以《机械工程材料》杂志3 a共36期刊物每期发表论文的基本情况统计数据为基础,根据BP人工神经网络算法原理,建立了科技期刊每期质量与载文量、基金论文比、研究院校作者比这3个评价指标之间的关系模型,并分析了各指标对每期质量的影响程度.结果表明:BP人工神经网络用于科技期刊每期质量的评价是可行的,所建模型能很好地评价每期杂志的实际质量,可以作为一种科技期刊质量评价方法推广应用,特别是可用于对科技期刊每期质量进行出版前预评:其中载文量是影响科技期刊每期质量的最主要指标,基金论文比和研究院校作者比居其次,且影响程度大致相同.

Based on the statistical data about basic situation of papers published in per issue among 36 issues in 3 years of Materials for Mechanical Engineering journal,the predicting model for the relation between per issue quality of the scientific and technical periodical and 3 evaluation factors which were paper quantity,rate of articles supported by foundations and rate of authors from academe and universities had been developed with back propagation (BP) artificial neural network method,and the extent of the influence caused by the factors on per issue quality of scientific and technical periodicals had been analyzed.The results show that it is feasible to evaluate the per issue quality of the scientific and technical periodicals by BP artificial neural network method,and the model which can primely evaluate the practical quality of per issue is worth of spreading and popularizing for the quality evaluation of scientific and technical periodicals,especially for the pre-evaluation of per issue quality of the scientific and technical periodicals before publication.The main influencing factor on per issue quality of scientific and technical periodicals is paper quantity,and the next ones whose influence degree are about the same are rate of articles supported by foundations and rate of authors from academe and universities.