CN116187594B

CN116187594B - Multi-target multi-task path scheduling cost optimization method, device, medium and equipment

Info

Publication number: CN116187594B
Application number: CN202310465168.9A
Authority: CN
Inventors: 文凯; 马寅; 王振明; 唐亮
Original assignee: Beijing Bose Quantum Technology Co ltd
Current assignee: Beijing Bose Quantum Technology Co ltd
Priority date: 2023-04-27
Filing date: 2023-04-27
Publication date: 2023-09-08
Anticipated expiration: 2043-04-27
Also published as: CN116187594A

Abstract

The application relates to a multi-target multi-task path scheduling cost optimization method, a multi-target multi-task path scheduling cost optimization device, a multi-target multi-task path scheduling medium and multi-target multi-task path scheduling equipment, belongs to the technical field of task management optimization, and solves the problem that the calculation time is long along with the increase of the problem scale in the existing optimization model. The technical scheme of the application mainly comprises the following steps: determining a decision space according to the targets, the tasks and the order, wherein the decision space is provided with a plurality of binary decision variables; acquiring the scheduling cost between any task pair; establishing an overall cost representation model according to the binary decision variables and the scheduling cost, wherein the overall cost representation model comprises the sum of all possible scheduling costs; determining a model constraint condition according to a scheduling requirement, constructing a penalty term according to the model constraint condition, and adding the whole cost representation model so that the whole cost representation model is a secondary unconstrained binarization model; and solving the overall cost representation model by adopting a quantum computer to determine the value of any decision variable, and distributing tasks and orders for the targets according to the decision variable.

Description

Multi-target multi-task path scheduling cost optimization method, device, medium and equipment

Technical Field

The application belongs to the technical field of task management optimization, and particularly relates to a multi-target multi-task path scheduling cost optimization method, device, medium and equipment.

Background

For the multi-objective multi-task path allocation scheduling optimization problem, in real applications such as optimization tasks scheduled by AGVs (Automated Guided Vehicle) in a port, a plurality of AGVs are needed to execute a plurality of cargo handling tasks in the port, and each task corresponds to a starting point and a transportation end point. The optimization problem is to calculate the optimal scheduling scheme of the AGVs to execute tasks (i.e. which AGVs each task is assigned to and the task execution sequencing of each AGV) so that the total cost of completing all tasks is minimized.

Current computing techniques are based on optimization algorithms of conventional computers, including for example, accurate solution algorithms (branch-and-bound, etc. algorithms) based on integer programming and mixed integer programming mathematical models, heuristic algorithms, deep reinforcement learning, etc.

The prior art has difficulty in calculating the large-scale port AGV scheduling, the calculation complexity is exponentially increased along with the increase of the problem scale (the increase of the number of solving tasks and the AGVs), the solving difficulty is rapidly increased, and the prior solving technology is difficult to finish solving in a short time.

The application aims to provide an allocation scheduling optimization method for a multi-target multi-task path, which solves the problem that in the prior art, the calculation time is longer along with the increase of the scale of the problem.

Disclosure of Invention

In view of the above analysis, the embodiments of the present application aim to provide a method, apparatus, medium and device for optimizing multi-objective multi-task path scheduling cost, so as to solve the problem that the existing cost optimization model has longer calculation time with the increase of the problem scale.

An embodiment of a first aspect of the present application provides a method for optimizing a multi-target multi-task path allocation scheduling cost, including the steps of:

determining a decision space according to a target, a task and an order, wherein the decision space is provided with a plurality of binary decision variables, and any binary decision variable represents whether a certain target executes a certain task in a certain order;

acquiring the scheduling cost between any task pair;

establishing an overall cost representation model according to the binary decision variable and the scheduling cost, wherein the overall cost representation model comprises the sum of all possible scheduling costs;

determining a model constraint condition according to a scheduling requirement, constructing a penalty term according to the model constraint condition and adding the whole cost representation model, so that the whole cost representation model is a secondary unconstrained binarization model, and generating penalty cost for decision variables which do not meet the model constraint condition by the penalty term; the objective function of the overall cost representation model is to minimize the sum of the penalty cost and the schedule cost;

And solving the overall cost representation model by adopting a quantum computer to determine the value of each binary decision variable, and distributing the tasks and the orders for the targets according to the binary decision variables.

