Multi-person acquisition scheduling method based on improved ant colony algorithm

Abstract

The invention provides a scheduling system and method for multi-person acquisition. According to the method, on the basis of an improved ant colony algorithm, personnel and scheduling information is coded, an ant colony is guided to perform heuristic search on a solution space according to a pheromone factor and a heuristic factor, an optimal path of the ant colony is continuously updated by setting an iteration threshold value until an optimal solution is obtained, and then the optimal solution is decoded to obtain a final acquisition scheme. The defects that traditional integer programming is difficult to solve and long in consumed time are overcome to a great extent. According to the system and the method, the acquisition scheme can be quickly generated according to different acquisition scenes and problems, and the system and the method have the characteristics of flexibility, operability, economy and the like. And technical support is provided for medical acquisition planning and workshop scheduling under specific conditions.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to a multi-personnel collection scheduling method based on an improved ant colony algorithm. Background Art

[0002] In modern military and industrial working environments, the collection, monitoring and analysis of human body data of relevant personnel is one of the main responsibilities of the health system. By collecting and applying key human body measurement data of relevant personnel, such as plantar pressure, body composition, bone density, psychological status and other characteristic data, it can play an important role in job allocation, scientific coordination, disease prevention, personnel management and other aspects.

[0003] To collect the above-mentioned characteristic data, the personnel being collected need to complete the measurement item by item. When there are a large number of personnel being collected, if they are not scheduled, or all the personnel being collected are collected in a fixed order, it is very easy to cause congestion in some collection items, resulting in low collection efficiency and huge work pressure on the staff of grassroots units such as medical institutions. At this time, the scheduling of the personnel being collected becomes particularly important.

[0004] The multi-personnel scheduling problem of medical service institutions mainly involves the scheduling of a single department and the serial scheduling of multiple departments. The scheduling problem of a single department is usually modeled and solved using queuing theory, while the traditional method for serial scheduling of multiple departments is the integer programming method. In the actual scheduling problem, in order to obtain an ideal scheduling solution, it is necessary to abstract the actual problem into a mathematical model and make some assumptions and constraints on the actual problem, such as the scheduled personnel are all positive integers greater than 0; a machine can only collect one person at the same time; the number of personnel collected in project k at any time must be less than or equal to the number of machines in the project, etc. At the same time, it is necessary to construct a time-optimal or space-optimal objective function to find the optimal solution, which is the ideal scheduling solution. However, when the problem scale is large, integer programming needs to solve a large number of inequalities and equations, and traverse all solution spaces at the same time, which requires a large amount of memory resources, takes a long time to solve, and even cannot find the ideal optimal solution. Therefore, meta-heuristic algorithms (such as ant colony algorithm, simulated annealing algorithm, genetic algorithm, etc.) are used to solve the problem. The ant colony algorithm has the advantages of positive feedback and distributed computing, but it is easy to fall into local optimality when searching for feasible solutions in large-scale scheduling. Therefore, it has been improved and applied to scheduling problems by many scholars in recent years. The dynamic transfer strategy proposed in it improves the convergence speed of the algorithm. By introducing a stage-based encoding method, the iterative process of the algorithm is greatly reduced, and the running time of the algorithm is reduced. By introducing the remaining work heuristic strategy, the quality of the solution is further improved, the search range of the solution space is narrowed, and the situation of falling into local optimality is solved. Summary of the invention

[0005] The purpose of the present invention is to provide a collection scheme system for large-scale medical human body collection that can meet the requirements of flexibility, operability, economy, etc. The system can quickly generate collection schemes according to different collection scenarios and problems, and provide technical support for medical collection planning and workshop scheduling arrangements under specific circumstances.

[0006] According to one aspect of the present invention, an improved ant colony algorithm is provided, which introduces a stage-based encoding method and uses the encoding as algorithm input. Secondly, a remaining work heuristic strategy is designed to reduce the algorithm's search space and improve the algorithm's efficiency.

