基于BP人工神经网络的潜在震灾预测系统

结合人工神经网络自身的特性和地震灾害预测研究的特点,应用BP人工神经网络模型,建立了潜在地震灾害预测系统。利用大样本数据对网络进行了训练,形成了有识别和记忆功能的非线性预测系统。通过对网络的测试和检验,论证了该系统在预测潜在地震灾害上的可行性和有效性。同时,从测试精度出发,探讨了这种预测网络存在的不足,并给出了相应的改进建议。虽然提出的神经网络模型预测精度还有待提高,但其量化指标仍可为地震灾区政府抗震减灾工作提供参考。

POTENTIAL SEISMIC DISASTER FORECASTING SYSTEM BASED ON BP ARTIFICIAL NEURAL NETWORK

We used model of BP artifical neural network to build potential seismic disaster system based on characteristics of artifical neural network and seismic disaster forecasting research.Used large sample datum to train the BP network,formed an identifiable and rememberable non-linear forecasting system.By testing and checking up,the feasibility and validity of the system for forecasting potential seismic disaster were proved.At the same time,according to testing precision,the shortage of forecasting network was discussed,and we give some suggestions.Although the forecasting precision of neural network model need to advance,its quantificational index can offer reference for earthquake resistance and disaster mitigation of government.