In some embodiments, the determining a decision space according to the goals, tasks, and orders includes:

defining a set V, v= { a1, a2, a3, … } of the targets;

defining a set R of the tasks, r= { task 1, task 2, task 3, … task n-1};

defining the order space T, t= {1,2, …, n };

defining the binary decision variable according to the objective, the task and the order, wherein If the kth object performs task r in order t +.>=1, otherwise->=0; all possible binary decision variables form the decision space.

In some embodiments, the obtaining the scheduling cost between any pair of tasks includes:

construct task pair set a, a= { (task 1, task 2), (task 1, task 3), … };

the tasks include a first task r and a second taskObtaining the path of the first task rDiameter and the second task->Is a starting point position of (2);

acquiring the target to complete the first task r and reach the second task Cost of movement of the origin position of (2)As the task pair (r,/>) Wherein (r,/-)>）/>A。

In some embodiments, the sum of the scheduling costs is expressed as:

, wherein />Representing task pairs (r,)>) Is used to determine the scheduling cost of (a),the decision variable representing the target k t-th order execution task r, A representing the task pair set, and V representing the target set.

In some embodiments, the scheduling requirement includes:

any one of the targets starts from a common starting point, and the corresponding model constraint condition is expressed as:defining task 0 as exiting the common origin;

moving any one of the targets to a common end point after completing all assigned tasks, wherein the corresponding model constraint condition is expressed as:defining a task n as reaching the common end point;

any one of the tasks is completed once by one of the targets, and the corresponding model constraints are expressed as:；

the same target performs only at most one task in one of the orders, the corresponding model constraints are expressed as:and->； and

any one of the targets sequentially executes the tasks according to the order, and the corresponding model constraint conditions are expressed as follows:；

wherein ,Decision variables representing target k t-th order execution task R, a representing task pair set, V representing target set, r= { task 1, task 2, task 3, … task n-1}, "for example>= { task 0, task 1, task 2, … task n-1}, +.>= { task 1, task 2, task 3, … task n }.

In some embodiments, the constructing a penalty term according to the model constraint and adding the overall cost representation model such that the overall cost representation model is a quadratic unconstrained binarization model comprises:

rewriting the model constraint into a form of f (x) =0, and then squaring the left side of the equation and multiplying the square by a penalty coefficient to form a penalty term;

adding the penalty term into the overall cost representation model to form a secondary unconstrained binarization model, wherein the model is expressed as:

；

wherein ,representing task pairs (r,)>) Scheduling costs of->Decision variables representing target k t-th order execution task r, A representing task pair sets, V representing target sets,/for>、/>All represent penalty coefficients, R= { task 1, task 2, task 3, … task n-1}, ∈ ->= { task 0, task 1, task 2, … task n-1}, +.>= { task 1, task 2, task 3, … task n }.

An embodiment of a second aspect of the present application provides a method for optimizing a multi-target multi-task path allocation scheduling cost, including the steps of:

acquiring the scheduling cost between any task pair;

determining model constraint conditions according to scheduling requirements; the objective function of the overall cost representation model is to minimize the sum of the scheduling costs when the model constraints are met;

and solving the overall cost representation model to determine the value of each binary decision variable, and distributing the tasks and the orders for the targets according to the binary decision variables.

An embodiment of a third aspect of the present application provides a multi-target multi-task path allocation scheduling cost optimization device, including:

the decision variable determining module is used for determining a decision space according to a target, a task and an order, wherein the decision space is provided with a plurality of binary decision variables, and any binary decision variable represents whether a certain target executes a certain task in a certain order;

The scheduling cost acquisition module is used for acquiring the scheduling cost between any task pair;

an overall cost representation module for building an overall cost representation model from the binary decision variables and the scheduling costs, the overall cost representation model comprising a sum of all possible scheduling costs;

the condition unconstrained module is used for determining a model constraint condition according to a scheduling requirement, constructing a penalty term according to the model constraint condition and adding the whole cost representation model, so that the whole cost representation model is a secondary unconstrained binarization model, and the penalty term generates penalty cost for decision variables which do not meet the model constraint condition; the objective function of the overall cost representation model is to minimize the sum of the penalty cost and the schedule cost;

and the quantum computing module adopts a quantum computer to solve the whole cost representation model to determine the value of any binary decision variable, and distributes the tasks and the orders for the target according to the binary decision variable.

