Task allocation optimization method based on multi-machine multi-task allocation problem model MDTAP

Abstract

The invention discloses a task allocation optimization method based on a multi-UAV multi-task allocation problem model (MDTAP), and the method comprises the steps: introducing a constraint condition, carrying out the N-ary conversion solving of an uncertain matrix in a multi-UAV multi-task allocation benefit model through employing a coding mode, and constructing a fitness function according to the total income, total cost and total risk; dividing an input value interval, and determining a separation point matrix and an interval midpoint matrix; and calculating and selecting a search iteration probability or an optimization iteration probability, selecting search iteration or optimization iteration through a roulette method, and jointly searching an optimal solution of the fitness function by using two iteration methods of search iteration and optimization iteration. According to the method, the constraint condition of the task is fused in the function, the universality, diversity and convenience of task allocation are improved, the fitness function combines the task cost with the unmanned aerial vehicle cost and the load cost, and the risk and unmanned aerial vehicle exposure length and the unmanned aerial vehicle and load cost are combined, so that the allocation is more reasonable and practical.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle task allocation, and in particular to a task allocation optimization method based on a multi-machine multi-task allocation problem model MDTAP. Background Art

[0002] In recent years, UAV technology has developed rapidly and has been highly valued for its flexibility and efficiency. With the increasing complexity of application scenarios and mission objectives, higher requirements have been put forward for the research on using multiple types of UAVs to collaboratively complete multiple tasks.

[0003] The traveling salesman problem model TSP is a classic task allocation model, but it can only describe the task allocation problem of one drone. The multi-traveling salesman problem model mTSP is the development of the classic traveling salesman problem model. Although it can describe the multi-task allocation problem of multiple drones, it has low adaptability in describing different types of drones and different types of tasks as well as the constraints of the tasks. The collaborative multi-task allocation model CMTAP can describe this type of problem more appropriately. However, it lacks diversity in calculating the benefits of various tasks, and requires separate judgments and trade-offs when considering constraints, making the problem-solving process incoherent and the model less versatile. When calculating the cost of the allocation plan, the existing technology only considers the cost of the drone itself, but cannot calculate the cost of the payload, making the practicality and rationality of the allocation model low.

[0004] For the solution of multi-task problems completed by multiple UAVs in collaboration, linear programming, integer programming, heuristic algorithms, multi-objective optimization, etc. are mainly used. Among the heuristic algorithms, genetic algorithms, simulated annealing algorithms, ant colony algorithms and particle swarm algorithms are widely used. Although the existing technology can eliminate solutions with poor effects and retain solutions with good effects, the eliminated solutions cannot provide guidance for the subsequent optimization process. The existing technology only retains the excellent solutions obtained by iteration, lacks analysis of bad solutions, and will cause waste of computing power and information. In the selection of methods at each stage of the solution, the existing technology cannot switch methods or only switches with the number of iterations, and cannot integrate the iteration results and the number of iterations and select the iteration method at the same time. This causes the existing technology to be unable to quickly traverse the entire feasible domain in the early stage of the solution, it is difficult to quickly jump out of the local optimal solution in the middle stage of the solution, and it is difficult to further optimize the known solution at the end of the solution. Summary of the invention

[0005] Purpose of the invention: The present invention aims to provide a task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP, which integrates the constraints of the task into the function.

[0006] Technical solution: The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP of the present invention comprises the following steps: (1) Construct a multi-UAV multi-task allocation benefit model, including payload model, task model, UAV model, and allocation relationship model; (2) Introducing constraints, the uncertainty matrix in the multi-UAV multi-task allocation benefit model is converted into N-ary form using coding method to solve it; (3) Based on total revenue Total cost and total risk Constructing the fitness function ; (4) According to the parameters of the multi-UAV multi-task allocation benefit model, initialize the task allocation optimization, divide the input value interval, and determine the separation point matrix and the interval midpoint matrix; (5) Calculate the probability of selecting a search iteration or an optimization iteration, and select the search iteration or the optimization iteration by the roulette wheel method; if the search iteration is selected, proceed to step (6); otherwise, proceed to step (7); (6) Calculate the iteration probability of each search iteration interval, select the search iteration interval by roulette method, and calculate the iteration probability according to the fitness function , calculate the fitness value , update the interval search iteration parameters, if the fitness value Not less than the global maximum fitness value , then update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters in sequence, and go to step (8); otherwise, directly update the interval and global parameters and go to step (8); (7) Calculate the input value of the optimization iteration according to the fitness function , calculate the fitness value , determine the interval where the input value of the optimization iteration is located, update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters, and enter step (8); (8) If the current global calculation times are less than the preset calculation times, return to step (5); otherwise, solve the task allocation scheme through the allocation relationship model based on the current global maximum input value and the global maximum fitness value.

