| Although the number of project management software is large and its function is powerful at home and abroad,it has many defects in optimization of network plan and resource leveling.The main reasons for causing the situation are that the theory and algorithms of optimization of network plan and resource allocation don't satisfy the demand of reality at present.Five new viewpoints are described in this doctoral dissertation,there are two new ideas in theory among them, the rest is three new methods in algorithm. Two new ideas in theory are the method of the compression of duration at least cost and resource allocation in optimization of network plan separately. Three new methods in algorithm are separately using orthogonal list efficiently in Network Chart ,the hybrid genetic algorithms on resource leveling as well as a programming model and algorithms on sequence optimization of optimal technology of network plan. The method of the compression of duration at least cost firstly figures out key critical path starting from the lowest cost and the normal time of per activity,and then chooses key activity which is the lowest cost rate in the key critical path to compress its duration,so that we can satisfy required time object. Using the method of the compression of duration at least cost to select decision plan,which can't distinguish the effect degree of compression of different key activity to quality and safety,only considers the duration and the cost rate of key activity in different decision plan,the solution which is calculated in this way can't reflect the real situation of project,the different of result and reality is large,therefore,it is impossible for us to give priority to compress the duration of key activity which effects slightly on project quality and safety,the new thinking which is directed against the problem is described in this doctoral dissertation. The main methods of resource leveling are now graphic , formula judgment and genetic algorithms at home and abroad. If network chart is simple and the number of resource needing banlanced is small, the three methods can be used for even though resource calendar were determined it would be more easy to revise it. If network chart is complicated and the number of resource needing banlanced is large,the three methods can't be used for the number of resource needing banlanced is very large and the contradiction of supply and requirement is also obvious.A new idea to solve this problem is put forward as follow:the resource leveling should be considered when a resource is allocated,the problem of resource leveling will be solved as long as each resource is used evenly. The hybrid genetic algorithms on resource leveling have overcome the defects existed formerly,they adopt reasonable code rule,build resource leveling model of genetic algorithms.The algorithms consider resource calendar,they solve the contradictory of resource leveling and unavailableness completely. The network chart of optimization of network plan is very complicated chart.Most of applied software don't provide the data structure of chart.Even they was supplied,it is a common chart which doesn't have a help to express a specific chart. Network Chart is a Directed Graph which applied software can not provide its data structure,this means that we must design the method of using network chart by ourself. We find that orthogonal list is the best method describing network chart by looking up a lot of data and carefully research to network chart,furthermore,the speed of applied software which was programmed by using orthogonal list is very fast.We have browsed many books and magazines after deciding to use orthogonal list,we find that only the book《DATA STRUCTURES WITH C++》written by YAN Wei-ming and WU Wei-min has introduced a little about orthogonal list.In this book only the input method was presented.Many other important methods such as adding node,deleting node,adding process,deleting process and so on haven't been presented.We have replenished the function of orthogonal list...
|
PDF Full Text Request