Researches On Urban Mass Transport Operation System Software Behavior Dynamic Evaluation Method

With the gradual progress of Suzhou subway engineering, people have more requestand expectation for their reliability,availability,safety,etc.. In order to solve the increasinglyprominent issue of credibility of Urban Mass Transport Operation software, it’s not enoughto make the traditional quality assurance(such as testing and validation), it needs to makeeffective analysis and trend prediction for interactive behavior.Urban Mass Transport Operation Software is different from traditional distributedsoftware,it exhibits characteristics such as a big scale and behavior complexity. Taking thesoftware behavior as the breakthrough point, the paper did some researches on softwarebehavior trace analysis and prediction of intention, which means to establish a softwarebehavior trust dynamic testing method.Through mapping the state variation with semantic sequence of events, Consideringinteractive events as the basic unit,the paper use Sequence Cycle Reduction algorithm tostreamline behavioral sequence and make use of the repeated sequences to procude thesequence which can reflect software behavior characteristics. Using the sequencecomparative analysis to make evaluation. HMM is to make trend prediction, this is calledbig granularity behavior credibility dynamic evaluation. At the same time, the paper paysattention to the experience of events knowledge and values of the parameters and otherdetailed information, using MEBN tools to build the Specific Situation of BayesianNetwork (SSBN), which describes the interaction of complex process and effect, this iscalled small granularity behavior study. In this paper, using small particle size measuringmethod to process complex cicurmstances group interaction creditability. Use large sizebehavior analysis to process general situation. The main work of this paper includes:(1) Study on software behavior characteristics of Urban Mass Transport OperationSystem. To explore the relationship among the intelligent subsystem and their informationrequirement. The paper aim to provide theory support for achieveing an efficient anddistributed UMT Operation System.(2) Propose a set of software process description method, translating the relevantcredible data captured by monitoring mechanism to be software behavior information, thendesign the software behavior information track format and a storage mechanism, towardsoftware entity interaction cycle subsequence problem, eliminating the repeated sequncetrail, and storing software behavior with the most concise and broadly representativeformat.(3) Presents a complex marked big granularity behavior dynamic testing method.Using sequence of two comparison algorithm to analysis the credibility of behavior.EM-based data reconstruction method is for training data.The using of HMM is for theprediction of behavior safety category. A simulation experiment is given to show theadvantage of this method.(4) Puts forward a kind of complex situation with small particle behavior trajectoryanalysis method.With the MEBN first-order logic semantic representation and probabilisticreasoning ability, it can effectively describe and analyze the multiple software entitiesinteracting complex behavior.Dynamic testing method study Based on the behavior of the software,it can not onlydevelop the dynamic trust evaluation theory, but also has the very good instructionsignificance.