多视域下交通管理数据应用成效评估

Effectiveness Evaluation of Traffic Management Data Application in Multiview

摘要:

交通管理系统平台在全国各地已经形成了规模化的应用布局。为全面评估交管数据应用成效,规避评价方案视域单一、不能客观系统的反映数据使用效果等问题,针对多视域下交管数据应用成效评估开展研究并设计评估方案。首先,提出多视域评估原则,建立包括技术视域下的数据质量、应用视域下的数据使用、绩效视域下的交管业务、公共视域下的社会效益等准则在内的准则层。其次,研究确定支撑各准则的具体指标。针对交管业务准则及社会效益准则,采用灰度关联分析方法(GRA)对某市交管部门绩效与具体业务指标排名数据进行相关性分析,选取高关联因素指标群并依据属性相近原则分类聚集形成评估指标。构造统一量纲的评估指标表达式,兼顾准确性和高效性。最后完成基于层次分析(AHP)的准则层权重分配。考虑分视域下指标赋权可信度问题,引入改进序关系分析法(iG1)进行分视域一次赋权,提出专业校正赋权法(PAW)对一次赋权结果进行校正,保证指标权重客观公正。算例验证显示,基于灰度关联分析(GRA)指标排序的评估可信度(38.9%)、基于序关系分析(G1)指标排序的评估可信度(30.95%)与基于iG1-PAW组合赋权策略指标排序的评估可信度(61.875%)有较明显差别。

Abstract:

A scaled distribution of traffic management platform has been accomplished across the country with the lack of effectiveness evaluation. Single-view evaluation couldn’t objectively reflect effectiveness of the data application. Focusing on effectiveness evaluation of data application in traffic management under the Multiview, evaluation scheme was designed as below. First, Multiview principle was set for establishing the criterion layer, comprising Data quality in the technical perspective,Data application in the utilization perspective, Traffic management business in the Performance perspective, Social benefit in the public perspective. Second, the specific indexes were designed to support the criteria. Contraposing traffic management criterion and social benefit criterion, grey relationship analysis (GRA) was adopted to analyze the correlation between performance index and business indexes about traffic management department in certain city. Then the indexes of high correlation were collected to merge into a classic index according to principle of close attribution. Index expression of unified dimension was constructed to show conciseness and efficiency. At last, weighting task was finished based on analytic hierarchy process (AHP). Considering the confidence level of weighting on indexes under the Multiview, the improved G1 method was adopted to weight indexes in every criterion, while the professional adjustment weighting (PAW) was introduced to adjust the first weighting to ensure the fairness and objectiveness. Illustrated from an example, result is obvious that the confidence degree of GRA, G1 and iG1-PAW respectively are 38.9%,30.95% and 61.875%.