An embodiment of a fourth aspect of the present application provides an electronic device, including a memory and a processor, where the memory stores a computer program that, when executed by the processor, implements a multi-objective multi-tasking path allocation scheduling cost optimization method according to any of the embodiments above.

A fifth aspect of the present application provides a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a multi-objective multi-tasking path allocation scheduling cost optimization method according to any of the embodiments above.

The embodiment of the application has at least the following beneficial effects:

1. and setting a binary decision variable according to a scheduling target to form a decision space, so that the decision space covers all possible decision possibilities, the representation of the cost optimization model is facilitated, and the construction of a secondary unconstrained binarization model through the binary decision variable is facilitated.

2. By constructing the cost optimization model into a secondary unconstrained binarization model, the solution of the model can be realized through a quantum computer, the problem of calculation efficiency of a traditional computer optimization algorithm in calculating a scheduling strategy problem is solved, and the solution efficiency of the cost optimization problem on a larger scale can be improved, so that the running cost is reduced, the efficiency is improved, and the effects of saving cost, energy consumption, time and the like are achieved.

Drawings

In order to more clearly illustrate the embodiments of the present description or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the embodiments of the present description, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.

Fig. 1 is a flow chart of a multi-objective multi-task path allocation scheduling cost optimization method according to an embodiment of the first aspect of the present application;

FIG. 2 is a schematic diagram of an exemplary AGV schedule according to an embodiment of the present application;

FIG. 3 is a graph of the relationship between the number of different tasks and the calculation time of the optimization problem in a conventional optimization model in a common computer;

FIG. 4 is a graph of the point model provided by the embodiment of the application for different task numbers and calculation time of an optimization problem;

FIG. 5 is a flowchart of a method for optimizing multi-objective multi-task path allocation scheduling cost according to a second embodiment of the present application;

FIG. 6 is a schematic diagram of a multi-objective multi-task path allocation and scheduling cost optimization device architecture provided by the present application;

fig. 7 is a schematic diagram of an electronic device architecture according to the present application.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. It should be noted that embodiments and features of embodiments in the present disclosure may be combined, separated, interchanged, and/or rearranged with one another without conflict. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Furthermore, when the terms "comprises" and/or "comprising," and variations thereof, are used in the present specification, the presence of stated features, integers, steps, operations, elements, components, and/or groups thereof is described, but the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof is not precluded. It is also noted that, as used herein, the terms "substantially," "about," and other similar terms are used as approximation terms and not as degree terms, and as such, are used to explain the inherent deviations of measured, calculated, and/or provided values that would be recognized by one of ordinary skill in the art.

The method and the device are applied to optimization of a transport vehicle dispatching scheme, such as port AGV dispatching, intelligent warehouse AGV dispatching, automatic driving vehicle dispatching and other scenes.

An embodiment of the first aspect of the present application provides a method for optimizing multi-objective multi-task path allocation scheduling cost, as shown in fig. 1, including the following steps:

acquiring the scheduling cost between any task pair;

It should be appreciated that scheduling cost optimization problem refers to calculating an optimal scheduling scheme for the vehicle to perform tasks (i.e., which vehicle each task is assigned to and the order in which the tasks of each vehicle are performed) such that the total cost of completing all tasks is minimized. For the convenience of understanding the present embodiment, the present embodiment of the present application is described with reference to the scheduling of AGVs in a port, where a plurality of cargo handling tasks are required to be performed by a plurality of AGVs, each task corresponding to a start point and a transport destination point, and the plurality of AGVs start from a common start point to complete the assigned transport tasks and then travel to the common destination point. Therefore, the sum of the moving costs between tasks becomes the optimization objective of the present embodiment, where the moving cost between tasks is defined as the moving cost generated in the process of moving the goods of the previous task to the goods position of the next task after the goods of the previous task are carried out, and the moving cost can be specific information such as the running time and the oil consumption of the AGV, and is related to the distance between the end point of the previous task and the start point of the next task.

