Abstract: Effective partition of the input and output spaces must be involved when sample data is transformed into some qualitative rules.When dealing with multidimensional sample data，the point mentioned above is much more difficult.For the purpose of handling that problem，an improvement on scale cloud transform method is proposed，named two-dimension scale cloud transform method with the purpose of handling two-dimension sample data.The two-dimension T-S cloud controller also presented.An example is researched to prove the validity of this new method，and the simulation results show that the new method is simple，powerful，and the new T-S cloud controller consisted of the rules generated by the new method can reproduce the original data very well，and that the number of intelligible rules is few.

Key words: cloud model, rules extraction, qualitative reasoning, cloud controller

摘要： 从定量数据到定性概念的转换必然涉及到样本数据的划分问题，当样本数据维数较高时，对其合理划分是从中提取定性规则的难点问题。针对此问题，改进了尺度云变换算法，提出二维尺度云变换算法，使之可以处理二维数据。还给出了二维T-S型云控制器的设计，并结合一组实例数据，验证了算法的有效性。仿真结果表明，该算法可以较好地实现对原始数据空间的划分，提取的定性规则数量少，易于理解；由该方法提取的规则构建的T-S型云控制器能较好地复现原始数据分布，误差较小。

关键词: 云模型, 规则提取, 定性推理, 云控制器