Abstract: A new scale cloud transform method is proposed firstly.Both the normal semi-cloud model and the shift operator are introduced to that method to make sure that data partition in the method can reflect some information about the actual data distribution，to reduce the number of rules generated by the new method，and to increase the rules’ fitness.Because key points and best breakpoints are exploited，the rules base generated by the proposed method is complete.This paper first presents T-S cloud controller，and gives the simple design process of this new type of intelligent controller based on scale cloud transform.The simulation results show the new method is simple，powerful，and the T-S cloud controller consisted of the rules generated by scale cloud transform can reproduce the original data very well，and can reduce the control error to any degree theoretically.

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

摘要： 提出了尺度云变换的新方法。该方法采用“半正态云”作为基本云，并引入位移因子，使基于数据的规则提取算法中的数据划分能充分反映数据的实际分布，减少提取的规则数量，提高规则的推理精度；通过引入数据关键点及最佳分割点以保证规则库的完备性。还首次提出T-S型云控制器的概念，并将尺度云变换方法应用到T-S型云控制器的设计中。仿真结果表明，该方法简单有效，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数据分布，误差能控制在给定范围内。

关键词: 定性推理, 云模型, 规则提取, 云控制器