Design And Application Of Genetic Algorithm For Testing Planning In Laboratory

Test planning in laboratory is an optimization process,which allocates test equipments according to given requirements in a defined time frame in order to optimize utilization of resource of laboratory.Study in the past mostly focused on test planning with assumption that environment applied is simplified and ideal,and it is based on defined Mathematical Model,thus there is a huge gap between theory and practice of test planning in laboratory.There are many kinds of constraints on test planning in laboratory management,such as specified starting time or finishing time of testing for certain samples,or specified final finish finishing time for all samples.There are also special application environments,such as a testing item with multiple testing equipments or a testing equipment enabled to conduct multiple testing items.Design and adjustment of algorithms according to constraints and application environment is the key process to work out a good test planning in laboratory.This dissertation mainly refers the study of constraints and application environment in laboratory,develops and designs Genetic Algorithm for test planning in laboratory.The process is proceeded by creating genetic representation,definition of fitness function,selection & optimization of GA and finalizing of test planning w/ Matlab programming,major focus in the dissertation are realization of two typical test planning and test planning under constraints and application environment in X company.The main research work lies in four aspects as follows:1.Analysis of application of classical heuristic algorithm and GA for simple test planning,selection of GA parameters,design of Partial-Mapped Crossover(PMX),programming of GA and final realization of programming with Matlab.2.In the process of common test planning,this dissertation raises the thought of genetic representation creation,definition of initial population,definition of fitness function and analysis of different crossover operator.Two GA based on different crossover operator investigated and realized,comparison is made for those two GA.3.Analysis of equipments test planning under actual constraints and application environment.Two typical constraints drawn from equipments test planning of real laboratory management: the test planning with given testing starting time and finish time.Further study made on a common application environment of equipments test planning: same test with several test equipments.4.Analysis of possible solution for equipments test planning with given constraints,bring forward an idea of algorithm that it discards unqualified population with weighted fitness function.Realizes equipments test planning by design of weighted fitness function and corresponding GA algorithm.Tries to design a specific GA algorithm with proper fitness function for equipments test planning,which has application environment of same test with several test equipments,then validated and realized via Matlab.5.Base actual situation of X company,three algorithms are analyzed and realized for test planning in laboratory: manual test planning based on test cycle,genetic algorithm based on position crossover and genetic algorithm based on components crossover.Comparison of three algorithms in terms of whole lead time,processing time of algorithm and utilization rate of test equipment.Based on the detail analysis and study above,genetic algorithm designed is applied,analyzed and validated in real application environment in laboratory.The effectiveness and efficiency of genetic algorithm are assessed as well.The results of complete study are summarized and further study to be done in future is proposed.