[0007] According to another aspect of the present invention, a multi-personnel collection scheduling system based on an improved ant colony algorithm is provided, which includes three submodules: a collection variable input module, an algorithm parameter input module, and a collection plan output module. The collection variable input module is used to input the collection conditions in the actual scene, including two parts, one part is the collection condition input, including the number of collection projects, the number of machines for each project, and the collection time required for each project; the other part is the collection personnel condition input, including the total number of people and the number of people in the team. The algorithm parameter input module is used to input the parameters required for the operation of the ant colony algorithm; the collection plan output module is used to generate the final collection plan, including two forms of Gantt chart output and text output.

[0008] Advantages of the present invention include:

[0009] (1) A large-scale medical collection scheduling algorithm is proposed. The ant colony algorithm is applied to collection scheduling by constructing a solution representation method. The pheromone heuristic information of the ant colony is used to provide a positive feedback mechanism for the ant colony, that is, the ants will gather on the path with more pheromones, and then the ant colony solution is continuously iterated through the fitness function to finally obtain an approximate optimal solution. It avoids the need for traditional integer programming methods to explore the entire solution space and set a large number of constraints on personnel, projects and machines, such as processing time upper limit constraints, collection order constraints, and machine quantity and collection time interval constraints. Through the above conditions, a series of inequalities and equations are obtained, an integer programming model is established, and then a large number of equations and inequalities are solved, which greatly shortens the solution time and improves the solution efficiency. At the same time, appropriate improvements are made to the traditional ant colony algorithm to improve the algorithm convergence performance and avoid falling into local optimality.

[0010] (2) A collection and scheduling system based on the ant colony algorithm is provided. By inputting problem conditions and algorithm parameters, a scheduling plan is quickly generated, which facilitates switching between different scenarios and improves operability. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1is a flowchart of a multi-personnel collection scheduling method based on an improved ant colony algorithm according to an embodiment of the present invention;

[0012] Figures 2 to 5 The invention discloses a multi-personnel collection and dispatching system based on an improved ant colony algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0013] Improved ant colony algorithm design for multi-personnel collection scheduling

[0014] Figure 1 This is a flowchart of a multi-personnel collection scheduling method based on an improved ant colony algorithm according to an embodiment of the present invention. For n collected personnel and s collection stages, first, n×s collection tasks are formed into a one-dimensional list and numbered from 1 to n×s according to the sequence number. According to the task transfer rule of the ant colony algorithm, one task is selected from the list each time until the ant colony traverses all elements of the list, which is a feasible solution. Then, the ant colony system parameters are updated through algorithm iteration until the algorithm converges. At this time, the path taken by the ants is the optimal solution. The specific implementation steps based on the improved ant colony algorithm according to an embodiment of the present invention are as follows:

[0015] S1: Initialize parameters such as pheromone, heuristic, global pheromone volatility factor, local pheromone volatility factor and the number of random solutions.

[0016] S2: Generate m ants and initialize a random task as the starting point of the ants. For a single ant, apply the ant colony system path selection function to select the next task until n×s tasks are selected. After each selection, local pheromones and heuristics are updated.

[0017] S3: Construct m random solutions and expand the number of solutions to 2m.

[0018] S4: Deserialize all solutions, that is, convert implicit phases into phase encodings; decode all solutions and select all solutions with the minimum collection completion time.

[0019] S5: Determine whether the solution is smaller than the previous optimal solution. If so, perform pheromone update according to the global pheromone update method and output the current optimal solution.

[0020] S6: Determine whether the minimum acquisition time for the reconnection is equal to the lower limit of time. If it is, directly output the optimal solution, otherwise continue to iterate

[0021] In S2, in order to avoid the algorithm from stagnating prematurely, a selection strategy combining deterministic selection and random selection is adopted, and the state transition probability is dynamically adjusted during the search process. That is, for ant k at task i, the next task is selected according to formula (1):

[0022]

[0023] in:

[0024] τ is is the pheromone concentration between task i and task s,

[0025] is the pheromone factor, which indicates the relative importance of the amount of information accumulated during the ant movement in the ant colony search.

