Message is delivered between nodes through single hop or multiple hops in wireless sensor networks(WSNs). Obtaining link quality information in advance which provides reference for the upper routing protocol to select link is the basis of delivering to the monitoring center accurately and in time. In this paper, a concrete analysis about the related works on WSNs link prediction based on intelligent learning is presented. A novel model, Cloud Model, is proposed to predict link quality. The large amounts of link quality samples are collected from different scenarios, and then adaptive gauss cloud transformation is applied to clustering training samples, such as RSSI, LQI, SNR and PRR. Taking the limit of node's resources into consideration, an Apriori algorithm is applied to mining the association rules from the RSSI, LQI, SNR and PRR which had been clustered. At last, three dimensional cloud model is employed in WSNs link prediction. Comparing with BP neural network prediction method, the proposed prediction model achieves higher accuracy as simulation experiment results show.