| The problem of population aging has become a social hotspot.General Secretary Xi Jinping emphasized in the 32nd collective study of the Political Bureau of the CPC Central Committee:adhere to the combination of meeting the needs of the elderly and solving the problem of population aging,and strive to meet the growing material culture of the elderly.To meet the needs of the elderly,promote the comprehensive,coordinated and sustainable development of the cause of aging,strengthen scientific research on aging,meet the various needs of a large number of elderly people,and properly solve the social problems brought about by population aging.With the advent of an aging society,the traffic participation of the elderly has been further enhanced,and the problem of street safety has become increasingly serious.However,traffic safety research is currently mainly focused on young people and drivers,and the research on elderly pedestrians is very limited;in addition,in the construction of accident severity prediction models,most studies select accident factors and construct accident network structures for application Most of the methods are based on experience judgment and expert scoring,which is highly subjective and involves too many accident factors,making it difficult to extract crowd characteristics and key scenarios.In response to the above problems,this thesis extracts 18,176 elderly pedestrian accident data based on 87,829 pedestrian accident data in Hong Kong from 1998 to 2017,and then divides it according to time after preprocessing.5011 accident data were input into subsequent research as training set and test set,the former was used for model construction and learning,and the latter was used for model evaluation and verification.The main research contents of this thesis are:1)Use factor analysis to calculate the weight of accident factors in the sample.The KMO test and the Bartlett sphericity test were used to verify that the training set data were suitable for factor analysis.In order to obtain the degree of influence of each factor on the traffic accident of elderly pedestrians,the accident data was standardized and then the factor analysis method was used to calculate the influence weight value of each factor..2)Use the Bayesian method to build a prediction model for the severity of elderly pedestrian traffic accidents.The 19 accident influencing factors extracted by factor analysis,corresponding to their accident severity,are input into the learning process of the Bayesian network as training set data.s network structure.Then the obtained Bayesian network structure is visualized in Netica software,and its parameters are learned to obtain the conditional probability distribution of each node variable.The final prediction model is tested by using the test set of elderly pedestrian traffic accident data,and the prediction accuracy of the model is obtained.3)Apply the model for sensitivity analysis and key causal chain reasoning.The constructed model is used for further analysis,sensitivity analysis and key causal chain reasoning are carried out on the accident severity nodes,and the single influencing factors and multi-factor combination chains that lead to higher accident severity are obtained,and effective safety prevention suggestions are put forward accordingly.Based on a large number of long-term real accident data,this paper uses objective data analysis methods to construct a Bayesian network model of the severity of traffic accidents for elderly pedestrians,and uses the test set data to test the prediction accuracy.The prediction error is 15.29%,while the usual prediction If the accuracy rate is above 80%,the model prediction result is considered to be very good.Using the constructed model to further analyze the degree of traffic accidents among elderly pedestrians,the key nodes and multi-factor combination causal chains leading to higher accident severity are obtained,and targeted improvement countermeasures are proposed to improve the travel safety of elderly groups. |
PDF Full Text Request