动态环境下复杂多源数据的局部增量式粒融合计算方法研究

With the development of information technology, data collection is becoming more and more convenient, the sources of data collection are widespread, the types of data are diverse and the structures of data are complex. How to fuse these complex multi-source data and study its uncertainty is a challenge in the field of artificial intelligence. Granular computing is a basic theoretical tool in artificial intelligence. This project draws on the theory of granular computing to carry out research on information granulation of complex multi-source data. Existing information granulation methods of complex multi-source data are mainly based on the information granulation from the perspective of binary relationship in static environment. This information granulation method lacks certain learning ability and the adaptability to dynamic environment. This project mainly considers the information granulation method of complex multi-source data with learning ability in dynamic environment, and then constructs the local granular computing model on the basis of information granulation, analyzes the uncertainty of complex multi-source data. Because the granularity selected of the local granular computing model in dynamic environment will change, to implement the fast granularity selection of dynamic complex multi-source data, an incremental mechanism is introduced to study the granularity selection and rule extraction of dynamic complex multi-source data, and evaluate the effectiveness of the algorithm, thus realize the goal of incremental information fusion for complex multi-source data.

随着信息技术的发展，数据采集越来越便捷，采集的数据来源广泛、类型多样并且结构复杂.如何对这些复杂多源数据融合，研究其不确定性,是人工智能领域面临的一项挑战.粒计算是人工智能领域的基本理论工具,本项目借鉴粒计算理论，针对复杂多源数据的信息粒化开展研究.现有的复杂多源数据的信息粒化主要在静态环境下从关系角度对多源数据信息粒化，这种信息粒化方法缺乏一定的学习能力和对动态环境的适应能力.本项目主要考虑动态环境下具有学习能力的复杂多源数据的信息粒化方法，并在此基础上构建局部粒计算模型，对复杂多源数据的不确定性进行分析.由于动态环境下局部粒计算模型所选粒度会随之变化，为了实现动态复杂多源数据的快速粒度选择，引入增量机制对动态复杂多源数据的粒度选择和规则提取进行研究，并对算法的有效性进行评价，从而实现复杂多源数据增量式信息融合的目标.

随着信息技术的发展，数据采集越来越便捷，采集的数据来源广泛、类型多样并且结构复杂.如何对这些复杂多源数据融合，研究其不确定性,是人工智能领域面临的一项挑战.粒计算是人工智能领域的基本理论工具,本项目借鉴粒计算理论，针对复杂多源数据的信息粒化开展研究.现有的复杂多源数据的信息粒化主要在静态环境下从关系角度对多源数据信息粒化，这种信息粒化方法缺乏一定的学习能力和对动态环境的适应能力.本项目主要考虑动态环境下具有学习能力的复杂多源数据的信息粒化方法，并在此基础上构建局部粒计算模型，对复杂多源数据的不确定性进行分析.由于动态环境下局部粒计算模型所选粒度会随之变化，为了实现动态复杂多源数据的快速粒度选择，引入增量机制对动态复杂多源数据的粒度选择和规则提取进行研究，并对算法的有效性进行评价，从而实现复杂多源数据增量式信息融合的目标.