论文部分内容阅读
有效预测震灾人员的存活情况是紧急配置应急资源和提高救援效率的首要工作。为提高震灾人员存活预测的精度,本文首先依据区域灾害系统理论和现有研究成果提出震灾人员存活预测指标。其次,针对震灾人员存活量指标数据的小样本、高维度、非线性特征,考虑将支持向量机(Support Vector Machine,SVM)模型引入震灾人员存活量预测中,为有效降低SVM在高维空间中非线性分类的误差,采用Mexican母小波核函数替换满足Mercer内积条件的核函数,以改变常规核函数缩小偏差的局限性,提出用于预测震灾人员存活量的Mexican小波SVM(Mexican Wavelet-SVM,Mexican Wv-SVM)模型。数值算例表明:相比于标准SVM、BP神经网络,Mexican WvSVM模型具有预测精度好、训练速度快和运行稳定性好的特征,证明了模型的可靠和有效。
Effective prediction of the survivals of earthquake-stricken people is the most important task of emergency allocation of emergency resources and improvement of rescue efficiency. In order to improve the prediction accuracy of earthquake survivors, this paper firstly puts forward the prediction index of survivals for earthquake disaster victims based on the regional disaster system theory and existing research results. Secondly, according to the small sample, high dimensionality and non-linearity of the data of the survivorship index of earthquake disaster victims, we consider introducing the Support Vector Machine (SVM) model into the prediction of the survivals of earthquake victims. In order to effectively reduce the SVM in the high dimension To solve the problem of nonlinear classification in space, a kernel function satisfying the inner product of Mercer is replaced by a Mexican mother wavelet kernel function to change the limitations of the conventional kernel function to reduce the deviation. A Mexican wavelet SVM (Mexican Wavelet-SVM, Mexican Wv-SVM) model. Numerical examples show that compared with the standard SVM and BP neural networks, the Mexican WvSVM model has the characteristics of good prediction accuracy, fast training speed and good running stability, which proves the reliability and effectiveness of the model.