combined bids, the current drone i makes a bid;

[0037] For the task with the highest profit in the task list currently being processed by drone A, increase the price by Δp:

[0038] Δp=p max -p smax +ε

[0039] Among them, p max is the profit of the task with the largest net profit in the task list being processed, p max The net profit of the task with the second largest net profit in the task list being processed is ε, which is a hyperparameter and needs to be adjusted according to the actual value. The larger the ε, the faster the convergence, but consistency problems may occur. If the tasks currently undertaken by the drone exceed the upper limit, the tasks will be assigned to other drones without giving a quotation.

[0040] In the sixth step, jump to the first step until the method iteration limit is reached.

[0041] The present invention provides a complex adversarial environment communication adaptive cluster intelligent streaming task allocation method, which has two processes that need to be described in detail. One is the improvement of the communication network. Since the communication network structure is unstable and the information of the tasks to be allocated that can be obtained in each local area network cannot be guaranteed to be consistent, the priority weight of communication is designed in each drone, and the communication object and frequency ratio of each drone are recorded. The weight of the communication with a lower frequency ratio is increased, which helps to obtain more overall data and make better judgments.

[0042] Suppose that within a certain communication time interval, i The n nodes in the connection communicated μ j times, then the next time a communication object is selected, the probability of each communication node being selected is:

[0043]

[0044] In addition, it is necessary to elaborate on the conflicts that arise when the two original communication networks are merged. If we observe the operation process of the auction method from an overall perspective, it is not difficult to find that at any time, it can be determined that any two connected nodes p i ∈N(i),p j ∈N(j) can represent the only restored communication node between the two communication networks N(i) and N(j) that were once broken into two parts, and R(i) and R(j) represent the distribution scheme that the two original networks reach a consensus on. Then, in p i ,p jIf the unified new allocation result R(i)′=R(j)′ after communication can be generated using the two original allocation schemes, the distributed operation of the method can be completed. The method hopes to construct a fluid distributed task allocation method, where new tasks are continuously generated and added to the list to be allocated, and the allocation results are continuously output to the execution unit for execution, rather than the naive distributed allocation method that distributes all known weight information, and then ensures global consistency through a consistency method, and obtains the processing results of the time section after a specific deadline.

[0045] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A method for intelligent streaming task allocation of adaptive cluster communication in complex adversarial environment, characterized in that include: 1) According to the communication history between the current drone A and other drones in the entire unmanned cluster, the drone B is selected as the target for this point-to-point communication; 2) According to the environmental information obtained by the current UAV A, determine whether to generate a new task to be assigned, and determine whether to add it to the list of tasks to be assigned of the current UAV A; 3) The target drone B generates a streaming task message body based on its stored list of pending tasks, list of tasks being processed, and list of completed tasks, and sends it to the current drone A; 4) Update the current drone A's own pending task list, in-process task list, and completed task list according to the streaming task message body sent by the target drone B; 5) According to the updated list of tasks being processed by the current drone A, the current drone A gives a bid to the one with the highest profit, forms a temporary allocation result, and stores it in the list of tasks being processed; 6) Determine whether the tasks in the task list being processed by the current drone A have reached the set iteration round. If so, the temporary allocation result is moved from the task list being processed to the completed task list of the current drone A; if not, return to step 1) to continue iteration.

2. According to claim 1, a complex confrontation environment communication adaptive cluster intelligent streaming task allocation method is characterized in that: In the step 1), based on the communication history between the current drone A and other drones in the entire unmanned cluster, the target drone B for this point-to-point communication is selected, including: using the communication frequency negative power function method to give the probability that other drones in the entire unmanned cluster except the current drone A are selected as the communication objects of the current drone A, and extracting the target drone B based on this probability distribution.

3. According to claim 1, a complex confrontation environment communication adaptive cluster intelligent streaming task allocation method is characterized in that: In the step 3), the streaming task message body generated by the object drone B includes the current list of tasks to be assigned, the list of tasks being processed, and the list of completed tasks. The list of tasks to be assigned stores the task label and task start time of each task; the list of tasks being processed stores the task label, task start time, current quotation, and quotation drone ID of each task.

4. According to claim 3, a complex confrontation environment communication adaptive cluster intelligent streaming task allocation method is characterized in that: The current quotation means that each task in the processing task list will be temporarily given a price for the current execution of the task by a certain drone.

5. According to claim 3, a complex confrontation environment communication adaptive cluster intelligent streaming task allocation method is characterized in that: The completed task list includes the task tag, task start time, current quotation and quotation drone ID; the task tag includes information for calculating the profit of the current drone A performing this task.

6. The method for intelligent streaming task allocation of adaptive clustering in complex confrontation environment communication according to claim 5 is characterized in that: In step 4), the task list to be assigned, the task list in progress and the task list completed of the current drone A are updated, including: Update the list of tasks to be assigned: remove duplicate tasks from the list of tasks to be assigned, set the quotation drone of each task in the list of tasks to be assigned to the current drone A and set the current quotation to 0, and then move the task to the list of tasks being processed; Update the list of tasks in progress: remove duplicate tasks from the list of tasks in progress, and merge the quotations in the list of tasks in progress of the current drone A with the quotations in the list of tasks in progress of the target drone B; Update the completed task list: remove duplicate tasks from the completed task list, and merge the quotations and quotation drone IDs of the completed task list.

7. The method for intelligent streaming task allocation of adaptive cluster in complex confrontation environment communication according to claim 6 is characterized in that: When completing the elimination of duplicate tasks in the list of tasks to be assigned, completing the elimination of duplicate tasks in the list of tasks being processed, and completing the elimination of duplicate tasks in the list of completed tasks, the judgment criteria for the duplicate tasks are: based on the description in the task label, determine whether the tasks from the current drone A and the target drone B are the same task.

8. The method for allocating intelligent streaming tasks in a complex adversarial environment communication adaptive cluster according to claim 6, characterized in that: The merging of the quotations in the processing task list of the current drone A and the quotations in the processing task list of the target drone B includes: for each task in the processing task list of the target drone B, if it also exists in the current drone A, the current quotation in the processing task list of the current drone A is updated to the higher quotation and the quoting drone ID in the processing task list of the current drone A is updated; if it is not in the processing task list of the current drone A, the task is added to the processing task list of the current drone A, and the quotation and the quoting drone are the same as the data in the target drone B.

9. The method for intelligent streaming task allocation of adaptive cluster in complex confrontation environment communication according to claim 4 is characterized in that: The merging of the quotations and quoting drone IDs of the completed task list includes: for each task in the completed task list of the object drone B, if it also exists in the current drone A, the current quotation of the completed task list of the current drone A is updated to the higher quotation and the quoting drone ID in the completed task list of the current drone A is updated; if it is not in the completed task list of the current drone A, the task is added to the completed task list of the current drone A, and the quotation and quoting drone are the same as the data in the object drone B.

10. The method for intelligent streaming task allocation of adaptive cluster in complex confrontation environment communication according to claim 1, characterized in that: In the step 5), according to the updated list of tasks being processed by the current drone A, the current drone A makes a bid for the one with the highest profit to form a temporary allocation result, including: the current drone A calculates the income of each task through the task tags in the list of tasks being processed, solves the net income of each current task according to the difference between the income of each task and the current bid, selects the task with the largest net income of the current drone A, bids for it, and modifies the current bid and the bidding drone ID of this task.

Images