A software multi-project scheduling platform based on a multi-population artificial bee colony algorithm
By optimizing multi-project scheduling in software using a multi-swarm artificial bee colony algorithm, combined with matrix management and online skills training, the problem of unreasonable resource allocation was solved, resulting in improved task matching rate and resource utilization, as well as increased employee skill proficiency.
Patent Information
- Application Number
- CN202311480535.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-08
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-11-08
AI Technical Summary
Existing technologies fail to effectively allocate resources in multi-project scheduling in software, resulting in task stagnation, extended project durations, and poor employee experience, and do not consider employee skills training to improve resource utilization.
We employ a multi-swarm artificial bee colony algorithm, combined with a matrix management model and online employee skills training. We optimize employee-task matching through the MCABC algorithm and improve resource utilization and employee skill proficiency through a multi-swarm co-evolutionary algorithm.
By rationally allocating tasks in a dynamic environment, improving the employee-task matching rate and human resource utilization, and enhancing employee skill proficiency, the algorithm achieves significantly better solution accuracy than existing algorithms.
Smart Images

Figure CN117391387B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of optimization algorithms and software engineering, and in particular to a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm. Background Technology
[0002] Software multi-project scheduling refers to the process of simultaneously managing and scheduling multiple software projects within an organization. These projects may share limited resources and may have interdependent and conflicting relationships. For organizations, effectively allocating and utilizing these resources to ensure the smooth progress of each project is crucial. However, in actual software project development, it is common for software development departments to handle multiple projects simultaneously. Human resources are limited, and when multiple projects are developed in parallel, resource competition can easily arise. Inappropriate allocation can lead to a chain reaction of tasks stalling, extended deadlines, and poor employee experience. Currently, there is a limited amount of research on software multi-project scheduling, but none of it has addressed the division of employee roles and the systematic allocation of tasks to suitable employees. Furthermore, it has not considered the role of employee learning to improve their skills. Therefore, in-depth research on software multi-project scheduling under a matrix management model and employee skills training mechanisms is essential.
[0003] Artificial bee colony optimization (AQC) is an optimization method that mimics the behavior of bees. It's a concrete application of swarm intelligence, characterized by its ability to converge quickly without requiring specific information about the problem. Instead, it compares the merits of different problems and, through the local optimization behavior of individual artificial bees, ultimately leads to the emergence of the global optimum within the colony. AQC is a flexible and efficient optimization algorithm now applied in optimization problems, data mining and machine learning, image processing and machine vision, network optimization, and communication. In recent years, search-based software engineering has rapidly developed, with more and more scholars both domestically and internationally researching how to model software project scheduling problems as search-based optimization problems. This involves using metaheuristic algorithms to search for optimal allocation schemes, thereby assisting software project managers in making final decisions. Software multi-project scheduling problems require scheduling multiple projects, involving a large number of employees and a large total number of tasks, resulting in a large dimensionality of the decision space. Co-evolutionary algorithms, based on problem decomposition, break down complex problems into multiple simpler subproblems, using multiple subpopulations for optimization and collaborative optimization to improve the model's solution efficiency. Therefore, this invention designs a multi-swarm artificial bee colony algorithm for the software multi-project scheduling model under the proposed matrix management model and employee skills training mechanism, which improves the matching rate of employees and tasks and the utilization rate of human resources. Summary of the Invention
[0004] The purpose of this invention is to enable the rational allocation of tasks in a dynamically changing environment, thereby improving the employee-task matching rate and the utilization rate of human resources.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] Firstly, a software-based multi-project scheduling platform based on a multi-population artificial bee colony algorithm is provided, including the following steps:
[0007] The receiving unit is used to receive project information and employee information. The project information includes the workload, skill requirements, skill difficulty, and task priority map for each task. The employee information includes the employee's existing skills, proficiency in existing skills, learning ability, maximum workload that can be contributed, normal monthly salary, and overtime monthly salary. The unit also initializes the project task information and employee information.
[0008] The processing unit is used to determine the matrix management mode and the online skills training mode for employees by combining the initialized employee information, wherein the matrix management mode and the online skills training mode for employees are used to establish a multi-project software project scheduling model.
[0009] The solution unit is used to solve the multi-project software project scheduling model using the MCABC algorithm and determine the optimal scheduling scheme.
[0010] The display unit is used to show the scheduling status of multiple software projects, statistical analysis of data streams running over a specific time period, and to provide each user with a platform operation entry point. Each user also has an independent personal space that only displays their own content, including recently run models, data, and running records.
[0011] Based on the above technical solutions, the present invention also provides the following optional technical solutions:
[0012] Further technical solution: The multi-project software scheduling of the multi-project software project scheduling model in the processing unit is specifically as follows:
[0013] Suppose that at time t there exists a multi-item set Q = {Q 1 Q 2 ,...,Q q ...,Q R}, where Q q Let q represent the q-th item in a set of multiple items, where q = 1, 2, ..., R, and R represents the number of items; each item in, Project Q q The j-th task, N q Project Q q The total number of tasks; the multi-project set Q has a total of S skill requirements. express The required skill set, including Taking 1 or 0 indicates Skill k required or not required; Employee E i (i = 1, 2, ..., M) at time t, the task... The level of proficiency is expressed as and in, It is E i Proficiency with skill k Taking 0 or C respectively indicates that Ei does not possess skill k at time t, or has fully mastered skill k. i (t) is the set of skills that Ei possesses at time t, and M represents the total number of employees.
