With the popularity of the cloud computing, more and more data owners are motivated to outsource their complex data from local sites to the cloud for great shared comuputing sources and economic savings. But for protecting data security and privacy, sensitive data have to be encrypted before outsourcing, which obsoletes some traditional data utilization, for example the multi-Keywords search. In this paper, we develop a enhanced multi-Keywords fuzzy search scheme. On one hand, a novel method is designed to transfor Keywords based on BloomFilter, which can reduce the index length effectively while keeping the fuzzy search rate. On the other hand, the similarity measure method is improved based on dynamic adaption of confusion parameters to achieve various stringent privacy requirements, which shows the data user's favoritism better. Experiments on real-world data show that the proposed scheme is feasible, effective and accurate.