Research On The Skilled Workforce Scheduling Problem In The Engineering Project

With the globalization of economic development and the individuation of user require-ments, it is a trend that the enterprises adopt the project-oriented production mode. Therehas emerged a number of construction projects about architecture, energy source, transporta-tion and water conservancy, as well as production manufacture projects, software projects,scientific research projects and so on. The workforce scheduling is a key problem of the engi-neering project management. Because of the engineering projects being complex increasingly,the labor cost being rising continuously, the multi-skilled workforce being a training trend andso on, the workforce scheduling becomes more dificult. The researchers from the science andpractice have further focused on the research of the skilled workforce scheduling problem inthe engineering project management. The project scheduling and stafing is a kind of complextask scheduling and human resource allocation problem, which belongs to a crossed researcharea involving machine scheduling, project scheduling and workforce scheduling.There are many infiuencing factors of the workforce scheduling needing to be considered inthe practical engineering project, which mainly include workforce skills, workforce eficiencies,learning forms, wage levels, as well as optimization objectives such as project duration, totaltardiness, salary cost, overhead and time cost etc. This dissertation systematically studiesfive types of project scheduling and stafing chiefiy from two dimensions of skill type andeficiency type, separately constructs mathematics models for each problem, and proposesthe corresponding algorithms to solve them. The results of a great of random numericalexperiments show that the algorithms have strong optimization capability and stability. Theprimary work and innovation achievements of this dissertation are described as follows:(1)The research framework of the workforce scheduling problem in the engineering projectmanagement is proposed, based on the analysis and comparison of parallel machine schedulingwith precedence constraints, resource-constrained project scheduling and workforce scheduling.(2)The single-skilled workforce scheduling problem is studied. For the workforce schedul-ing with homogeneous eficiencies minimizing project duration, a zero-one integer linear pro-gramming model is constructed. Since this problem can be considered as a special case ofclassical resource-constrained project scheduling problem, a serial insertion schedule genera-tion scheme is proposed. Numerical experiments show that the proposed algorithm distinctlyoutperforms another any-order schedule generation scheme with respect to the quality andeficiency of solution. For the workforce scheduling with heterogeneous eficiencies minimizing project duration, an integer linear programming model is constructed, and a priority rule-basedheuristic is first designed to obtain a single initial solution, then a hybrid simulated annealing(HSA) is proposed to further explore the solution space. Numerical experiments show that theHSA can find good solution more accurately and eficiently than the conventional simulatedannealing.(3) The complete-skilled workforce scheduling problem is studied. For the condition ofworkforce with heterogeneous eficiencies, three optimization objectives are considered. Thefirst objective is to minimize project duration, for which an integer linear programming modelis constructed, and a priority rule-based heuristic serial schedule algorithm is designed. Nu-merical experiments show that the proposed algorithm clearly outperforms another existingalgorithm with respect to the quality and eficiency of solution. The second objective is tominimize total tardiness, for which an integer linear programming model is constructed, anda priority rule-based heuristic is first designed to obtain multiple initial solutions, then ahybrid genetic algorithm (HGA) based on probability evolutionary strategy is proposed tofurther explore the solution space after experimentally comparing the probability evolutionarystrategy with the elitist evolutionary strategy. Numerical experiments show that the HGAcan get better result than the conventional genetic algorithm within the same runtime. Thethird objective is to minimize the sum of increasing type salary and time cost, for which aninteger nonlinear programming model is constructed, and an amended DP algorithm is pro-posed through combining heuristic rule with conventional dynamic programming. Numericalexperiments show that introducing the bound of every task completion time in the proposedalgorithm can greatly improve the computational eficiency.(4) The multi-skilled workforce scheduling problem is studied. For the workforce schedul-ing with homogeneous eficiencies minimizing the sum of overhead and difierence type salarycost, an integer nonlinear programming model is constructed. A hybrid genetic algorithm isproposed through combining the priority rule-based parallel schedule generation scheme withthe conventional genetic algorithm. Numerical experiments show that several priority rules canhelp the proposed algorithm converge to ideal points. For the learning workforce schedulingwith heterogeneous eficiencies minimizing project duration, a zero-one integer nonlinear pro-gramming model is constructed, and a priority rule-based heuristic is first designed to obtaina good initial solution, then a hybrid particle swarm optimization (HPSO) algorithm is pro-posed to further explore the solution space through introducing discrete operators to modifythe classical particle's velocity and position equations. Numerical experiments show that theHPSO can convergence to better solution than the conventional particle swarm optimizationalgorithm within the same runtime. (5) A practical software development project of electronic archive is selected as study case.Two methods including microsoft project 2003 and hybrid genetic algorithm are separatelyapplied to solve the case. The results show that the using of hybrid genetic algorithm canmarkedly reduce the project duration, which indicates the achievements of this dissertationhave application and extension values.