Abstract: Because traditional radial or non-radial DEA（Data Envelopment Analysis） models employ different perspective when they pursuit of maximum improvement on input or output, and in most cases they employ strong disposability assumption of pollution, considering this cause assessment deviated from the actual production process, poor distinguish and other problems, it proposes the radial and non-radial DEA comprehensive efficiency models based on grey correlation under assumption of weak disposability of environment. First, it proposes a comprehensive efficiency model framework which combines radial and non-radial characters based on environmental DEA technology. At the same time, it constructs the affinity indices based on grey correlation of input variables serials, obtains characteristic parameters of framework which is linked and measured radial and non-radial models, then puts them in the framework to obtain evaluation results. At last, an example real datum for 29 provinces of China in 2010 has been employed to describe their comprehensive efficiency, and the results show the validity of the model.

Key words: Data Envelopment Analysis（DEA）, radial and non-radial, affinity indices, environmental DEA technology

摘要： 由于传统的径向或非径向DEA（Data Envelopment Analysis）模型在追求投入或产出最大化改进时角度不同，并且模型中的污染物往往采用强可处置假设，考虑因此造成的评价偏离实际生产过程、区分力差等问题，提出环境弱处置性下的基于灰关联的DEA径向与非径向综合效率模型。基于环境DEA技术提出了一个径向与非径向特征结合的综合效率模型研究框架；基于投入变量序列的灰关联构造了接近度指标，从而获得框架中连接与测量径向与非径向模型的特征参数，将其置于框架中得到最终评价结果。以2010年中国29个省份综合效率评价为算例，结果验证了模型的有效性。

关键词: 数据包络分析（DEA）, 径向与非径向, 接近度指标, 环境DEA技术