[0007] Furthermore, in step (2), the uncertainty matrix is ​​the decision matrix of the distribution relationship model , Allocation Quantity Matrix , unordered allocation matrix , Task Allocation Matrix , Task Time Matrix ; Task execution time matrix of task model , Task expected benefit matrix , Complete the task benefit matrix ; UAV flight time matrix for UAV model , UAV mileage matrix , Drone exposure duration matrix .

[0008] Decision Matrix of the Allocation Relationship Model No. i Line j Elements of a column for ;in, Indicates drone With the task The distribution relationship, Representation Task Assign to drone ; Representation Task Not assigned to drones ; express The j-th element of It is a base-N value, obtained by rounding down the decimal input value x. ,Will Convert to get; Allocation Quantity Matrix middle for ;in, Indicates drone The number of tasks assigned; Unordered Allocation Matrix middle for ;in, Indicates drone Tasks assigned out of order; Indicates that no drone was given Assigned to tasks; Indicates that the task Assign to drone ,and , The decision matrix The column number of the jth element with a value of 1 in the i-th row; task allocation matrix By unordered distribution matrix Sure; Task Time Matrix middle for ;in, Indicates drone Execute the task time, Indicates drone Starting point and mission The distance between ; Indicates drone speed, Representation Task With the task The distance between , .

[0009] Task execution time matrix of task model middle for ;in, Representation Task Execution time; Task Estimated Profit Matrix middle for ; ; In the formula, Constraint parameters for the importance of the task; and is the urgency constraint parameter of the task, where It is manifested as the linear change of task value with execution time. It is manifested as the exponential change of task value with execution time; and is the periodic constraint parameter of the task, where It is manifested as the cycle of changes in task value over execution time; and is the execution time window constraint parameter of the task, where The value of reflects the reward for executing the task within the execution time window. Reflects the penalty for executing tasks outside the execution time window; The time interval required to perform the task; Completion of task benefit matrix middle for ;in, Indicates completion of task of income; For drones Is there a load? , , Representation Task Type, Indicates drone Loading load ; Indicates drone No load ; Indicates that the task is not taken Required payload to perform the task The penalty value.

[0010] Furthermore, the UAV flight time matrix of the UAV model middle for ;in, Indicates drone The total time of the voyage to complete all tasks and return as required, Indicates drone The time when the last task was executed; Indicates drone Starting point and mission The distance between , Indicates drone Assigned tasks in order , that is, the last drone assigned to the task, , ; Indicates drone speed; Indicates drone Round trip requirements, Indicates drone After completing all tasks, you need to return to the starting point. Indicates drone There is no need to return to the starting point after completing all tasks; Drone Mileage Matrix middle for ;in, Indicates drone The total mileage of the voyage in which all missions are completed and returned as required; Drone exposure duration matrix middle for ;in, Indicates drone The total exposure time to complete all tasks, Representation Task The exposure radius of .

[0011] Furthermore, in step (3), the fitness function for ; ; ; ; ; ; in, Indicates drone Costs; Indicates drone Is there a load? ; Indicates load Costs; Indicates the cost of drones; Indicates the load cost; is the UAV mileage cost coefficient; is the UAV flight time cost coefficient; is the load cost factor.

[0012] Furthermore, in step (4), the separation point matrix for ; ; ; in, Represents the input value of the i-th separation point; express The fitness value of is the maximum value of the domain of the input value x, and N is a constant greater than 1; Interval midpoint matrix for ; ; ; in, Representation interval The input value of the midpoint of express The fitness value of .