Correspondingly, the target is the AGV in the embodiment, the task is the task that the AGV needs to transport the goods from the location to the destination, and the order is the task that the AGV needs to execute. The decision is thus targeting the tasks assigned by each AGV and their order of execution. In this embodiment, a binary decision variable is set for each action content of each vehicle, which indicates whether the decision of executing a task in a certain order of the AGVs occurs, so that the path allocation decision of each AGV can be represented by a sequence of binary decision variables, and the decision of the whole scheduling cost optimization problem can be understood as a set of decision sequences of all AGVs.

Since the starting point and the ending point of each task are known, the moving cost between any two task pairs is also known, adding the moving between all the task pairs after the scheduling cost between the task pairs is obtained is the target to be optimized in this embodiment, and accordingly, an overall cost representation model is established (where the overall cost representation model only includes the cost optimization target, that is, the sum of all possible scheduling costs), and then, the model constraint condition can be determined according to the scheduling requirement or the formulated scheduling logic, so that the objective function of the overall cost representation model is to minimize the sum of the scheduling costs when the model constraint condition is satisfied.

In some embodiments, for the AGV scheduling cost optimization problem, the scheduling requirements include:

1. at the beginning of the schedule, the AGVs start from a common starting point.

2. One task is accomplished by and only by one AGV.

3. Each AGV performs tasks in turn.

4. Each AGV performs at most one task per order.

5. The AGV moves to the same destination after completing the task assigned to it.

And finally, solving the constructed overall cost representation model to determine the value of each decision variable, and distributing the tasks and the orders for the targets according to the decision variables, so that the AGV can complete the cargo handling task under the condition of lowest overall cost.

The combined optimization is a mathematical optimization technology, and consists of discrete decision variables, objective functions and constraint conditions, and aims to solve the optimal values of the objective functions and the corresponding optimal solutions. If the objective function is quadratic and without constraints and the decision variables can only take 0 or 1, then this combined optimization is called a quadratic unconstrained binary optimization model (QUBO, quadratic UnconstrainedBinary Optimization). The QUBO model form can be solved by a quantum annealing machine or a special quantum computer such as coherent i Xin Ji. The problem is solved in an acceleration way by using the quantum computer, so that the calculation efficiency is improved, and the problem that the complex scheduling problem is difficult to calculate or the calculation consumes long time in the prior art is solved.

In this embodiment, a decision space is formed by setting a binary decision variable according to a scheduling target, so that the decision space covers all possible decision possibilities, which is favorable for the representation of a cost optimization model and is favorable for constructing a mathematical model which is easy to solve, such as a secondary unconstrained binarization model, through the binary decision variable.

Preferably, in some embodiments, the determining a decision space according to the goals, tasks, and orders includes:

defining a set V, v= { a1, a2, a3, … } of the targets;

defining a set R of the tasks, r= { task 1, task 2, task 3, … task n-1};

defining the order space T, t= {1,2, …, n };

Specifically, the set and data used by the port AGV scheduling optimization problem are defined as follows:

an AGV set V for port scheduling is defined, such as V= { AGV1, AGV2, … }, here assuming that all AGVs used by the port perform the same.

A set of scheduled tasks R is defined, such as r= { task 1, task 2, … }, a task referring to the process of an AGV transporting goods from its location to a specified location. Since each AGV starts from a common start point, task 0 may be defined as a virtual task departure start point. Similarly, the AGVs complete the respective tasks and then aggregate to a common endpoint, and the virtual task can be defined to return to the endpoint as task n. In order to define more precisely the different task set ranges in the subsequent mathematical model, r= { task 1, task 2, …, task n-1} is defined here as the actual task set, = { task 0, task 1, …, task n-1} is the set containing starting virtual task 0, +.>= { task 1, task 2, …, task n } is a set containing endpoint virtual task n, and R total= { task 0, task 1, …, task n } is a total task set.

