The Research Of BDI Agent Test Case Generation Algorithm Based On AUML Model

With the continuous development of Agent Technology and the growingpopularity of Agent system application, testing of Agent system is also increasinglybecoming a hot.At present, Most scholars and researchers at home and abroad havepresented their methods to test Agent and Multi-Agent System. Although those testmethods and test theory is effective to promote agent software testing technologydevelopment and continuous improvement, there is lack of some methods in thetesting fields of BDI Agent.In order to improve test efficiency and test results of Agent, this paper presents aAUML sequence graph model-based method to test BDI agent.In our model, tostudy BDI reasoning model,using goal-plan tree to determine agent behavior logic ofhandling complex problem.Although traverse algorithm to determine agent behaviortrace. On this basis, analyze the relationship between agent. Firstly, using thegoal-plan tree struct to analysis the operation of BDI in order to determine thetrajectory of the individual agent; then, analysis the AUML model to test interactionamong agents. Proposing the algorithm to extract message from SD to determine themessage event set which will be handled by each agent which locates in MAS andaccording to message event set to generate goal plan tree; Finally, the paper proposesalgorithm to determine test scenarios by test paths to generate test cases.Experimental validation shows that our solution can generate completed test casesbased on AUML SD.