[0013] Furthermore, in step (5), the probability of optimizing iteration is selected and the probability of selecting a search iteration as follows: ; ;in, To preset the number of calculations, is the number of global calculations, Represents the probability calculation parameters for optimization iterations.

[0014] Furthermore, in step (6), i The iteration probability of the search iteration interval for ; ; ;in, Representation interval The maximum input value among ; express The fitness value of Representation interval The number of calculations.

[0015] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. The present invention converts the uncertainty matrix in the multi-UAV multi-task allocation benefit model into N-ary form according to the constraints by coding, solves the benefit function of each task according to the task completion time and the task constraints, integrates the task constraints into the function, and increases the versatility, diversity and convenience of the task allocation model; 2. The present invention constructs a fitness function according to the total benefit, total cost and total risk, combines the task cost with the UAV cost and the payload cost, making the allocation more reasonable, and combines the risk with the UAV exposure length and the cost of the UAV and the payload, making the MDTAP model more practical; 3. The present invention uses two iterative methods, search iteration and optimization iteration, to jointly find the optimal solution of the fitness function. The search iteration divides the input value into multiple intervals, and can quickly and as much as possible traverse each interval segment. The search iteration reflects the best solution of each interval to the iteration probability of each interval, so that more computing power is invested in the interval with excellent iteration value, and the iteration interval can be better selected; the optimization iteration uses the known optimal solution and the corresponding sorting method to predict the global optimal solution, and can optimize the possible global optimal solution near the known optimal solution, and will continue to converge to the known optimal solution; 4. The present invention calculates the search iteration probability and the optimization iteration probability, so as to select the appropriate iteration method, and reflects the result of each iteration on the search iteration probability and the optimization iteration probability in the next iteration, so that the result of each iteration can guide the next iteration, so that the search iteration is used as much as possible in the early stage of the calculation to quickly search for a better solution, and the optimization iteration is used as much as possible in the late stage of the calculation to continuously optimize the nearby better solution; when the search iteration is used to search for a better solution, the known optimal solution of the optimization iteration will be updated, and the optimization area will be quickly moved to the area where the new optimal solution is located. When the result of the optimization iteration is poor, the probability of selecting the search iteration can be increased, so as to continue to dig for a better solution in other areas. The overall solution process can be quickly and effectively switched between the two methods, making the solution faster and reducing the loss of computing power. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of the present invention; Figure 2 This is a graph of the results of the embodiment. DETAILED DESCRIPTION

[0017] The present invention will be further described below in conjunction with the accompanying drawings.

[0018] The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP of the present invention comprises the following steps: (1) Construct a multi-UAV multi-task allocation benefit model, including payload model, task model, UAV model and allocation relationship model.

[0019] The model is constructed as N drones with K types of loads cooperating to perform M tasks. The load model is mainly composed of the load cost matrix The task model mainly includes: task type matrix ; Task exposure radius matrix ; Distance matrix between tasks Etc. The UAV model mainly includes: UAV speed matrix ; Drone cost matrix ; Drone round trip requirements matrix ; UAV payload type matrix ; The distance between the drone’s starting point and the mission The distribution relationship model mainly includes: the decision matrix of the distribution relationship model , Task Allocation Matrix wait.

[0020] (2) Constraints are introduced and a coding method is used to convert the uncertainty matrix in the multi-UAV multi-task allocation benefit model into N-ary system to solve it.

[0021] The uncertainty matrix is ​​the decision matrix of the distribution relationship model , Allocation Quantity Matrix , unordered allocation matrix , Task Allocation Matrix , Task Time Matrix ; Task execution time matrix of task model , Task expected benefit matrix , Complete the task benefit matrix ; UAV flight time matrix for UAV model , UAV mileage matrix , Drone exposure duration matrix .