Each AGV performs tasks in a certain order. The present embodiment indicates the serial number of the task order executed by the AGV with t=1, 2, …. An AGV performs a task corresponding to and only corresponding to a sequence number. In order to facilitate the establishment of a subsequent mathematical model, the sequence number ranges of the AGVs may be unified, and according to the definition of the task set, the sequence number value of each AGV may be unified from 1 to n, and if one AGV has completed all assigned tasks in a certain order, the subsequent sequence number remains, although the AGV does not perform any tasks at this time.

Specifically, for the goal of minimizing overall task execution costs, the present embodiment builds the following QUBO model based on the above logic. Defining binary variablesRepresents whether the task r is executed by the kth order of the kth AGV, if executed +.>Taking 1, otherwise taking 0. One sequence of AGVs corresponds to the process of completing the transport of the corresponding task from the start position of one task point and going to the start position of the next task point.

Preferably, in some embodiments, the acquiring the scheduling cost between any pair of tasks includes:

construct task pair set a, a= { (task 1, task 2), (task 1, task 3), … };

the tasks include a first task r and a second taskAcquiring the path of the first task r and the second task +.>Is a starting point position of (2);

acquiring the target to complete the first task r and reach the second taskCost of movement of the origin position of (2)As the task pair (r,/>) Wherein (r,/-)>）/>A。

Specifically, a set a formed between two different tasks is defined, such as a= { (task 1, task 2), (task 3, task 2), … }, from the perspective of graph theory, if each task is a node, the elements of a may be referred to as directed arcs, i.e., directed task pairs. The directed arc set a contains only valid arcs, and if the AGV cannot directly go from one task to another during actual dispatch operation, the corresponding arc is not contained in the set a.

Then constructing cost data between task pairs, namely, AGVs move goods of the task r from the position to the task destination to the next task The cost of the goods in the position of the goods can be specifically information such as AGV running time, oil consumption and the like, and can be expressed as +.>Wherein r, & gt>Belonging to a task set R.

As shown in fig. 2, which is a schematic diagram of an AGV with a small scheduling problem, fig. 2 is an example of scheduling of 2 AGVs with 4 tasks, wherein the square indicates the cargo positions of different tasks, and the parallelogram indicates the destination corresponding to the tasks. From a common starting point, one AGV performs tasks 1 and 3 in the first and second order, respectively, and the other AGV performs task 2, task 4, respectively. After the task is performed, the two AGVs move to a common end point.

In some embodiments, the sum of the scheduling costs is expressed as:

, wherein />Representing task pairs (r,)>) The scheduling cost of the task r, i.e. AGV moves the goods of the task r from the position to the task destination to the next task +.>Where the cost can be a known amount for a particular problem for a particular AGV run time, fuel consumption, etc., for the cargo location>The decision variable representing the target k t-th order execution task r, A representing the task pair set, and V representing the target set.

It should be appreciated that due to Only 0 or 1 can be taken, here +.>If equal to 1, it means that the kth AGV performs tasks r and +.1 in the order of t and t+1, respectively>At the same time the expression +.>The secondary expression of (2) meets the requirement of QUBO. The costs corresponding to all possible tasks performed in all consecutive order by all AGVs are summed up here, due to +.>Is a variable, this cost sum is also a variable, is obtained by solving all + ->The minimum target of the sum of the formulas is obtained, namely the minimum scheduling cost is obtained, and the cost optimization target is completed.

Further, in some embodiments, the scheduling requirement includes:

the same target performs only at most one task in one of the orders, the corresponding model constraints are expressed as: And->； and

Preferably, in some embodiments, said constructing a penalty term according to said model constraints and adding said overall cost representation model such that said overall cost representation model is a quadratic unconstrained binarization model comprises:

；

wherein ,representing task pairs (r,)>) Scheduling costs of->Decision variables representing target k t-th order execution task r, A representing task pair sets, V representing target sets,/for>、/>All represent penalty coefficients, R= { task 1, task 2, task 3, … task n-1}, ∈ - >= { task 0, task 1, task 2, … task n-1}, +.>= { task 1, task 2, task 3, … task n }.

