论文部分内容阅读
震伤人员的快速评估对统筹配备医疗资源和进一步预测废墟受伤和失踪人数十分关键。为提高震后受伤人员评估结果的可靠性,通过对影响震伤的关键性影响因子的提取,采用能够有效处理模糊性和非线性指标的RBF神经网络模型对震后受伤人员进行快速评估。从分析震后造成人员受伤的影响因子入手,从承载体减抗风险能力、暴露性和敏感性三个维度提出震伤人员预测指标体系;在受伤人员预测方法上,考虑指标的小样本性、非线性和部分指标的模糊性特征,将模糊逻辑与神经网络方法结合起来,采用动态优化的径向基(RBF)神经网络方法,以提高评估模型的全局搜索和优化能力,避免常规BP神经网络较早陷入局部优化的不足;案例结果显示:与BP神经网络训练的绝对误差3.24%相比,RBF神经网络震伤人员评估模型的绝对误差能够降低至1.71%,精度提高47.2%,说明本文评估模型评估可靠性高、模型鲁棒性强,能够推广于震灾应急的管理决策之中。
The rapid assessment of people injured in the earthquake is critical for co-ordinating the deployment of medical resources and for further predicting the number of people injured and missing in the rubble. In order to improve the reliability of the evaluation results of injured people after earthquake, RBF neural network model that can effectively deal with the ambiguity and non-linear indicators was used to evaluate the post-earthquake injured personnel rapidly by extracting the key influence factors of earthquake injury. Beginning with the analysis of the influencing factors of human injuries caused by the earthquake, this paper puts forward the forecasting index system of earthquake victims from three dimensions: the risk reduction ability, the exposure and the sensitivity of the carriers; considering the small samples’ Nonlinearity and some indicators of the fuzziness characteristics, the fuzzy logic and neural network methods combine the use of dynamically optimized radial basis (RBF) neural network method to improve the global search and optimization model assessment ability to avoid the conventional BP neural network The results show that compared with the absolute error of BP neural network training of 3.24%, the absolute error of the RBF neural network earthquake injury assessment model can be reduced to 1.71% and the precision can be improved by 47.2% Model evaluation has high reliability and strong model robustness and can be popularized in the management decision-making of earthquake disaster emergency.