[0022] Decision Matrix of the Allocation Relationship Model No. i Line j Elements of a column for ;in, Indicates drone With the task The distribution relationship, Representation Task Assign to drone ; Representation Task Not assigned to drones ; express The j-th element of It is a base-N value, obtained by rounding down the decimal input value x. ,Will Convert to get; Allocation Quantity Matrix middle for ;in, Indicates drone The number of tasks assigned; Unordered Allocation Matrix middle for ;in, Indicates drone Tasks assigned out of order; Indicates that no drone was given Assigned to tasks; Indicates that the task Assign to drone ,and , The decision matrix The column number of the j-th element with a value of 1 in the i-th row; Task Allocation Matrix By unordered distribution matrix Determine; Task Allocation Matrix Elements Indicates drone Assign tasks in order. Indicates that no drone was given Assign tasks; Indicates that the task Assign to drones in order l The unordered assignment matrix uses the integer part of x to assign all M tasks to N drones, where The number of tasks assigned is , then drone Total The sorting scheme executes the assigned missions, all drones have The M tasks are executed in order using the arrangement scheme. The permutation schemes are mapped to unit 1, and the permutation scheme is determined according to the decimal part of x, thus obtaining the task allocation matrix The mapping scheme is as follows: calculate the minimum interval segment where x is currently located, and divide the minimum interval segment into segment, determine the position of the bisecting segment where x is located, and The corresponding position elements in are extracted and assigned to , fill forward The empty element in the interval and update the minimum interval where x is currently located to the current bisection segment to determine the drone Have all the tasks been assigned? If yes, proceed to the next UAV assignment. Otherwise, proceed to the next task assignment; determine whether all UAVs have completed the task assignment. If yes, output ; Otherwise, allocate the next drone.

[0023] Task Time Matrix middle for ; in, Indicates drone Execute the task time, Indicates drone Starting point and mission The distance between ; Indicates drone speed, Representation Task With the task The distance between , .

[0024] Task execution time matrix of task model middle for ;in, Representation Task Execution time; Task Estimated Profit Matrix middle for ; ; In the formula, Constraint parameters for the importance of the task; and is the urgency constraint parameter of the task, where It is manifested as the linear change of task value with execution time. It is manifested as the exponential change of task value with execution time; and is the periodic constraint parameter of the task, where It is manifested as the cycle of changes in task value over execution time; and is the execution time window constraint parameter of the task, where The value of reflects the reward for executing the task within the execution time window. Reflects the penalty for executing tasks outside the execution time window; The time interval required to perform the task; Completion of task benefit matrix middle for ;in, Indicates completion of task of income; For drones Is there a load? , , Representation Task Type, Indicates drone Loading load ; Indicates drone No load ; Indicates that the task is not taken Required payload to perform the task The penalty value.

[0025] UAV flight time matrix for UAV model middle for ;in, Indicates drone The total time of the voyage to complete all tasks and return as required, Indicates drone The time when the last task was executed; Indicates drone Starting point and mission The distance between , Indicates drone Assigned tasks in order , that is, the last drone assigned to the task, , ; Indicates drone speed; Indicates drone The round trip requirement, Indicates drone After completing all tasks, you need to return to the starting point. Indicates drone There is no need to return to the starting point after completing all tasks; Drone Mileage Matrix middle for ;in, Indicates drone The total mileage of the voyage in which all missions are completed and returned as required; Drone exposure duration matrix middle for ; in, Indicates drone The total exposure time to complete all tasks, Representation Task The exposure radius of .

[0026] (3) Based on total revenue Total cost and total risk Constructing the fitness function .

[0027] Fitness function for ; ; ; ; ; ; in, Indicates drone Costs; Indicates drone Is there a load? ; Indicates load Costs; Indicates the cost of drones; Indicates the load cost; is the UAV mileage cost coefficient; is the UAV flight time cost coefficient; is the load cost factor.

[0028] (4) According to the parameters of the multi-UAV multi-task allocation benefit model, initialize the task allocation optimization, divide the input value interval, and determine the separation point matrix and the interval midpoint matrix.

[0029] Separator Point Matrix for ; ; ; in, Represents the input value of the i-th separation point; express The fitness value of is the maximum value of the domain of the input value x, and N is a constant greater than 1; Interval midpoint matrix for ; ; ; in, Representation interval The input value of the midpoint of express The fitness value of .

[0030] (5) Calculate the probability of selecting the search iteration or the optimization iteration, and select the search iteration or the optimization iteration by the roulette method; if the search iteration is selected, proceed to step (6); otherwise, proceed to step (7).

[0031] Probability of choosing optimization iterations and the probability of selecting a search iteration as follows: ; ;in, To preset the number of calculations, is the number of global calculations, Represents the probability calculation parameters for optimization iterations.