Specifically, for the above constraint, in order to make it a part of the objective to satisfy the QUBO form, each equation is rewritten first to the form of f (x) =0, then the left side is squared, and a penalty coefficient large enough is added to put in the objective. So that when the equation is 0, i.e. the constraint is satisfied, the target term is 0, no significant penalty occurs. When the constraint is not satisfied, the penalty will take effect. A first constraint such as that described above may be written asWhere M is a sufficiently large penalty factor, since the constraint applies to any AGVs, all AGVs can be summed in the target, written as. For the second constraint, due to +.>The value is 0 or 1->Is equivalent to->So write objective function is +>. The overall cost representation of the above described QUBO form is finally obtained. Preferably, the penalty coefficient of each penalty term can be the same or different, and is adjusted according to the actual requirement.

The QUBO model of the scheme is suitable for special quantum computers such as a quantum annealing machine, coherent I Xin Ji (Coherent Ising Machine, CIM) and the like. The physical machine is implemented by taking CIM based on a degenerate optical parametric oscillator (Degenerate OpticalParametric Oscillator, DOPO) as an example, which is a mixed quantum computing system consisting of an optical part and an electrical part. The optical portion includes a laser, an amplifier, a periodically poled lithium niobate crystal (Periodically Poled LithiumNiobate, PPLN), and a fiber loop. The laser uses a femtosecond pulse fiber laser and is matched with an amplifier system. The amplified laser is first frequency doubled by using PPLN crystal. The frequency-doubled laser is used as a pumping source to synchronously pump a PPLN crystal in an optical fiber loop, so as to form degenerate optical parametric oscillation, and hundreds of oscillation pulses can exist in the optical fiber loop at the same time. The electrical section includes FPGA (Field Programmable GateArray), AD/DA (digital analog/digital conversion), and a phase detecting section. The laser output in the optical fiber loop and the laser with the fundamental frequency are measured by a phase detector, so that the phase of the output light can be tested. And the FPGA is matched with the high-speed AD/DA to measure and feedback control the optical pulse.

Unlike classical computers running on semiconductor integrated circuits, CIM uses laser pulses in an optical fiber as qubits for computation. In DOPO, the pump light is incident on the nonlinear optical crystal to split two beams of light, the polarization direction of the two beams of light is the same, the frequency is half of that of the pump light, and the pump light is in a compressed state and can be used as a qubit. When the power of the pumping light is increased gradually and exceeds the oscillation threshold, the generated light becomes coherent state, the phase of the light is divided into two states (phase 0 state and pi state), and the phase can be set correspondingly to spinAnd 1, solving the scheduling cost optimization problem of the QUBO model.

Next, experimental data will be described for the QUBO scheduling cost optimization model provided by the present embodiment and for the experimental data comparison using a conventional optimization algorithm, which takes a mixed integer programming (mixed integer programming, MIP) mathematical model as an example. The scheduling cost optimization model provided by the present embodiment is hereinafter referred to as a point (Node) model since its variables are set for the target, task nodes, and intersection points of the order. The following will make experimental comparisons of the impact of two dimensions on computation time from the scale of the scheduling optimization problem (number of tasks, number of targets) and computing device (using conventional computers or quantum computation).

Firstly, the relation between the task number and the calculation time is a relation chart of the calculation time along with the task number in the problem of optimizing the scheduling cost on a traditional computer by adopting MIP as shown in FIG. 3; as shown in fig. 4, in order to obtain a graph of the task number and the calculation time calculated on a common computer using the point model provided in this embodiment, two experiments were performed for the same number of AGVs. It can be seen that when the task number is greater than 9, the calculation time of the MIP model increases sharply, and the change trend of the point model provided in this embodiment is similar, but the overall time consumption is far lower than that of the MIP, the calculation time is not longer than 12 seconds when the task number is 13, and the calculation time of the MIP model is approximately 30 seconds when the task number is 10.

