改进神经回归算法的数据质量优化与预测

Data Quality Optimization and Prediction of Improved Neural Regression Algorithm

为了满足智慧交通管理系统向智能化方向发展,一般采用毫米波交通雷达对交通流进行实时并准确地检测。由于受到时间、天气和通信故障等因素的影响,以及雷达最大作用距离的限制,数据常常存在缺失异常等情况,从而对交通流变化规律分析带来很大的影响。在研究城市道路中交通流数据质量优化及交通流变化规律的问题时,对原始交通流数据分成三份,首先采用组合检验和阈值检验,非线性回归分析、线性插值、历史均值法、粒子群多层前馈(PSO-BP)神经网络算法,分别完成异常数据的识别、更正及部分缺失数据的填补,然后建立时间序列的交通流模型、交通流参数关系模型,根据交通流内在的变化规律建立改进后的神经网络回归预测模型,完成所有缺失数据的填补,最终获得完整的高质量数据。本算法的精确度可达到95.54%,能有效进行数据质量优化与预测。

In order to satisfy the development of intelligent traffic management systems in the direction of intelligence,millimeter-wave traffic radars are generally used to detect traffic flow in real time and accurately.Due to the influence of time,weather,communication failures and other factors,as well as the limitation of the maximum range of radar,there are often missing abnormalities in the data,which have a great impact on the analysis of traffic flow changes.When it comes to studying the problem of traffic flow data quality optimization and traffic flow change law in urban roads,the original traffic flow data was divided into three parts in this article.It could be firstly used combination test and threshold test,nonlinear regression analysis,linear interpolation,historical mean method,particle swarm optimization-back propagation(PSO-BP)neural network algorithm.It could respectively complete the identification and correction of abnormal data and the filling of some missing data,and then it could establish a time series traffic flow model and a traffic flow parameter relationship model,and establish an improved neural network regression according to the inherent change law of traffic flow Forecast model,complete the filling of all missing data,and finally obtain complete high-quality data.The result shows that the accuracy of this algorithm can reach 95.54%,which can effectively optimize and predict data quality.