[0014] Further technical solutions: The matrix management model and employee online skills training model specifically include:
[0015] The matrix management model and the online employee skills training model specifically include:
[0016] The matrix management model involves dividing all employees into functional groups based on their skills, and selecting suitable employees from the corresponding functional groups to perform project-related tasks based on project requirements. Each employee can possess multiple skills and can be assigned to multiple functional groups simultaneously. Each employee can participate in multiple tasks and projects concurrently, but the number of different projects an employee participates in at the same time cannot exceed two. S functional groups are formed based on the number of skills, and employees with different skills are assigned to the corresponding functional groups. fug k (t) represents the kth functional group at time t;
[0017] The online skills training model for employees is as follows: each employee can register for multiple online skills training courses at the start of the project and arrange the order of the courses themselves. Employees can also register for training courses corresponding to skills they do not currently possess or whose proficiency has not reached the threshold α·C (α=0.8). When the proficiency of a skill reaches the threshold α·C or the training course ends, the next skills training course can begin. The employee skills proficiency growth trend includes:
[0018] Training skill proficiency growth: Assuming employee E i (i = 1, 2, ..., M) registered for skill z (z ∈ traS) i Training courses, E represents i The skills set for the training program I've enrolled in, and the learning order I've already planned out, 0 or 0 respectively represent employee E iRegistered or unregistered training courses for skill k; using the hyperbolic tangent function tanh(·) to fit the proficiency growth trend of the trained skill, i.e.:
[0019]
[0020] Among them, t z The time taken to learn skill z is represented by the formula t. z =(t-t_l z )·tra i _time, t_l z Indicates the start time of learning skill z, tra i _time represents E i The percentage of time each month available for online skills training courses, tra i _time∈(0,1),tra i _time = 0.2 indicates E i 0.2 x 30 days each month can be used for online skills training; Indicates employee E i The growth rate of proficiency for skill z Indicates employee E i Interest in learning skill z, employee learning ability The larger the value, the greater the interest in skill z. The larger the skill level, the greater the difficulty (SD). z The smaller the value, the higher the growth rate. The larger; Indicates employee E i The similarity of proficiency between skill k and new skill z, where t0 represents the project start time, and r kz r represents the degree of correlation between skill k and new skill z. kz ∈[0,1], r kz The larger the value, the more related the two skills are;
[0021] If at time t, E i If the training and assessment requirements for skill z are met ahead of schedule, then And t z <train k At that time, train k t indicates the duration of the training course for skill k; or that the training course for skill z has ended. z =train k When, all represent E i Learning of skill z is complete; at this point, traS i =traS i \z, and update tra_in at the same time.i (t), tra_in i (t) represents E i The skill being trained at time t, tra_in i (t)∈{traS i};
[0022] Increase in existing skill proficiency: Using the ln(·) function to fit the changing trend of employees' existing skills; Employee E i Existing skill proficiency of (i = 1, 2, ..., M) The growth function is as follows:
[0023]
[0024] in, Indicates employee E i The growth rate of proficiency for skill k, t k This indicates the time when skill k begins to improve slightly, which can be either the start of the project or the end of the training for skill k.
[0025] Further technical solution: The multi-project software project scheduling model includes decision variables, objective function, and constraints that the scheduling scheme needs to satisfy, as detailed below:
[0026] The decision variable is the engagement matrix X of all employees on all projects in the multi-project set Q, which is defined as follows:
[0027]
[0028] in, This indicates that all employees are supportive of Project Q. q The allocation matrix, Indicates employee E i For project Q q Medium task The level of commitment, when Indicates employee E i Q, who did not participate in the project q Medium task This indicates that they participated in the task; E represents i Maximum commitment to the project E represents i You can dedicate all your working hours each month to the project. E represents i I can accept working overtime. E represents i They may have been assigned other work outside of this project or have part-time jobs;
[0029] The objective function is to minimize the duration and cost of all projects, expressed as:
[0030]
[0031] in, and These are weighting parameters used to adjust Q. q The relative importance of construction period and cost; when When the time and cost are on the same order of magnitude, the project is considered to be more likely to be completed in a shorter time than the cost after weighting. Conversely, it is considered to be more likely to be completed in a lower cost than the cost.
[0032] Item Q in the objective function q Construction period q , is represented as:
[0033]
[0034] in, and Representing task T respectively j The start and end times (in months), and
[0035] Item Q in the objective function q cost q , is represented as:
[0036]
[0037] in, For period t′, pay employee E i Salary is expressed as:
[0038]
[0039] Where t′ represents the time unit in the project development process, E represents i Normal monthly salary E represents i Monthly overtime pay, T_active_set(t′) represents the set of tasks being developed during period t′; if the employee's total investment in this task set during t′ is... This indicates that the employee worked overtime and should be paid overtime wages;
[0040] The constraints that the scheduling scheme needs to satisfy include:
[0041] The set of employee skills assigned to a task to be performed in a project must include the skills required for that task, that is:
[0042]
[0043] Among them, skill i (t) represents E i The set of skills possessed
[0044] At any time t, the total engagement of each employee with the current set of executable tasks must not exceed that employee's maximum engagement, that is:
[0045]
[0046] Where T_active_set(t) represents the set of currently executable tasks;
[0047] The number of participants in a task cannot exceed the maximum number of participants allowed for that task, nor can it be less than the minimum number of employees required to meet the skill constraints of that task.
[0048]
[0049] in, Indicates participation in the task Number of employees; and These represent the maximum number of people who can participate in the task and the minimum number of employees required to meet the skill constraints of the task, respectively.