[0026] η is is the heuristic function, β is the heuristic function factor, which reflects the relative importance of heuristic information in ant colony search.

[0027] q 0 is a random number in the interval [0,1], an algorithm parameter (0≤q 0 ≤1), when q>q 0 When , ant k determines the target task j to which task i is transferred downward according to formula (2).

[0028] allowed k is the set of tasks to be visited by the kth ant. At the initial moment, there are n×s-1 tasks, which means that the tasks other than the tasks that the ant is in at the beginning are excluded. As time goes by, the number of tasks becomes smaller and smaller until it is empty, indicating that all tasks have been traversed;

[0029]

[0030] In the process of constructing the solution, every time the ant passes an edge (i, j), it will immediately call the local update rule to locally update the pheromone on the edge. The local update rule is shown in formula (3):

[0031] τ ij =(1-ρ)τ ij +ξτ 0 (3)

[0032] ξ and τ 0 are two parameters, ξ satisfies 0<ξ<1; τ 0 is the initial value of the pheromone amount. ρ is the pheromone volatility factor, which reflects the disappearance level of the pheromone, 0<ρ<1. The role of local update is that every time the ant passes through the edge (i, j), the pheromone τ of the edge ij will decrease, thus reducing the probability of other ants selecting this edge.

[0033] Finally, in order to prevent falling into the local optimal solution, the pheromone constraint of the minimum and maximum ant system is introduced. At the end of the update, if τ ij Greater than τ max , then make it equal to τmax ; Similarly, if it is less than τ min , then τ ij Equal to τ min .

[0034] The heuristic information of the ant colony algorithm helps the algorithm converge to a high-quality solution region by imposing problem-specific knowledge. The "most remaining work heuristic" is introduced in the collection scheduling problem, which gives priority to the stage with more jobs to be processed, as shown in formula (4):

[0035]

[0036] in:

[0037] η ij is the heuristic function, which represents the expected degree of ant transfer from task i to task j;

[0038] is the number of personnel who have not yet been collected in stage s represented by task i;

[0039] n is the total number of people.

[0040] In S5, the global pheromone is updated as follows:

[0041]

[0042] in:

[0043] S upd is the set of all optimal solutions,

[0044] u soln is the decision variable,

[0045] If c ij If it is a path from task i to task j in the optimal solution, it is 1, otherwise it is 0.

[0046] soln is the optimal solution for this iteration, c ij is the path corresponding to the optimal ant from task i to task j. User interaction interface of multi-personnel collection and scheduling system based on improved genetic algorithm

[0047] like Figures 2 to 5 As shown, it is a schematic diagram of the user interface of a multi-personnel data collection and scheduling system based on an improved genetic algorithm in a specific example of an embodiment of the present invention.

[0048] After entering the team collaboration status assessment system based on support vector machine, the initial interface is as follows: Figure 2 The system interface consists of three parts: acquisition variable input, algorithm parameter input, and acquisition plan output.

[0049] Before the experiment starts, click on the collection variable input module and enter parameters such as "Number of collection projects", "Number of machines included in each project" and "Collection time required for each project". The input of collected personnel is divided into two parts, one based on individuals and the other based on teams. When choosing based on individuals, you need to enter the "Total number of people collected". When choosing based on teams, you need to fill in the "Number of people in the team" parameter, such as Figure 3 As shown. After completing the input of the collected variables, click Next to enter the algorithm parameter input module. You need to select the two solution encoding methods based on operation or project, and then enter the algorithm parameters such as "ant colony size", "total number of iterations", "pheromone factor", and "inspiration factor". After the input is completed, you can click "Start calculation". Figure 4 shown.

[0050] After the calculation is completed, you can enter the acquisition plan output module, which has two output methods, one is "Gantt chart output", after clicking, the system will automatically save it as a picture for reference, and the other is "text output", after clicking, the system will generate a txt file. After selecting the output method, the optimal acquisition plan time will be displayed. Figure 5 shown.

