基于BKW状态方程的爆轰产物及参数的改进算法

为了降低爆轰产物及爆轰参数的求解难度,通过对质量守恒方程的基本可行解进行线性组合,得到了爆轰产物的平衡组成,并在此基础上进一步获得了爆轰参数。其主要实现方法为:由最小自由能原理对基本可行解进行筛选,然后根据最大放热原则确定初始解,并在最小自由能原则的引导下,由初始解和基本可行解的线性组合获得爆轰产物的平衡组成,以上操作步骤均由自编程序完成。应用支持向量机(SVM)线性模型对BKW状态方程参数进行了调整,并详细介绍了其主要步骤。使用此方法预测了PETN、CL-20和含铝炸药的爆轰产物及爆轰参数,经参数调整后,发现预测结果与实验值吻合良好;通过与单质炸药爆轰实验数据对比,发现调整BKW状态方程参数时,应当尽可能使用爆轰产物中气体含量相近的含能材料对SVM模型进行训练;若预测含铝炸药,应当使用铝氧比接近待测炸药的样品来训练SVM模型。

Improved Algorithm of Detonation Products and Parameters Based on the BKW Equation of State

To reduce the difficulty of predicting the detonation products and solving detonation parameters, the equilibrium compositions of detonation products were achieved by linear combination of the basic feasible solutions, which were obtained from the mass conservation equations;and the detonation parameters were further obtained based on equilibrium compositions.The major process was executed as follows: the basic feasible solutions were selected out by the principle of minimum free energy, and the initial solution was given by the principle of largest heat release.The equilibrium compositions of detonation products were linearly searched by uniting the initial solution with the basic feasible solutions, and the above-mentioned operation steps were completed by using self-made program.The parameters of the BKW equation of state were adjusted applying the linear support vector machine (SVM), and its main steps were introduced in detail.The detonation products and parameters of PETN, CL-20 and aluminized explosives were predicted with this method, and after parameter adjustment, it is found that the predicted results and the experimental ones are in better agreement.In comparison with the detonation experiment data of single compound, it is found out that when the BKW equation parameters are adjusted, the energetic materials with more similar percentage of gas fraction in detonation mass to the explosives predicted should be used as the training set of the LS-SVM model.If the detonation parameters of aluminized explosives are predicted, it should use the Al/O ratio close to measured explosive to train the SVM model.