[0050] At any time t, each employee can participate in no more than 2 projects, that is:
[0051] |E_Project i (t)|≤2,i=1,2,...,M
[0052] in, E represents i The set of projects participated in at time t It is a binary variable. Taking 1 or 0 respectively represents E i Q: Whether or not someone participated in the project at time t q .
[0053] Further technical solution: The steps for solving the multi-project software project scheduling model using the MCABC algorithm are as follows:
[0054] S401, an initial population is generated using a population initialization strategy based on functional groups;
[0055] S402, grouping nectar source populations based on sub-target ranking values;
[0056] S403 guides bees to conduct global searches using subpopulation interaction mechanisms, while follower bees conduct local searches of individuals based on heuristic information;
[0057] S404, the scout bee discards useless nectar sources and randomly generates new nectar sources;
[0058] S405, if the number of evaluations of the target reaches the maximum value, output the global optimal value; otherwise, go to step S402.
[0059] A further technical solution: The initial population obtained by the population initialization strategy based on functional groups consists of 50% heuristically constructed solutions and 50% random solutions;
[0060] Employees are assigned to S functional groups based on S skills. k (t), k=1,2,...,S; Therefore, at the initial time t0 of the project, selecting employees from the corresponding functional groups to join the task according to the skill requirements of the project task can ensure that the obtained initial solution is a feasible solution, thereby increasing the number of initial feasible solutions;
[0061] The random solution is as follows: each employee's level of engagement with all project tasks is randomly generated within the feasible region. When the number of participants in a task exceeds the maximum allowed number, the participating employees are reduced. When an employee is overworked, the employee's level of engagement is normalized.
[0062] The heuristic solution is constructed as follows: for the task j = 1, 2, ..., N q q = 1, 2, ..., R, based on its required skill set The skills requirements are derived from the relevant functional groups. Selecting employees to join Find and own The employee with the highest required skills; if multiple employees exist, then the one with the most required skills will be selected. proficiency The highest employee E i And randomly generate the level of investment within its investment range. calculate To address the missing skills, repeat the above steps and select additional staff from the remaining employees in the corresponding functional groups. Until the task skill requirements are met; make a judgment Does the number of participants exceed If the maximum number of participants is not reached, members will be selected from the remaining members of the functional team required for the task based on their proficiency in the task, using a roulette wheel strategy, until the maximum number of participants for the task is reached; when employees are overworked, their level of engagement will be normalized.
[0063] Further technical solution: The steps for grouping nectar source populations based on sub-target ranking values are as follows:
[0064] Before grouping, there are SN honey source individuals. For each honey source individual, in project Q... q The objective function value f p (Q q Summing the sub-target values, we obtain the target value F(X) for each individual nectar source. p );
[0065] For each project, pair all SN honey source individuals with project Q. q The sub-objective values are sorted from smallest to largest (minimizing the objective) to obtain the value of each nectar source individual in project Q. q sort value sortf p (Q q For each individual ∈ [1,2,...,SN], there exists a set o_set consisting of R sorted values. p ={sortf p (Q 1 ),sortf p (Q 2 ),...,sortf p (Q q ),...,sortf p (Q R The smaller the ranking value, the better the target value of the corresponding project for that individual.
[0066] The nectar source population is divided into R subpopulations based on the number of projects, and then individuals are gradually added to the subpopulations. For each individual, the index of the minimum value among its R sorted values is taken. If multiple minimum value indices exist simultaneously, one is randomly selected.
[0067] Individuals are placed into subpopulations until all individuals have been traversed.
[0068] Further technical solution: The mechanism by which the leading bee performs a global search using subpopulation interaction includes:
[0069] When individuals in a subpopulation interact with information from other subpopulations, based on probability... Determine if interactive information is needed; for each nectar source individual in the subpopulation, generate a random number between 0 and 1 for each subpopulation. If the random number is not greater than... This indicates that information needs to be exchanged with the subpopulation; individuals in the subpopulation are sorted in ascending order according to the target value of the project, and the sorting number is recorded. An individual is selected from the subpopulation as its interaction object through a sorting selection strategy; when the information exchange with other subpopulations ends, a greedy selection is performed based on the individual target values before and after the update, leaving the individual with the better target value.
[0070] Secondly, a computing device for a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm is provided, comprising:
[0071] One or more processors, memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the platforms described.
[0072] Thirdly, a computer-readable storage medium is provided for storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any of the platforms.
[0073] Beneficial effects
[0074] (1) The present invention provides a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm, which can reasonably allocate tasks in a dynamically changing environment, improve the matching rate of employees and tasks and the utilization rate of human resources.
[0075] (2) The present invention provides a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm, which allows each employee to freely choose skills for training, improve skill proficiency, and enhance their own value.
