Research On Improved Artificial Bee Colony Algorithm For Staff Scheduling Problems In Call Centers

Call centers can be broadly described as facilities designed to offer responsive and effective service for customers via telephone communications,such as business consultation,information inquiry,after-sales support and transaction processing.According to the industry research,there are more and more call center enterprises in recent years in China,and the scale of the industry expands rapidly,with over 200 million employees.However,call center staff scheduling problems are often highly dependent on actual enterprise scenarios,taking into account of various labor factors.In essence,it is a very complex combinatorial optimization problem,which needs to be solved by means of operational research and combinatorial optimization techniques.Thus,research on staff scheduling problems of call centers has its important theoretical significance and application value.Based on many practical scenarios of outsourcing call center enterprises in China,the thesis extracts some specific staff scheduling problems from the perspective of labor cost,service level and employee satisfaction,and first attempts to construct a series of mathematical programming models,which considers same-shift-dominant assignments,global service level,fair weekends off among staff and employees’ working preferences,respectively.Faced with various kinds of constraints and complex problem structures,the optimization models mentioned above are NP-hard from a complexity of view.Thus,it is also the emphasis of this thesis to design efficient optimization methods for solving these call center staff scheduling problems.Specifically,the main research contents of the thesis are as follows:(ⅰ)To solve the staff scheduling problem with consideration of same-shift-dominant assignments,it is decomposed into three relatively independent sub-problems,which are monthly off-duty scheduling problem,daily shift scheduling problem and personnel assignment problem respectively,through the analysis of the model logical structure and optimization properties.Firstly,two integer programming models are utilized to solve the lists of daily shift assignmens and available employees.Then,an artificial bee colony algorithm is designed to realize the shift assignments of employees,in order to meet the requirement of same shift within the week as much as possible.To improve the quality of the initial population of artificial bee colony algorithm,a heuristic algorithm is constructed to generate a same-shift-dominant scheduling scheme,combining the problem optimization property and greedy rule.Experimental results show that the proposed algorithm can balance the enterprise’s management demand of same shift and labor costs.Finally,the influence of this human resource allocation mode on the labor cost of enterprises is discussed.(ⅱ)To solve the staff scheduling problem with the consideration of multi-dimensional service level,a hybrid integer programming and artificial bee colony algorithm is proposed.Firstly,an artificial bee colony algorithm is designed to generate a scheduling scheme that satisfies the constraints such as the requirement of service level of each interval,so as to minimize labor cost.Four kinds of neighborhood structures are elaborated,from two perspectives of employee attendance and employee redundancy of intervals.Then,integer programming employed under several ruin-and-recreate principles is utilized to refine the scheduling scheme,so as to improve the global service level.These principles can be chosen with the same probability.The effectiveness of the hybrid algorithm is verified by computational experiments based on some problem instances.Moreover,the quantitative analysis and research on labor costs and global service level are carried out,which gives management insights and implications for making staff scheduling schemes of call centers.(ⅲ)To solve the staff scheduling problem with consideration of weekend-off fairness,an enhanced artificial bee colony algorithm is proposed.Firstly,the problem structure is decomposed based on the classification of intraday,week and cross-week constraints.Based on the derivation of the lower bound of the problem and the analysis of its optimization properties,a basic artificial bee colony algorithm is designed to solve the problem.Then,a series of techniques are incorporated into the algorithm.For example,a good-quality solution generated by a heuristic algorithm is introduced to the initial population.Four elaborate neighborhood structures are designed based on the decomposed structure of the problem and a series of quantitative indicators of schemes.Furthermore,the food source evolution mode is modified in the algorithm,and an improved abandoning mode is employed in the onlooker stage.Using real instances from a call center enterprise,the experimental results show that the proposed algorithm can achieve(sub-)optimal solutions for large-scale problems;at the same time,the influence of weekend-off fairness on labor costs is discussed.(ⅳ)To solve the staff scheduling problem with consideration of employees’ working preferences,a two-stage approach is proposed,where the former generates preliminary solutions satisfying all the hard constraints,and the latter refines the solutions by fulfilling soft constraints as much as possible.The proposed approach employs artificial bee colony algorithm as the main search framework in both stages,and appropriate neighborhood structures are designed,respectively.Specifically,in the latter stage of the artificial bee algorithm,integer programming employed under a suitable ruin-and-recreate principle is embedded in the scouts bee stage,which helps intensify and diversify food sources.The experimental results show that the proposed approach is effective and efficient in achieving good solutions for large-scale problem instances.Finally,the relationship between employee satisfaction and labor costs is discussed,and the influence of employees’ working preference on generating staff scheduling schemes is also explored.Finally,this thesis also discusses some challenging topics which deserve further research in the future based on the above research results.