改进的GA-BP神经网络模型在财产犯罪预测中的应用

发布时间：2018-05-30 16:21

本文选题：时空分析 + BP神经网络模型　；参考：《武汉大学学报(信息科学版)》2017年08期

【摘要】：发现犯罪时空分布规律并预测犯罪发生,是提高警务策略有效预防、控制犯罪的重要方法。在分析财产犯罪时空规律的基础上,利用BP神经网络模型自动学习训练各因子与财产犯罪的非线性关系,建立了财产犯罪预测模型。针对BP神经网络模型易陷入局部最优和模型不稳定的缺陷,提出了利用遗传算法(GA)选择各因子最优的初始化权重和参数,并以此作为BP神经网络模型的初始化权重矩阵,通过对历史数据的学习及训练建立了改进后的GA-BP神经网络模型。利用某市2007~2012年财产犯罪、人口、GDP、土地利用等35个综合影响因子数据,对改进前后的模型进行了预测对比试验。结果表明,改进后的GA-BP神经网络模型成功克服了BP模型的缺陷,收敛迭代最小次数从117次改进到8次;10次计算收敛迭代次数最大误差从370次提高到5次;模型预测精度(RMES)从0.043 0提高到0.019 95。
[Abstract]:It is an important method to improve the police strategy to prevent and control the crime by discovering the temporal and spatial distribution of the crime and predicting the occurrence of the crime. Based on the analysis of temporal and spatial laws of property crime, a property crime prediction model is established by using BP neural network model to automatically learn and train the nonlinear relationship between each factor and property crime. In view of the defects that BP neural network model is prone to fall into local optimum and the model is unstable, the genetic algorithm (GA) is used to select the optimal initialization weights and parameters of each factor, which is used as the initialization weight matrix of BP neural network model. An improved GA-BP neural network model is established by learning and training historical data. Based on the data of 35 comprehensive influencing factors, such as property crime, population GDPand land use in a certain city from 2007 to 2012, a comparative prediction experiment was carried out on the model before and after the improvement. The results show that the improved GA-BP neural network model successfully overcomes the defects of BP model and the minimum number of convergent iterations is improved from 117 times to 8 times from 10 times to 10 times. The prediction accuracy of the model is improved from 0.043 to 0.019 95.
【作者单位】：华南师范大学地理科学学院;
【基金】：国家自然科学基金(41171141)~~
【分类号】：D917

【参考文献】