[0076] (3) The present invention provides a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm. The proposed algorithm has significantly better solution accuracy than representative existing algorithms in the vast majority of cases. Attached Figure Description
[0077] Figure 1 This is a block diagram of the software multi-project scheduling platform based on the multi-population artificial bee colony algorithm of the present invention;
[0078] Figure 2 This is a comparison chart of Best values in an embodiment of the present invention;
[0079] Figure 3 This is a comparison chart of Avg values in an embodiment of the present invention;
[0080] Figure 4 This is a convergence curve of the Best value as a function of the number of target evaluations in an embodiment of the present invention. Detailed Implementation
[0081] The technical solutions in the embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0082] Example
[0083] Please see Figures 1-4 A multi-project scheduling platform based on a multi-population artificial bee colony algorithm, including:
[0084] The receiving unit is used to receive project information and employee information. The project information includes the workload, skill requirements, skill difficulty, and task priority map for each task. The employee information includes the employee's existing skills, proficiency level, learning ability, maximum workload that can be contributed, normal monthly salary, and overtime monthly salary. The unit also initializes the project task information and employee information; among which, learning ability... The larger the value, the greater the value of E. i The stronger one's learning ability; E i The highest level of commitment to a project is expressed as E represents i You can dedicate all your working hours each month to the project; E represents i I can accept working overtime, which also means E i They may have been assigned other work or part-time jobs outside of this project;
[0085] The processing unit is used to determine the matrix management mode and the online skills training mode for employees by combining the initialized employee information, wherein the matrix management mode and the online skills training mode for employees are used to establish a multi-project software project scheduling model.
[0086] The solution unit is used to solve the multi-project software project scheduling model using the MCABC algorithm and determine the optimal scheduling scheme.
[0087] The display unit is used to show the scheduling status of multiple software projects, statistical analysis of data streams running over a specific time period, and to provide each user with a platform operation entry point. Each user also has an independent personal space that only displays their own content, including recently run models, data, and running records.
[0088] Specifically, the multi-project software scheduling in the multi-project software project scheduling model of the processing unit includes:
[0089] Suppose that at time t there exists a set of multiple items Q = {Q...}1 Q 2 ,...,Q q ...,Q R}, where Q q Let q represent the q-th item in a set of multiple items, where q = 1, 2, ..., R, and R represents the number of items; each item in, Project Q q The j-th task, N q Project Q q The total number of tasks; the multi-project set Q has a total of S skill requirements. express The required skill set, including Taking 1 or 0 indicates Skill k required or not required; Employee E i (i = 1, 2, ..., M) at time t, the task... The level of proficiency is expressed as and in, It is E i Proficiency with skill k Taking 0 or C respectively represents E i In the case where t lacks skill k or has fully mastered k, skill i (t) is E i The set of skills possessed at time t, where M represents the total number of employees in the project.
[0090] Specifically, the matrix management model and the online employee skills training model include:
[0091] The matrix management model involves dividing all employees into functional groups based on their skills, and selecting suitable employees from the corresponding functional groups to perform project-related tasks based on project requirements. Each employee can possess multiple skills and can be assigned to multiple functional groups simultaneously. Each employee can participate in multiple tasks and projects concurrently, but the number of different projects an employee participates in at the same time cannot exceed two. S functional groups are formed based on the number of skills, and employees with different skills are assigned to the corresponding functional groups. fug k (t) represents the kth functional group at time t;
[0092] The online skills training model for employees is as follows: each employee can register for multiple online skills training courses at the start of the project and arrange the order of the courses themselves. Employees can also register for training courses corresponding to skills they do not currently possess or whose proficiency has not reached the threshold α·C (α=0.8). When the proficiency of a skill reaches the threshold α·C or the training course ends, the next skills training course can begin. The employee skills proficiency growth trend includes:
[0093] Training skill proficiency growth: Assuming employee E i (i = 1, 2, ..., M) registered for skill z (z ∈ traS) i Training courses, E represents i The skills set for the training program I've enrolled in, and the learning order I've already planned out, 0 or 0 respectively represent employee E i Registered or unregistered training courses for skill k; using the hyperbolic tangent function tanh(·) to fit the proficiency growth trend of the trained skill, i.e.:
[0094]
[0095] Among them, t z The time taken to learn skill z is represented by the formula t. z =(t-t_l z )·tra i _time, t_l z Indicates the start time of learning skill z, tra i _time represents E i The percentage of time each month available for online skills training courses, tra i _time∈(0,1),tra i _time = 0.2 indicates E i 0.2 x 30 days each month can be used for online skills training; Indicates employee E i The growth rate of proficiency for skill z Indicates employee E i Interest in learning skill z, employee learning ability The larger the value, the greater the interest in skill z. The larger the skill level, the greater the difficulty (SD). z The smaller the value, the higher the growth rate. The larger; Indicates employee E i The similarity of proficiency between skill k and new skill z, where t0 represents the project start time, and r kzr represents the degree of correlation between skill k and new skill z. kz ∈[0,1], r kz The larger the value, the more related the two skills are;
[0096] If at time t, E i If the training and assessment requirements for skill z are met ahead of schedule, then and At that time, train k t indicates the duration of the training course for skill k; or that the training course for skill z has ended. z =train k When, all represent E i Learning of skill z is complete; at this point, traS i =traS i \z, and update tra_in at the same time. i (t), tra_in i (t) represents E i The skill being trained at time t, tra_in i (t)∈{traS i};
[0097] Increase in existing skill proficiency: Using the ln(·) function to fit the changing trend of employees' existing skills; Employee E i Existing skill proficiency of (i = 1, 2, ..., M) The growth function is as follows:
[0098]
[0099] in, Indicates employee E i The growth rate of proficiency for skill k, t k This indicates the time when skill k begins to improve slightly, which can be either the start of the project or the end of the training for skill k.
