基于支持向量机的泄爆压力峰值预测模型

Prediction model of venting peak pressure in vessel based on support vector machine

在现有的气体爆炸泄爆实验及理论研究的基础上,归纳总结泄爆过程中影响容器内压力峰值的主要因素,将这些因素作为输入,对压力峰值与各因素之间的内在非线性关系进行模拟,提出一种基于支持向量机的容器内气体爆炸泄爆压力峰值预测方法。对模型的有效性及预测性能进行验证,表明模型预测的结果与实验值基本一致;将模型的预测性能与现有的经验、半经验公式以及泄爆设计准则进行对比,表明建立的模型具有较高的准确性,为容器泄爆设计提供了一种新的途径。

Based on the experimental and theoretical research on gas explosion venting, major factors influencing on the peak pressure of vented vessel were summarized. A prediction method of pressure peak for a vented explosion based on the support vector machine （SVM） was proposed. By using the major factors as inputs and training the inner non-linear relationship between the pressure peak and each factor, validity and predictability of the SVM model were checked by the prediction values and experimental ones.Compared with the empirical formula and semi-empirical formula, the SVM-based model had higher accuracy, which would provide a new way for the design of gas explosion venting.