Research On The Correlation Performance Of Multi-resource Leveling And Non-dimensional Objective Functions

As one of the most important instruments of resource management,it can balance the resource usage during the project process.Resource leveling is helpful to reduce the project cost,improve the project quality and guarantee the project time schedule,which means a lot to the project comprehensive benefit.Thereby the problem of resource leveling has great value in terms of theory and practice.Based on the research of the reference,considering the difficult point and the shortcoming of the current resource leveling model,this paper analyzes the resource allocation characteristic,the correlations among the resources,and present a resource pre-treatment process.A new objective function is created and the algorithms are improved to solve the problem.Also,some new problems concerning the resource leveling problem are studied here.The innovative points of the paper are as the following:(1)A heuristic method is designed to draw an AOA network with minimal dummy activities and dummy nodes.On the basis of the analysis to the influence of the node type on the activity float algorithms,some conclusions are drawn as the following:The node type algorithms for activity safety float and free float are found to have error for out-dummy node and in-dummy node separately.Amendments are made and the example result shows that the conclusion in this paper is correct and the amendment algorithms are right.(2)The correlations among resources are studied,and a pre-treatment model is designed for the multi-resource leveling,which includes the following points:A resource allocation network is designed,and the resources are classified as Non-leveling resource,local-leveling resource,segment-leveling resource and global-leveling resource.On the basis of the resource allocation networks,methods of judgment are designed to identify each resource classifications.Different treatment should be made to each class to optimize the multi resource leveling model.Eliminate the non-leveling resource,confirm the leveling interval of the local-leveling resource, and treat the segment-leveling resource as several local-leveling resources.An index system and a comprehensive index are designed to evaluate the importance of the resource.The importance index system consists of the cost importance index,the time importance index,the activity importance index and the float importance index.Then the AHP is applied to confirm the weight of each index and a comprehensive importance index is calculated for each resource.The resources are clustered by the gray correlation clustering method.Then,the representative resource is selected and the others are eliminated.This method can not only simplify and optimize the resource leveling model,but also consider most of the resources.(3)The resource entropy is defined.As a distribution based on ratio,this objective function not only overcomes the resource dimension limit,but also has a good performance character to reflect the smooth level.Thereby we defined the multi-resource leveling model based on resource entropy,the resource classification,clustering and reductions are considered in this model.The PSO is improved and a comprehensive process for multi-resource leveling is created.The case result shows that the performance of the comprehensive process is good.(4)The model of Resource Leveling under Multi-mode activity is presented and a heuristic method based on the minimum deviation value is created.The case study shows that this method is easy and workable.The Model for he RCSP with Resource Leveling Requirement is created and a heuristic method based on leveling interval and barcentre is presented.The case study shows that this method is easy and practical.