[0100] Specifically, the multi-project software project scheduling model includes decision variables, objective functions, and constraints that the scheduling scheme needs to satisfy, as detailed below:
[0101] The decision variable is the engagement matrix X of all employees on all projects in the multi-project set Q, which is defined as follows:
[0102]
[0103] in, This indicates that all employees are supportive of Project Q. q The allocation matrix, Indicates employee E i For project Qq Medium task The level of commitment, when Indicates employee E i Q, who did not participate in the project q Medium task This indicates that they participated in the task; E represents i Maximum commitment to the project E represents i You can dedicate all your working hours each month to the project. E represents i I can accept working overtime. E represents i They may have been assigned other work outside of this project or have part-time jobs;
[0104] The objective function is to minimize the duration and cost of all projects, expressed as:
[0105]
[0106] in, and These are weighting parameters used to adjust Q. q The relative importance of construction period and cost; when When the time and cost are on the same order of magnitude, the project is considered to be more likely to be completed in a shorter time than the cost after weighting. Conversely, it is considered to be more likely to be completed in a lower cost than the cost.
[0107] Item Q in the objective function q Construction period q , is represented as:
[0108]
[0109] in, and Representing task T respectively j The start and end times (in months), and
[0110] Item Q in the objective function q cost q , is represented as:
[0111]
[0112] in, For period t′, pay employee E i Salary is expressed as:
[0113]
[0114] Where t′ represents the time unit in the project development process, E represents i Normal monthly salary E represents i Monthly overtime pay, T_active_set(t′) represents the set of tasks being developed during period t′; if the employee's total investment in this task set during t′ is... This indicates that the employee worked overtime and should be paid overtime wages;
[0115] The constraints that the scheduling scheme needs to satisfy include:
[0116] The set of employee skills assigned to a task to be performed in a project must include the skills required for that task, that is:
[0117]
[0118] Among them, skill i (t) represents E i The set of skills possessed
[0119] At any time t, the total engagement of each employee with the current set of executable tasks must not exceed that employee's maximum engagement, that is:
[0120]
[0121] Where T_active_set(t) represents the set of currently executable tasks;
[0122] The number of participants in a task cannot exceed the maximum number of participants allowed for that task, nor can it be less than the minimum number of employees required to meet the skill constraints of that task.
[0123]
[0124] in, Indicates participation in the task Number of employees; and These represent the maximum number of people who can participate in the task and the minimum number of employees required to meet the skill constraints of the task, respectively.
[0125] At any time t, each employee can participate in no more than 2 projects, that is:
[0126] |E_Project i (t)|≤2,i=1,2,...,M
[0127] in, E represents i The set of projects participated in at time t It is a binary variable. Taking 1 or 0 respectively represents E i Q: Whether or not someone participated in the project at time t q .
[0128] Specifically, the optimal scheduling scheme determined in step S3 based on the multi-population artificial bee colony algorithm and problem model is as follows:
[0129] S401, an initial population is generated using a population initialization strategy based on functional groups;
[0130] S402, grouping nectar source populations based on sub-target ranking values;
[0131] S403 guides bees to conduct global searches using subpopulation interaction mechanisms, while follower bees conduct local searches of individuals based on heuristic information;
[0132] S404, the scout bee discards useless nectar sources and randomly generates new nectar sources;
[0133] S405, if the number of evaluations of the target reaches the maximum value, output the global optimal value; otherwise, go to step S402.
[0134] Specifically, an individual is represented as:
[0135] Using integer encoding, for a software project scheduling problem involving M employees and N tasks, each individual is encoded as an M×N matrix Y = (y ij ) M×N ,i=1,2,...,M,j=1,2,...,N; where y ij ∈{0,1,2,...,h}, where h represents the granularity of the solution, i.e., employee E i Maximum investment Divided into h equal parts; element y ij The meaning is employee E i Give your maximum effort of y ij / h Dedicated to Mission T j .
[0136] Specifically, to satisfy its constraints, feasible individuals are generated after decoding. The decoding process is as follows:
[0137] Through each task T j Occupying each employee's E i Maximum contribution The proportion y ij / h, transform the individual matrix into an employee-task allocation matrix X = (x ij ) M×N The specific conversion method is as follows:
[0138]
[0139] in,
[0140] Specifically, the initial population obtained by the functional group-based population initialization strategy consists of 50% heuristically constructed solutions and 50% random solutions;
[0141] Employees are assigned to S functional groups based on S skills. k (t), k=1,2,...,S; Therefore, at the initial time t0 of the project, selecting employees from the corresponding functional groups to join the task according to the skill requirements of the project task can ensure that the obtained initial solution is a feasible solution, thereby increasing the number of initial feasible solutions;
[0142] The random solution is as follows: each employee's level of engagement with all project tasks is randomly generated within the feasible region. When the number of participants in a task exceeds the maximum allowed number, the participating employees are reduced. When an employee is overworked, the employee's level of engagement is normalized.
[0143] The heuristic solution is constructed as follows: for the task j = 1, 2, ..., N q q = 1, 2, ..., R, based on its required skill set The skills requirements are derived from the relevant functional groups. Selecting employees to join Find and own The employee with the highest required skills; if multiple employees exist, then the one with the most required skills will be selected. proficiency The highest employee E i And randomly generate the level of investment within its investment range. calculate To address the missing skills, repeat the above steps and select additional staff from the remaining employees in the corresponding functional groups. Until the task skill requirements are met; make a judgment Does the number of participants exceed If the maximum number of participants is not reached, members will be selected from the remaining members of the functional team required for the task based on their proficiency in the task, using a roulette wheel strategy, until the maximum number of participants for the task is reached; when employees are overworked, their level of engagement will be normalized.
