基于协同过滤的农业平台学习资源个性化推荐方法探究

由于传统方法未对不同用户之间的行为数据进行考虑,导致资源推荐满意度较低。为解决这一问题,提出基于协同过滤的农业平台学习资源个性化推荐方法研究。研究先引入影响值参数——Inf来表示不同关联用户之间影响关系的大小;然后,对时序评论和回复行为数据进行融合,构建农业平台用户资源兴趣矩阵;最后,以此为基础,采用协同过滤方式实现资源推荐。经过测试后,发现所提方法具有良好的推荐效果,可以有效保障用户对推荐资源的满意度。

Influence of various volute designs on volute overall performance

This paper discusses the influence of various volute designs on volute overall performance for a certain centrifu- gal compressor with both vaned and vaneless diffuser. Firstly, based on a free vortex flow pattern and the assumption of a circumferentially uniform flow at the design condition, a corrected method for volute design is adopted. By means of this method, corresponding to five geometric parameters affecting the volute overall performance, ten volute cases are designed. Secondly, the numerical simulation is employed and the detailed flow field and losses in different volutes with different geometric parameters are analyzed. The numerical investigation reveals that in all of the five geometric parameters, the radial location of the cross-section has the strongest influence on the performance of the volute. The non-uniform volute inlet formed by the upward vaned diffuser outlet flow is another important factor. Finally a relatively better value of D1/D2 is concluded.