The relationship between the target number and the calculation time is shown in the following table 1, and the calculation time for solving the scheduling cost optimization problem for the same task number and different AGV numbers is executed on a common computer by adopting MIP or Node respectively, wherein the instance (instances) condition is represented by a mode of "task number-target number", for example, 10-2 represents a scheduling cost optimization instance of two AGVs with 10 tasks, and the time unit is seconds. From the graph, the calculation time of using MIP is far higher than that of the point model of the embodiment of the application.

Table 1 cost optimization problem Using a general computer when calculated (Unit s)

Experimental data for the MIP model and the Node model run in a quantum computer to solve the scheduling cost optimization problem described above are shown in table 2 below, with examples expressed in milliseconds as described above. It can be known that the dot model provided in this embodiment is more suitable for solving in a quantum computer, the total computation time is far lower than that in a common computer, and compared with the MIP model solved by using a quantum computer, the dot model in this embodiment has no increase in computation time with the number of tasks in the example range, and the MIP model has a significant trend of increasing computation time.

Table 2 cost optimization problem Using Quantum computers when calculated (in ms)

In summary, the cost optimization model is constructed as the secondary unconstrained binarization model, so that the solution of the model can be realized through a quantum computer, the problem of the calculation efficiency of the traditional computer optimization algorithm in calculating the scheduling policy problem is solved, the solution efficiency of the cost optimization problem on a larger scale can be improved, the running cost is reduced, the efficiency is improved, and the effects of saving the cost, reducing the energy consumption, saving the time and the like are achieved.

An embodiment of the second aspect of the present application provides a method for optimizing multi-objective multi-task path allocation scheduling cost, as shown in fig. 5, including the following steps:

acquiring the scheduling cost between any task pair;

and solving the overall cost representation model to determine the value of any binary decision variable, and distributing the tasks and the orders for the target according to the binary decision variable.

The specific manner of this embodiment may be implemented in a model and unconstrained manner as provided in the embodiment of the first aspect, and the QUBO is constructed and solved by using a quantum computer, which will not be described in detail herein. The binary decision variables defined in this embodiment may also be used to construct other specific solution models for cost solutions. For example, modifications and extensions to the model are possible. The above model is applicable to AGVs having a unified start node and a unified end node. In a realistic scenario, the AGV has different start and end nodes, and the model described above is easily scalable for these types of situations. The solution is to relate to the number of virtual start and end tasks, depending on the number of AGVs. The proposed model is a pure scheduling problem that does not consider path optimization, since it is currently very difficult to map the large-scale optimization problem to the QUBO form because of the small number of bits available for quantum computation. Of course, in the case of mature technology, this possibility can be taken into account later. And a two-layer planning model can be established on the basis of the existing model, and the scheduling and path planning problems are solved alternately in the two sub-models, so that the modeling difficulty and the bit number used for solving a single model are reduced.

The quantum computer with the ultra-large number of bits can be introduced into AGV scheduling, so that the actual problem is solved better, and the calculation characteristics under the background of a large example are researched.

In addition, the quantum computer and the traditional computer can be combined to study the problem of AGV scheduling cost optimization, and the respective characteristics of the quantum computer and the traditional computer can be better utilized to improve the efficiency of solving the problem.

An embodiment of a third aspect of the present application provides a multi-target multi-task path allocation scheduling cost optimization apparatus, as shown in fig. 6, including:

An embodiment of a fourth aspect of the present application provides an electronic device, as shown in fig. 7, including a memory and a processor, where the memory stores a computer program, where the computer program, when executed by the processor, implements a multi-objective multi-task path allocation scheduling cost optimization method according to any of the embodiments above.

Computer-readable storage media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, read only compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by the computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of function in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the application, and is not meant to limit the scope of the application, but to limit the application to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the application are intended to be included within the scope of the application.

