基于博弈论综合赋权改进的交通事故组合预测模型研究  

Research on traffic accident combination prediction model based on game theory comprehensive empowerment improvement

作　　者：单小曼 刘尊青 辛宁 侯金超 姚亮 王逸飞 SHAN Xiaoman;LIU Zunqing;XIN Ning;HOU Jinchao;YAO Liang;WANG Yifei(School of Transportation and Logistics Engineering,Xinjiang Agricultural University,Urumqi 830052,China;Traffic Police Headquarters,Xinjiang Uygur Autonomous Region Public Security Bureau,Urumqi 830011,China)

机构地区：[1]新疆农业大学交通与物流工程学院,新疆乌鲁木齐830052 [2]新疆维吾尔自治区公安厅交警总队,新疆乌鲁木齐830011

出　　处：《现代电子技术》2025年第5期120-126,共7页Modern Electronics Technique

基　　金：国家自然科学基金项目(52268072);新疆农业大学校级重点学科项目(XJAUTE2022G09);新疆农业大学科研项目(2523HXKT3)。

摘　　要：为改善交通安全,提升交通事故预测模型的精度。基于熵权法、反熵权法、CRITIC法和变异系数法进行权重计算,继而通过等权法、乘积合成法和博弈论思想进行综合赋权,对组合模型进行改进。研究结果表明:基于博弈论综合赋权改进的GM(1,1)-指数平滑-BP组合预测模型的预测效果最佳,其决定系数(R^(2))高达0.994,平均绝对误差百分比(MAPE)仅为1.426%,较单项模型和直接组合模型预测程度更好,误差率更低。基于博弈论思想离差最小化求解最优权重组合系数,能够有效结合不同赋权方法的优势,权衡各方法之间的最优解,从而避免模型出现过度拟合、预测误差大及缺乏动态适应性等问题。同时,组合预测模型以数据最大优化为核心原则,可有效克服单项预测模型的局限,使其能更好地适应复杂变化的环境及数据,显著提升了预测的准确性和稳定性,增强了模型的鲁棒性和适应性。This paper strives to improve traffic safety and enhance the accuracy of traffic accident prediction model.The weight is calculated on the basis of entropy weight method(EWM),anti-entropy weight method,CRITIC(criteria importance through intercriteria correlation)method and coefficient of variation method,and then the combination model is improved by comprehensive empowerment with the ideas of equal weight method,product synthesis method and game theory.The results of the study show that the GM(1,1)-exponential smoothing-BP combination prediction model,which is improved based on the comprehensive empowerment of game theory,has the best prediction effect.Its coefficient of determination(R^(2))is as high as 0.994,and its mean absolute percentage error(MAPE)is only 1.426%,which is a higher accuracy of prediction and lower error rate than those of the single model and the direct combination model.The optimal weight combination coefficients are solved based on the game theory idea of minimization of deviation,which can effectively combine the advantages of different empowerment methods and weigh the optimal solutions among the methods,so as to avoid model overfitting,large prediction error and lack of dynamic adaptability.In addition,the combination prediction model is based on the idea of maximum optimization of data,so it can overcome the limitations of the single prediction model effectively,and better adapt to the complex changes in the environment and data,thereby improving the accuracy and stability of the prediction significantly.To sum up,the robustness and adaptability of the model are enhanced.

关 键 词：交通安全 交通事故 博弈论 GM(1 1)-指数平滑-BP 赋权改进 组合预测模型 

分 类 号：TN919-34[电子电信—通信与信息系统] U491.31[电子电信—信息与通信工程]