[0032] (6) Calculate the iteration probability of each search iteration interval, select the search iteration interval by roulette method, and calculate the iteration probability according to the fitness function , calculate the fitness value , update the interval search iteration parameters, if the fitness value Not less than the global maximum fitness value , then update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters in sequence, and go to step (8); otherwise, directly update the interval and global parameters and go to step (8).

[0033] No. i The iteration probability of the search iteration interval for ; ; ; in, Representation interval The maximum input value among ; express The fitness value of Representation interval The number of calculations.

[0034] The method to calculate the search iterations is ; The method to update the interval search iteration parameters is ; ; The method for updating the optimization iteration probability parameters is .

[0035] (7) Calculate the input value of the optimization iteration according to the fitness function , calculate the fitness value , determine the interval where the input value of the optimization iteration is located, update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters, and enter step (8).

[0036] The method for calculating the optimization iteration is ; ; ; ; ; ; ; .

[0037] The method for updating the optimization iteration reference matrix is ​​to first update the value through the following formula: ; ; ; Then arrange the input values ​​in ascending order to meet the conditions: .

[0038] (8) If the current global calculation times are less than the preset calculation times, return to step (5); otherwise, solve the task allocation scheme through the allocation relationship model based on the current global maximum input value and the global maximum fitness value.

[0039] In order to verify the effectiveness and rationality of the method proposed in the present invention, the following examples are given: In the Python environment, a simulation experiment is conducted to verify the task allocation optimization method for the multi-machine multi-task allocation problem model MDTAP proposed in the present invention. First, the MDTAP model parameters are set in Python as shown in Table 1, Table 2 and Table 3.

[0040] Table 1 ; Table 2 ; Table 3 ; In Python, set the preset number of calculations The inverse of the global maximum value is obtained by simulation experiment. With the number of iterations The result of the change is as follows Figure 2 As shown, the present invention can quickly and as much as possible traverse each interval segment in the early stage of the experiment, so that a better allocation plan can be quickly iterated in the early stage. And the plan will be optimized and iterated, so that the result is further reduced. The experiment fell into the local optimal solution from the 10,000th to 20,000th iterations, and it was impossible to further optimize a better solution. At this time, the present invention increases the probability of search iterations and explores the possibility of other better solutions in other areas. At about the 23,000th iteration, the search iteration experiment successfully jumped out of the local optimal solution and found a better solution. In the later stage of the experiment, the present invention increases the probability of optimization iterations, optimizes the known optimal solution, reduces the possibility of searching other areas, and finally further optimizes a better solution at about the 96,000th iteration. The effectiveness and rationality of the method proposed by the present invention can be verified through simulation experiments.

Claims

1. A task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP, characterized in that: The following steps are involved: (1) Construct a multi-UAV multi-task allocation benefit model, including payload model, task model, UAV model, and allocation relationship model; (2) Introducing constraints, the uncertainty matrix in the multi-UAV multi-task allocation benefit model is converted into N-ary form using coding method to solve it; (3) Based on total revenue Total cost and total risk Constructing the fitness function ; (4) According to the parameters of the multi-UAV multi-task allocation benefit model, initialize the task allocation optimization, divide the input value interval, and determine the separation point matrix and the interval midpoint matrix; (5) Calculate the probability of selecting the search iteration or the optimization iteration, and select the search iteration or the optimization iteration by the roulette method; if the search iteration is selected, proceed to step (6); otherwise, proceed to step (7); (6) Calculate the iteration probability of each search iteration interval, select the search iteration interval by roulette method, and calculate the iteration probability according to the fitness function , calculate the fitness value , update the interval search iteration parameters, if the fitness value Not less than the global maximum fitness value , then update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters in sequence, and go to step (8); otherwise, directly update the interval and global parameters and go to step (8); (7) Calculate the input value of the optimization iteration according to the fitness function , calculate the fitness value , determine the interval where the input value of the optimization iteration is located, update the optimization iteration reference matrix, update the optimization iteration probability parameter, update the interval and global parameters, and enter step (8); (8) If the current global calculation times are less than the preset calculation times, return to step (5); otherwise, solve the task allocation scheme through the allocation relationship model based on the current global maximum input value and the global maximum fitness value.

2. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 1 is characterized in that: In step (2), the uncertainty matrix is ​​the decision matrix of the distribution relationship model , Allocation Quantity Matrix , unordered allocation matrix , Task Allocation Matrix , Task Time Matrix ; Task execution time matrix of task model , Task expected benefit matrix , Complete the task benefit matrix ; UAV flight time matrix for UAV model , UAV mileage matrix , Drone exposure duration matrix .

3. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 2 is characterized in that: Decision Matrix of the Allocation Relationship Model No. i Line j Elements of a column for ； in, Indicates drone With the task The distribution relationship, Representation Task Assign to drone ; Representation Task Not assigned to drones ; express The j-th element of It is a base-N value, obtained by rounding down the decimal input value x. ,Will Convert to get; Allocation Quantity Matrix middle for ； in, Indicates drone The number of tasks assigned; Unordered Allocation Matrix middle for ； in, Indicates drone Disorderly assigned tasks; Indicates that no drone was given Assigned to tasks; Indicates that the task Assign to drone ,and , The decision matrix The column number of the j-th element with a value of 1 in the i-th row; Task Allocation Matrix By unordered distribution matrix Sure; Task Time Matrix middle for ； in, Indicates drone Execute the task time, Indicates drone Starting point and mission The distance between ; Indicates drone speed, Indicates the task With the task The distance between , .

4. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 3 is characterized in that: Task execution time matrix of task model middle for ； in, Indicates the task Execution time; Task Estimated Profit Matrix middle for ； ； In the formula, Constraint parameters for the importance of the task; and is the urgency constraint parameter of the task, where It is manifested as the linear change of task value with execution time. It is manifested as the exponential change of task value with execution time; and is the periodic constraint parameter of the task, where It is manifested as the cycle of changes in task value over execution time; and is the execution time window constraint parameter of the task, where The value of reflects the reward for executing the task within the execution time window. Reflects the penalty for executing tasks outside the execution time window; The time interval required to perform the task; Completion of task benefit matrix middle for ； in, Indicates completion of task of income; For drones Is there a load? , , Representation Task Type, Indicates drone Loading load ; Indicates drone No load ; Indicates that the task is not taken Required payload to perform the task The penalty value.

5. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 4 is characterized in that: Drone flight time matrix for drone models middle for ； in, Indicates drone The total time of the voyage to complete all tasks and return as required, Indicates drone The time when the last task was executed; Indicates drone Starting point and mission The distance between , Indicates drone Assigned tasks in order , that is, the last drone assigned to the task, , ; Indicates drone speed; Indicates drone The round trip requirement, Indicates drone After completing all tasks, you need to return to the starting point. Indicates drone There is no need to return to the starting point after completing all tasks; Drone Mileage Matrix middle for ； in, Indicates drone The total mileage of the voyage in which all missions are completed and returned as required; Drone exposure duration matrix middle for ； in, Indicates drone The total exposure time to complete all tasks, Representation Task The exposure radius of .

6. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 5 is characterized in that: In step (3), the fitness function for ； ； ； ； ； ； in, Indicates drone Costs; Indicates drone Is there a load? ; Indicates load Costs; Indicates the cost of drones; Indicates the load cost; is the UAV mileage cost coefficient; is the UAV flight time cost coefficient; is the load cost factor.

7. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 6 is characterized in that: In step (4), the separation point matrix for ； ； ； in, Represents the input value of the i-th separation point; express The fitness value of is the maximum value of the domain of the input value x, and N is a constant greater than 1; Interval midpoint matrix for ； ； ； in, Representation interval The input value of the midpoint of express The fitness value of .

8. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 7 is characterized in that: In step (5), the probability of selecting the optimization iteration is and the probability of selecting a search iteration as follows: ； ； in, To preset the number of calculations, is the number of global calculations, Represents the probability calculation parameters for optimization iterations.

9. The task allocation optimization method based on the multi-machine multi-task allocation problem model MDTAP according to claim 8 is characterized in that: In step (6), i The iteration probability of the search iteration interval for ； ； ； in, Representation interval The maximum input value among ; express The fitness value of Representation interval The number of calculations.

Images