[0144] Specifically, the population grouping method based on sub-target ranking values is implemented through the following steps:
[0145] Before grouping, there are SN honey source individuals. For each honey source individual, in project Q... q The objective function value f p (Q q Summing the sub-target values, we obtain the target value F(X) for each individual nectar source. p );
[0146] For each project, pair all SN honey source individuals with project Q. q The sub-objective values are sorted from smallest to largest (minimizing the objective) to obtain the value of each nectar source individual in project Q. q sort value sortf p (Q q For each individual ∈ [1,2,...,SN], there exists a set o_set consisting of R sorted values. p ={sortf p (Q 1 ),sortf p (Q 2 ),...,sortf p (Q q ),...,sortf p (Q R The smaller the ranking value, the better the target value of the corresponding project for that individual.
[0147] The nectar source population is divided into R subpopulations based on the number of projects, and then individuals are gradually added to the subpopulations. For each individual, the index of the minimum value among its R sorted values is taken. If multiple minimum value indices exist simultaneously, one is randomly selected.
[0148] Individuals are placed into subpopulations until all individuals have been traversed.
[0149] Specifically, the process of selecting individuals for updating based on sequence number using a sorting and selection strategy involves the following steps:
[0150] Individuals in each subpopulation are sorted according to the merits of the target value and assigned a sorting number. Follower bees then use a sorting selection strategy based on the sorting number to select individuals for updating.
[0151] Considering the preferences of different projects, three different heuristics are used to assist the algorithm in the search. The three heuristics are as follows:
[0152] For projects that are time-sensitive: the more proficient the employees involved in a task are, the faster the task will be completed. These employees have a greater impact on the completion time of the corresponding task, thus indirectly affecting the project's completion time. Therefore, when optimizing a task, searching for the input values of employees with a high probability of proficiency in that task can increase the likelihood of improving the scheduling scheme.
[0153] For cost-oriented projects: the cost calculation is the labor cost of employees, i.e. wages; when the project has low time requirements and is biased towards low cost, the input of low-wage employees in the project will have a significant impact on the project's target value; therefore, when searching for scheduling solutions for cost-oriented projects, increasing the search probability of low-cost employee input can increase the possibility of improving the scheduling solution.
[0154] For projects where time and cost are equally important: When a project considers time and cost to be equally important, the skill level and wages of the employees involved in the task should be considered simultaneously. The higher the skill level of the employees involved in the task, the shorter the task completion time, the lower the employee wages, and the lower the task cost. Therefore, employees with high skill level and low wages are more suitable for the task and have a greater impact on the quality of task completion.
[0155] Specifically, the mechanism by which the leading bee uses subpopulation interaction to perform a global search includes:
[0156] When individuals in a subpopulation interact with information from other subpopulations, based on probability... Determine if interactive information is needed; for each nectar source individual in the subpopulation, generate a random number between 0 and 1 for each subpopulation. If the random number is not greater than... This indicates that information needs to be exchanged with the subpopulation; individuals in the subpopulation are sorted in ascending order according to the target value of the project, and the sorting number is recorded. An individual is selected from the subpopulation as its interaction object through a sorting selection strategy; when the information exchange with other subpopulations ends, a greedy selection is performed based on the individual target values before and after the update, leaving the individual with the better target value.
[0157] Specifically, a computing device for a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm is provided, comprising:
[0158] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described.
[0159] Following the conventional approach in existing ABC algorithm literature, the proposed multi-population artificial bee colony algorithm MCABC sets the nectar source population size SN and the maximum update failure count limit to 100 and 200, respectively. This paper uses 10 software multi-project scheduling model test cases, including 9 software multi-project cases generated by the case generator and 1 real-world commercial software project combination instance. The 9 artificially synthesized cases are named P#1_E#2_T#3, where P#1 represents the number of projects (selected from 2, 3, and 4), E#2 represents the number of project employees (selected from 5, 10, and 15), and T#2 represents the number of project tasks (selected from 5, 10, and 15). The total number of skills S required for the project is randomly selected from 5 to 10. The training course duration for skill k is train. k It is randomly and uniformly generated within the range [1,3], E i The percentage of non-working time available each month for online skills training courses. i `_time` is uniformly and randomly generated between (0,1). Priority relationships between multiple projects are randomly generated. Three existing software project instances are grouped into a single multi-project instance `Real_group`, with the priority relationships between the three projects randomly generated.
[0160] To verify the performance of the MCABC algorithm in solving a multi-project scheduling model for software, it was compared with five representative existing algorithms: (1+1)EA, GA, IDABC, ABCMSSCE, and MPPCEDE. The encoding method of GA was replaced with the discrete encoding method of (1+1)EA, while the rest remained the same as in the original literature. (1+1)EA and GA used the same individual encoding and decoding, and the other comparison algorithms used the same individual encoding and decoding method as the proposed algorithm. The parameter values of the comparison algorithms were the same as in the original literature. All comparison algorithms used the same methods for target value calculation and constraint handling, while the rest of the algorithms were the same as in the original literature. These algorithms were then applied to the model established in this invention and compared with the proposed algorithm. The termination condition for all algorithms was set to the maximum number of target evaluations reaching 50,000. Experimental results are attached. Figures 2-3 As shown.
[0161] From the appendix Figures 2-3 As can be seen, compared with the other five algorithms, the proposed MCABC algorithm achieved the best average value (Avg) and optimal value (Best) on most multi-project software tests of varying scales. To facilitate observation of the convergence of all algorithms, taking the Real_group example, the convergence curves of the optimal objective value (Best) of each algorithm in a single run as a function of the number of objective evaluations are plotted, as shown in the attached figure. Figure 4As shown in the curve, the use of a functional group-based population initialization strategy gives MCABC a better starting point for the search. Compared to the five comparison algorithms, MCABC achieved the best value in the early, middle, and late stages of the algorithm, demonstrating better evolution speed and search accuracy.
