Research Of Intelligent Computing Strategy Based On The Cloud Model

There are a lot of uncertain things in nature and society. How to represent and dealwith them has been a bottle-neck in the scientific community. Random and fuzzy aretwo main property of the uncertainties. For the disadvantage of probability theory andfuzzy set theory on dealing with uncertain knowledge, the cloud model was proposedby Professor Deyi Li. Cloud model is the new theory that we use to process andanalyze the uncertainties by means of organically combining the randomness andfuzziness, including the definition of cloud model, digital feature, cloud generator,virtual cloud, cloud transform, uncertainty reasoning of cloud and so on. Cloud modelis the uncertainty model that makes the interchange of the qualitative concepts ofnatural language and quantitative expression. It is its superior characteristics ofdealing with the uncertain knowledge that the research on the theory of the cloudmodel and its applications are gradually becoming a hot spot in just ten years.However, the research on the theory of the cloud model and its application are still inits infancy at home and abroad at present, not yet mature and complete. On the basisof the analysis and summary of the existing theory and research situation, this papermakes a more in-depth discussion on the theory of cloud model, applies the cloudmodel into new areas, and proposes new methods to expand the scope of applicationof cloud model.This paper is focus on two applications of cloud model in time series similaritymeasurement and automatic generation of software test data:(1)The cloud model is applied in time-series similarity measurement. We mainly usethe superior characteristics of cloud model on qualitative knowledge representation toqualitatively describe the macro characteristics of time series. Then, we also add thegeneral information of the sequence as the description of the local trend of timeseries，according to the trend information that is essential for time series. According tothe two aspects of information, the experiment finally proves the new method iseffective to time series similarity measurement.(2)The cloud model is applied in automatic generation of software test data. Wemainly use the superior characteristics of cloud model on qualitative knowledgerepresentation and mutual transformation of qualitative and quantitative, which hasadvantages on evolutionary computing applications, to automatically generatesoftware test data. According to the requirements of automatic generation of test data,and the characteristics of evolutionary algorithms based on cloud model, theevolutionary algorithm is applied to automatically generate software test data, selectand construct the appropriate fitness function, to achieve the automatic generation oftest data. The proposed algorithm is compared with other methods, and experimentalresults show that the algorithm is effective and stable to automatically generate test data.Finally, the research results and shortage are summarized. Future work of thecloud model applied in time series similarity measurement and automaticgeneration of software test data, are discussed.