Claims

1. The multi-target multi-task path allocation scheduling cost optimization method is characterized by being applied to optimization of a transportation vehicle scheduling scheme and comprising the following steps of:

determining a decision space according to a target, a task and an order, wherein the decision space is provided with a plurality of binary decision variables, and any binary decision variable represents whether a certain target executes a certain task in a certain order; the determining a decision space according to the targets, tasks and orders comprises:

defining a set V, v= { a1, a2, a3, … } of the targets;

defining a set R of the tasks, r= { task 1, task 2, task 3, … task n-1};

defining the order space T, t= {1,2, …, n };

defining the binary decision variable according to the objective, the task and the order, wherein If the kth object performs task r in order t +.>=1, otherwise->=0; all the binary decision variables form the decision space;

acquiring the scheduling cost between any task pair, including:

construct task pair set a, a= { (task 1, task 2), (task 1, task 3), … };

the tasks include a first task r and a second taskAcquiring the path of the first task r and the second task +. >Is a starting point position of (2);

acquiring the target to complete the first task r and reach the second taskMoving costs of origin position->As the task pair (r,/>) Wherein (r,/-)>）/>A, the moving cost comprises moving time or oil consumption;

establishing a whole cost representation model according to the binary decision variable and the scheduling cost, wherein the whole cost representation model comprises the sum of all the scheduling costs, and the sum of the scheduling costs is expressed as:

, wherein />Representing task pairs (r,)>) Scheduling costs of->A decision variable representing a target k t-th order execution task r, wherein A represents a task pair set, and V represents a target set;

determining a model constraint condition according to a scheduling requirement, constructing a penalty term according to the model constraint condition and adding the whole cost representation model, so that the whole cost representation model is a secondary unconstrained binarization model, and generating penalty cost for decision variables which do not meet the model constraint condition by the penalty term; the objective function of the overall cost representation model is to minimize the sum of the penalty cost and the schedule cost; the scheduling requirement includes:

wherein ,decision variables representing target k t-th order execution task R, a representing task pair set, V representing target set, r= { task 1, task 2, task 3, … task n-1}, "for example>= { task 0, task 1, task 2, … task n-1}, +.>= { task 1, task 2, task 3, … task n };

2. The multi-objective multi-tasking path allocation scheduling cost optimization method of claim 1 wherein: the method comprises the steps of constructing a punishment item according to the model constraint condition and adding the whole cost representation model, so that the whole cost representation model is a secondary unconstrained binarization model, and comprises the following steps:

；

3. The multi-target multi-task path allocation scheduling cost optimization method is characterized by being applied to optimization of a transportation vehicle scheduling scheme and comprising the following steps of:

determining a decision space according to a target, a task and an order, the decision space having a number of binary decision variables, any of the binary decision variables representing whether a certain target performs a certain task in a certain order, the determining a decision space according to a target, a task and an order comprising:

Defining a set V, v= { a1, a2, a3, … } of the targets;

defining a set R of the tasks, r= { task 1, task 2, task 3, … task n-1};

defining the order space T, t= {1,2, …, n };

acquiring the scheduling cost between any task pair, including:

construct task pair set a, a= { (task 1, task 2), (task 1, task 3), … };

determining model constraint conditions according to scheduling requirements; the objective function of the overall cost representation model is to minimize the sum of the scheduling costs when the model constraints are met; the scheduling requirement includes:

any of the targets completes the assigned stationAfter the tasks are carried out, the tasks are moved to a common end point, and the corresponding model constraint conditions are expressed as follows:defining a task n as reaching the common end point;

wherein ,decision variables representing target k t-th order execution task R, a representing task pair set, V representing target set, r= { task 1, task 2, task 3, … task n-1}, "for example >= { task 0, task 1, task 2, … task n-1}, +.>= { task 1, task 2, task 3, … task n };

4. A multi-target multi-task path allocation scheduling cost optimizing apparatus for performing the multi-target multi-task path allocation scheduling cost optimizing method according to any one of claims 1 to 3, comprising:

the overall cost representation module is used for building an overall cost representation model according to the binary decision variable and the scheduling cost, wherein the overall cost representation model comprises the sum of all the scheduling costs;

5. An electronic device comprising a memory and a processor, the memory storing a computer program that when executed by the processor implements the multi-objective multi-tasking path allocation scheduling cost optimization method of any of claims 1-3.

6. A computer readable storage medium, having stored thereon a computer program which when executed by a processor implements the multi-objective multi-tasking path allocation scheduling cost optimization method according to any of the claims 1-3.