油墨预设LS-SVM模型研究

Research on the model of ink presetting based on LS-SVM for offset printing

针对胶印机油墨预设关键技术中非线性、高维数、系统性的特点,使用常规测量方法难以解决油墨精确预设的问题,提出了在考虑输墨系统串墨影响作用下,建立基于最小二乘支持向量机的多元非线性映射关系下的油墨预设模型,解决印品墨区图文网点覆盖率与墨键开度之间的非线性问题,并通过胶印机实际数据仿真实验,对通过样本训练的油墨预设模型的预测结果和实际生产中胶印机油墨供给的设置结果比较分析。分析结果表明,本文模型具有较高的预测精度和稳定性。

According to the characters of nonlinear, high dimension and systemic of offset ink presetting technology, the problem of accurate ink presetting is difficult to be solved by using con- ventional methods. With the consideration the effect of ink vibrator in inking system, a model of ink presetting for offset printing is proposed based on LS-SVM, in order to solve nonlinear problems between the ink coverage percentage of ink regions and ink key opening degree. The simulation experiment is executed with the actual offset printing data. Comparing the prediction results of ink presetting models by training sample with the results of actual ink supply of offset, the experimental simulation shows that the ink presetting model based on LS-SVM is of high accuracy and stability.