基于云平台的大米产地确证方法研究

Research on Confirmation Method of Rice Producing Area Based on Cloud Platform

针对大数据背景下地理标志大米产地真伪鉴别的算法模型与实现技术,以大米中矿物质元素含量数据为基础,运用Hadoop分布式集群技术,构建了基于MapReduce的并行化随机森林、支持向量机、人工神经网络与线性判别分析算法模型。结果表明,并行化随机森林模型的判别准确率为97.55%,与相同条件下并行化构建的支持向量机、人工神经网络与线性判别分析模型相比具有更好的产地判别精度,同时依托并行化随机森林模型构建的云平台能获取到较好加速比,不仅能够实现对未知地区大米数据进行准确的产地鉴别,而且能够通过提升数据量或计算节点数,更高效地处理大规模数据。

Aiming at the arithmetic model and implementation technology of authenticity and falsity identification of geographical indication rice origin under the background of large data,a parallel random forest arithmetic model based on MapReduce was constructed by using hadoop distributed cluster technology on the basis of mineral element content data in rice.The results show that the discriminant accuracy of parallel random forest model is97.55%.Compared with the linear discriminant analysis model and parallel support vector machine under the same conditions,the parallel random forest model has better discriminant accuracy.At the same time,the parallel random forest model has a growing acceleration ratio,which can achieve rapid and accurate identification of unknown area data.