Claims

1. A multi-personnel collection scheduling method based on an improved ant colony algorithm, characterized in that Among them for There are n collected personnel and s collection stages. The n×s collection tasks are formed into a one-dimensional list and numbered from 1 to n×s. According to the task transfer rule of the ant colony algorithm, one task is selected from the list each time until the ant colony traverses all elements in the list, which is a feasible solution. Then, the ant colony system parameters are updated through algorithm iteration until the algorithm converges. At this time, the path taken by the ants is the optimal solution, which includes the following steps: S1: Initialize pheromones, heuristics, global pheromone volatility factors, local pheromone volatility factors and the number of random solutions; S2: Generate m ants, initialize a random task as the starting point of the ants, and for each ant, apply the ant colony system path selection function to select the next task until n×s tasks are selected. After each selection, update the local pheromone and heuristic. S3: construct m random solutions and expand the number of solutions to 2m; S4: Deserialize all solutions, that is, convert implicit phases into phase encodings; decode all solutions and select all solutions with the minimum collection completion time; S5: Determine whether the solution is smaller than the previous optimal solution. If so, perform pheromone update according to the global pheromone update method and output the current optimal solution. S6: Determine whether the minimum acquisition time for the reconnection is equal to the lower limit of time. If so, directly output the optimal solution, otherwise continue to iterate.

2. The multi-personnel data collection scheduling method according to claim 1, characterized in that: In S2, in order to avoid the algorithm from stagnating prematurely, a selection strategy combining deterministic selection and random selection is adopted, and the state transition probability is dynamically adjusted during the search process, including selecting the next task for ant k at task i according to formula (1): in: τ is is the pheromone concentration between task i and task s, is the pheromone factor, which indicates the relative importance of the amount of information accumulated during the ant movement in the ant colony search. η is is the heuristic function, β is the heuristic function factor, which reflects the relative importance of heuristic information in ant colony search. q0 is a random number in the interval [0,1], which is an algorithm parameter (0≤q0≤1). When q>q0, ant k determines the target task j to which task i is transferred downward according to formula (2). allowed k is the set of tasks to be visited by the kth ant. At the initial moment, there are n×s-1 tasks, which means that the tasks other than the tasks that the ant is in at the beginning are excluded. As time goes by, the number of tasks becomes smaller and smaller until it is empty, indicating that all tasks have been traversed; In the process of constructing the solution, every time the ant passes an edge (i, j), it will immediately call the local update rule to locally update the pheromone on the edge. The local update rule is shown in formula (3): t ij =(1-ρ)τ ij +ξτ0 (3), in: ξ and τ0 are two parameters, ξ satisfies 0<ξ<1; τ0 is the initial value of the pheromone amount, ρ is the pheromone volatility factor, reflecting the disappearance level of pheromone, 0<ρ<1, The role of local update is that every time the ant passes through the edge (i, j), the pheromone τ of the edge ij will decrease, thus reducing the probability of other ants selecting this edge. In order to prevent falling into the local optimal solution, the pheromone constraint of the minimum and maximum ant system is introduced. At the end of the update, if τ ij Greater than τ max , then make it equal to τ max ; Similarly, if it is less than τ min , then τ ij Equal to τ min , The "most remaining work heuristic" is introduced into the collection scheduling problem, as shown in formula (4): in: η ij is the heuristic function, which represents the expected degree of ant transfer from task i to task j; is the number of personnel who have not yet been collected in stage s represented by task i; n is the total number of people.

3. The multi-personnel data collection scheduling method according to claim 1 or 2, characterized in that: In S5, the global pheromone is updated as follows: in: S upd is the set of all optimal solutions, u soln is the decision variable, If c ij If it is a path from task i to task j in the optimal solution, it is 1, otherwise it is 0. soln is the optimal solution for this iteration, c ij is the path corresponding to the optimal ant from task i to task j.

4. A computer-readable storage medium storing a computer-executable program, wherein the computer-executable program enables a processor to execute the method according to any one of claims 1 to 3.

Images