Research on management and control strategy of E-bikes based on attribute reduction method

At present, the control strategy of urban mixed traffic is widely adopted by related departments. Some responsible researches shown that if pedestrians walk on the outermost lane, it would be beneficial to provide multiple channelised lanes for motor vehicles, and plan non-motorised roads for bicycle, motorcycles and electric bicycle. It is well known that traffic accidents often caused by these vehicles which were forced park on the motorised lanes because non-motorised vehicle lanes are often parked by motor vehicles. Hence, congestion is more likely to occur at peak periods; hidden danger increased greatly by traffic conflicts because the speed in both electric motor vehicles and non-motor vehicles are fast in flat peak and trough.

As some cities implement the ‘Motorcycle Ban’ policy in China, Traffic participant in the non-motorised lanes mainly are bicycle and electric bicycle, the E-bike is a weakly embedded alternative to mainstream auto mobility [1, 2]. On a substantial scale, E-bikes are not necessarily displacing cars, but are replacing walking, traditional bicycling, and the bus [3]. People use E-bikes to travel more than bicycle and most riders would travel by bus if E-bikes were unavailable [4]. Because the bicycles have low speed and flexible steering, the traffic conflicts often happen mainly between bicycles and the pedestrians [5, 6]. Electric bicycle can always be seen in traffic accidents in flat peak period and low peak period. Therefore, electric bicycle has become an unavoidable traffic participant. Electric bikes need to be considered as an important factor in urban traffic planning and traffic control strategy formulation.

As one of the most likely alternative vehicles for fuel vehicles, the research of E-bikes has attracted the attention of researchers worldwide. In general, the research of Asian countries (such as China) is mainly focused on safety, environment and selection patterns with E-bikes [7]. In contrast, in western countries, the main use of electric bicycles is the older people with higher income. The purpose of their use is entertainment and leisure. Most of the studies in Europe and North America focus on the impact of E-bikes on emerging markets, user behaviour, health and ageing [8, 9]. In recent years, for most urban decision-makers in China, pedalling models such as bicycles and E-bikes have been more and more attracted to people [10]. The E-bikes are electrically operated two-wheeled bicycles. Compared with traditional bicycles, people who need a long distance travel are more willing to choose it as their mode of travel [11]. In the field of urban transportation, compared with other modes of transport, E-bikes are more effective in reducing traffic problems, such as traffic jams, air pollution caused by traffic, lack of sports, excessive use of motor vehicles and so on [12]. According to ‘2017–2022 years China E-bike industry operation status and operation situation forecast report‘, by the end of 2014, there were more than 200 million E-bikes in China. According to ‘the analysis of the current situation of the market demand of the E-bike industry in China for 2018–2023 years and the Research Report on the prospect of investment development‘, in 2015 and 2016, the total output of E-bikes in China was 3080 and 32 million 150 thousand, respectively.

In Melbourne, Australia, 4,225 cyclists’ red-light crossing behaviour on videotape was used by Johnson, to analyse and summarise the factors affecting illegal crossing behaviour, including: driving direction, other cyclists, other vehicles and vehicle flow. Besides, for cyclists and pedestrians, personal characteristics such as gender and age are also the determinants of red-light crossing behaviour [13]. Field observation method was applied by Phillips et al., to study the factors and mitigation strategies about intersection conflicts between vehicles and bicycle [14].

In the past, the causes of traffic jams and accidents were often explained and described based on people's subjective cognition and judgement. These deficiencies will affect the evaluation based on traffic accidents [15]. With the development of computer video recognition technology, the video detection technology is used to automatically identify the traffic conflict data in the video [15]. Since videos are recorded at certain angles, it is necessary to convert the distance through the mathematical formula using the field reference to achieve more accurate data [16]. There have been studies on automatic conflict detection by video recognition after taking video by UAV flying high above the research object [17]. Some scholars believe that the most important thing in traffic safety research is to prevent accidents rather than to predict accidents. Traffic collision technology can be used as a tool to diagnose and evaluate road traffic congestion and safety [18]. The traffic conflict prediction models widely used at home and abroad are established according to the traffic accident prediction model, including regression model, generalised linear model (Poisson distribution, negative binomial distribution, zero pile-up probability distribution), BP neural network, time series, probability model etc. Traffic conflict prediction models mainly focus on the number of conflicts, and the severity of conflicts is less. The accident prediction models based on traffic conflict mainly include regression model, causal model and probability model of trajectory based on video [19,20,21,22].

There have also been some studies on participants in multiple traffic conflicts [23, 24]. For vehicles running in non-free flow, when a conflict occurs between a pair of participants, the driver immediately takes the avoidance behaviour, resulting in the change of vehicle movement state, which is likely to affect the adjacent vehicles in the lane and other lanes, causing other vehicles to have traffic conflicts, which is a ‘regional chain reaction‘. Expansion of the traditional pair of participants in traffic conflict to a larger scope and the mechanism of interaction among more participants may provide a basic model for the regional Internet of Vehicles (ROV) early warning system [25].

On the basis of the above research, this paper plans to analyse the types of human-vehicle conflicts involving E-bikes, and study the judgement criteria of traffic conflicts involving E-bikes based on time interval. Based on the field survey data, a statistical table summarising the number of conflicts under various road traffic conditions was formed. Then, the redundancy attribute reduction method based on rough set theory is used to analyse the main influencing factors of traffic conflicts of E-bikes under the traffic conditions. It provides a scientific decision-making basis for the development of traffic conflict control strategies.

A decision-making table is established to reflect the influencing factors of road traffic conflicts and the amount of conflicts. The attribute reduction method of rough set theory is used to remove redundant rules in decision tables, and the main factors causing traffic conflicts in the concerned road section are found. This method can effectively analyse the important factors of mitigating traffic conflicts, and provide a scientific decision-making basis for formulating effective control strategies and facility planning schemes to reduce road traffic conflicts.

Conflicts between E-bikes and motorists and pedestrians increase the chances of causing congestion and increasing traffic safety risks. This paper analyses the types of human-vehicle conflicts involving electric vehicles and puts forward a conflict determination method for designing E-bikes in mixed traffic. Then, in terms of traffic control strategy, the relationship decision table reflecting the influencing factors of road traffic conflicts and the number of conflicts was established. Based on the attribute reduction method of rough set theory, the redundant rules were removed from the decision table, so as to find the main influencing factors of road traffic conflicts. The results show that this algorithm can effectively analyse the important factors for mitigating traffic conflicts, and provide a scientific decision-making basis for making effective control strategies and facility planning schemes to reduce the amount of traffic conflicts.