[0162] This invention proposes a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm. This platform can rationally allocate tasks in dynamically changing environments, improving the employee-task matching rate and the utilization rate of human resources. Simultaneously, each employee can freely choose skills for training, enhancing their proficiency and increasing their value. Furthermore, comparisons between the proposed algorithm and five representative existing algorithms show that the proposed algorithm's solution accuracy is significantly superior to the other compared algorithms.
[0163] Embodiments of this application may be provided as methods or computer program products. Therefore, this application may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application may be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.
[0164] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0165] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0166] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0167] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
Claims
1. A software multi-project scheduling platform based on a multi-population artificial bee colony algorithm, characterized in that, include: The receiving unit is used to receive project information and employee information. The project information includes the workload, skill requirements, skill difficulty and task priority map of each task. The employee information includes the employee's existing skills, proficiency in existing skills, learning ability, maximum workload that can be devoted, normal monthly salary and overtime monthly salary. And initialize project task information and employee information; The processing unit is used to determine the matrix management mode and the online skills training mode for employees by combining the initialized employee information, wherein the matrix management mode and the online skills training mode for employees are used to establish a multi-project software project scheduling model. The matrix management model involves dividing all employees into functional groups based on their skills, and selecting suitable employees from the corresponding functional groups to perform project-related tasks based on project requirements. Each employee can possess multiple skills and can be assigned to multiple functional groups simultaneously. Each employee can participate in multiple tasks and projects at the same time, but the number of different projects an employee participates in simultaneously cannot exceed two. The division is based on the number of skills. S Employees with different skills are assigned to corresponding functional groups. , express t The first moment k One functional group; The solution unit is used to solve the multi-project software project scheduling model using the MCABC algorithm and determine the optimal scheduling scheme. The steps for solving the multi-project software project scheduling model using the MCABC algorithm are as follows: S401, an initial population is generated using a population initialization strategy based on functional groups; S402, grouping nectar source populations based on sub-target ranking values; S403 guides bees to conduct global searches using subpopulation interaction mechanisms, while follower bees conduct local searches of individuals based on heuristic information; S404, the scout bee discards useless nectar sources and randomly generates new nectar sources; S405, if the number of times the target is evaluated reaches the maximum value, output the global optimal value; otherwise, go to step S402. The initial population obtained by the population initialization strategy based on functional groups consists of 50% heuristically constructed solutions and 50% random solutions. according to S Each skill will be assigned to employees. S functional groups , Therefore, at the initial stage of the project By selecting employees from the corresponding functional groups to join the project task based on the skill requirements of the project task, it can be ensured that the initial solution obtained is a feasible solution, thereby increasing the number of initial feasible solutions. The random solution is as follows: each employee's level of engagement with all project tasks is randomly generated within the feasible region. When the number of participants in a task exceeds the maximum allowed number, the participating employees are reduced. When an employee is overworked, the employee's level of engagement is normalized. The heuristic solution is constructed as follows: for the task... , , Based on its required skill set The skills requirements are derived from the relevant functional groups. Selecting employees to join Find and own The employee with the highest required skills; if multiple employees exist, then the one with the most required skills will be selected. proficiency Top employee And randomly generate the level of investment within its investment range. ;calculate To address the missing skills, repeat the above steps and select additional staff from the remaining employees in the corresponding functional groups. Until the task skill requirements are met; determine Does the number of participants exceed If the maximum number of participants is not exceeded, members will be selected from the remaining members of the functional team required for the task based on their proficiency in the task, using a roulette wheel strategy, until the maximum number of participants for the task is reached; when employees are overworked, their level of engagement will be normalized. The display unit is used to show the scheduling status of multiple software projects, statistical analysis of data streams running over a specific time period, and to provide each user with a platform operation entry point. Each user also has an independent personal space that only displays their own content, including recently run models, data, and running records.
2. The software multi-project scheduling platform based on a multi-population artificial bee colony algorithm according to claim 1, characterized in that... The multi-project software scheduling model in the processing unit specifically includes: Assuming in t Multiple project sets always exist ,in, Represents the first item in a multi-item set. q One project, , R Indicates the number of items; each item ,in, Indicates project The j One task, Indicates project Total number of tasks; multi-project collection Q Total S Each skill requirement express The required skill set, including Taking 1 or 0 indicates Skills required or not required k ;staff ( i =1,2,… M )exist t Always on the task The level of proficiency is expressed as ,and ,in, , yes skills k proficiency Taking 0 or C respectively represents exist Lacking skills k Or have already fully mastered k , yes exist t The set of skills one possesses at any given moment M This indicates the total number of employees.
