差异化结构下排版软件运行缺陷优化预测仿真

Optimization and Prediction Simulation of Operational Defects in Typesetting Software under Differentiated Structure

针对现有的排版软件运行缺陷预测方法存在的预测准确性较低、实时性较差等问题,提出了一种基于聚类和偏相关关系的排版软件运行缺陷优化预测方法,利用排版软件运行数据的属性值判定数据属性类别,随机选取一些数据对象作为聚类中心,将与聚类中心相似的对象分配到相应的位置,利用K均值聚类找出令标准测度函数最小的对象,使后续分类中每个属性间尽可能达到属性独立。根据彼此间独立的核心属性集计算排版软件运行数据类的先验概率,划分排版软件运行数据集类别,在不同的运行数据集上分析软件运行标识函数与缺陷数据之间的关系,给出两者之间的偏相关函数值,由值计算出软件运行数据复杂密度属性值,根据复杂密度属性值构建排版软件运行缺陷优化预测模型,实现对排版软件运行缺陷的预测。实验结果证明,所提方法能够准确、高效的预测出排版软件的运行缺陷。

Due to low prediction accuracy and poor real-time performance, this article presented a method to optimize and predict the operational defect of typesetting software based on clustering and partial correlation. Firstly, we used the attribute value of typesetting software running data to determine the data attribute category. And then, we randomly selected some data objects as the cluster center. Secondly, we assigned the objects similar to the cluster center to the corresponding positions. Thirdly, we used K-means clustering to find the object that minimized the standard measure function, so that the attribute independence among attributes was achieved as much as possible in the subsequent classification. According to the independent core attribute set, we calculated the prior probability of the typesetting software running data, so as to divide the category of typesetting software running data set. Based on different running data sets, we analyzed the relationship between the software running identification function and the defect data, and then we gave the value of partial correlation function. On the basis of this value, we calculated the complex density attribute value of operation data. According to the complex density attribute value, we built the optimization prediction model of the typesetting software operation defect. Finally, we could predict the running defect of typesetting software. Simulation results show that the proposed method can accurately and efficiently predict the operational defect of typesetting software.