Improved HMM congestion based on multi-source data fusion evaluation model
To precisely evaluate the capacity of the road network and resolve issues with traffic congestion,in response to the intricate spatial and temporal correlation of traffic flows and their own uncertainties,an improved hidden Markov model that fuses multi-source data is proposed to evaluate the traffic congestion dynamics.Firstly,the traffic flow data are feature extracted and multi-input observation features are introduced to obtain the road characteristic state variables.Then,the state parameters of the road are determined while the traffic flow is divided into four states.Ultimately,an enhanced Welch's algorithm is used to estimate parameters and infer the state of the hidden Markov model,utilizing historical data for the first n moments to produce the optimized model.The location where a road segment located in Shenzhen city is used as an example to confirm the model's validity and application.According to the research's findings,the approach can reliably assess the condition of the road network and has a high accuracy of almost 97％,which is 4.7％percentage points better than the unimproved model.The difference in minimum accuracy between the baseline model and the improved model was 9.4％,9.7％,and 9.8％,respectively,with the change in sampling frequency.
traffic engineeringtraffic condition assessmentmulti-source datahidden Markov model
万方数据
维普