3. The software multi-project scheduling platform based on a multi-population artificial bee colony algorithm according to claim 2, characterized in that, The matrix management model and the online employee skills training model specifically include: The aforementioned online skills training model allows each employee to register for multiple online skills training courses at the start of the project and arrange the order of these courses themselves. Employees can also register for skills they do not currently possess or whose proficiency level has not reached the required threshold. Training courses corresponding to the skills, When the proficiency of a certain skill reaches a threshold Alternatively, the next skills training course can begin at the end of the current training course; trends in employee skill proficiency include: Training skill proficiency growth: Assuming employees I enrolled in a skills training course, including , , express The skills set for the training program I've enrolled in, and the learning order I've already planned, 0 or 0 respectively represent employees Skills registered or not k Training courses; utilizing the hyperbolic tangent function To fit the growth trend of training skill proficiency, that is: in, Representation skills z The formula for calculating study time is: , Indicates learning skills z The start time, express The percentage of time available each month for online skills training courses. , express 0.2 x 30 days each month can be used for online skills training; Indicates employees skills z The growth rate of proficiency Indicates employees Learning skills z Interest level, employee learning ability The larger the skill level, the greater the skill level. z interest The larger the size, the greater the skill difficulty. The smaller the value, the higher the growth rate. The larger; , , Indicates employees skills k With new skills z The similarity in proficiency between them Indicates the project start time. For skills k With new skills z The degree of correlation between them , The larger the value, the more related the two skills are; If in t time, Achieve skills ahead of time z The training and assessment requirements are as follows: and hour, Representation skills k Training course duration; or skills z The training course has ended. At times, all represent skills z The learning has ended; at this time, Simultaneously update , express exist t The skills that are being trained at all times ; Increased proficiency in existing skills: Utilizing The function is used to fit the changing trends of employees' existing skills; employees Existing skill proficiency The growth function is as follows: in, Indicates employees skills The growth rate of proficiency Representation skills k The time when you begin to see small improvements is either at the start of a project or in a skill. k The training ended.
4. The software multi-project scheduling platform based on a multi-population artificial bee colony algorithm according to claim 1, characterized in that, The multi-project software project scheduling model includes decision variables, objective functions, and constraints that the scheduling scheme must satisfy, as detailed below: The decision variable is the set of multiple items for all employees. Q Input allocation matrix for all projects X Its definition is: in, This indicates that all employees are committed to the project. The allocation matrix, Indicates employees For the project Medium task The level of commitment, when , indicating employees Not involved in the project Medium task , This indicates that they participated in the task; express Maximum commitment to the project express You can dedicate all your working hours each month to the project. express I can accept working overtime. express They may have been assigned other work outside of this project or have part-time jobs; The objective function is to minimize the duration and cost of all projects, expressed as: in, and These are weight parameters used to adjust... The relative importance of construction period and cost; when , When the time and cost are on the same order of magnitude, the project is considered to be more likely to be completed in a shorter time than the cost after weighting. Conversely, it is considered to be more likely to be completed in a lower cost than the cost. Items in the objective function Construction period , represented as: in, and Representing tasks The start and end times, in months, and ; Items in the objective function Cost , represented as: in, for Pay employees during the period Salary, expressed as: in, This refers to a unit of time used during the project development process. express Normal monthly salary express Monthly overtime pay Indicates in The set of tasks under development during this period; if employees are Total investment in this task set This indicates that the employee worked overtime and should be paid overtime wages; The constraints that the scheduling scheme needs to satisfy include: The set of employee skills assigned to a task to be performed in a project must include the skills required for that task, that is: in, express The set of skills possessed ; At any time t, the total engagement of each employee with the current set of executable tasks must not exceed that employee's maximum engagement, that is: in, This represents the set of currently executable tasks; The number of participants in a task cannot exceed the maximum number of participants allowed for that task, nor can it be less than the minimum number of employees required to meet the skill constraints of that task. in, Indicates participation in the task Number of employees; and These represent the maximum number of people who can participate in the task and the minimum number of employees required to meet the skill constraints of the task, respectively. At any time t, each employee can participate in no more than 2 projects, that is: in, express exist t The collection of projects participated in at any given time. It is a binary variable. Taking 1 or 0 respectively represents exist t Whether you are always involved in the project or not .
5. The software multi-project scheduling platform based on a multi-population artificial bee colony algorithm according to claim 1, characterized in that, The steps for grouping nectar source populations based on sub-target ranking values are as follows: Grouping before SN Each individual nectar source, and for each individual nectar source in the project objective function value Summing yields the target value for each individual nectar source. ; For each project, all SN Individual honey source for the project The sub-target values are sorted from smallest to largest to obtain the performance of each individual nectar source in the project. sorting value in Therefore, the individual possesses R A set of sorting values The smaller the ranking value, the better the target value of the corresponding project for that individual. Honey source populations are divided according to the number of projects. R First, create a subpopulation, then gradually add individuals to the subpopulation; for each individual, take its... R The index of the minimum value among the sorted values is selected randomly if multiple minimum value indices exist. Individuals are placed into subpopulations until all individuals have been traversed.
6. The software multi-project scheduling platform based on a multi-population artificial bee colony algorithm according to claim 5, characterized in that, The mechanism by which the leader bee uses subpopulation interaction to perform a global search includes: When individuals in a subpopulation interact with information from other subpopulations, based on probability... Determine if interactive information is needed; for each nectar source individual in the subpopulation, generate a random number between 0 and 1 for each subpopulation. If the random number is not greater than... If the target value of the subpopulation is not specified, it indicates that information needs to be exchanged with the subpopulation. The individuals in the subpopulation are sorted in ascending order according to the target value of the project, and the sorting number is recorded. An individual is selected from the subpopulation as its interaction object through a sorting selection strategy. When the interaction with other subpopulations ends, a greedy selection is performed based on the individual target values before and after the update, leaving the individual with the better target value.
7. A computing device for a software multi-project scheduling platform based on a multi-population artificial bee colony algorithm, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for executing any of the platforms according to claims 1-6.
8. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform the method of any of the platforms according to claims 1-6.
