Edge computing-based method for fine determination of urban traffic state

An edge computing-based method for fine determination of an urban traffic state, comprising the following steps:

• step 1:
• preprocessing lane-level multi-source data collected by edge nodes, and inputting preprocessed data into a database as input data for a subsequent step;
• step 2:
• dividing a complete road segment between intersections into an approach region and a remaining road segment region;
• step 3:
• computing an average delay per vehicle of each lane by using a computing capability of the edge nodes based on a signal cycle of a current intersection;
• step 4:
• determining input values and output values of a fuzzy logic model, and levels of the input values and levels of the output values, wherein the levels of the input values and the levels of the output values represent different traffic states, inputting the average delay per vehicle of the each lane obtained in step 3 into the fuzzy logic model to perform a computation, and outputting a traffic state of the approach region, wherein a method for computing the traffic state of the approach region comprises the following steps:
• step 4.1:
• performing a cluster analysis on the input values by K-Means clustering based on historical data, to obtain classification indexes of the input values, wherein the input values are historical average delays per vehicle of the each lane, and the classification indexes are levels representing different traffic states;
• step 4.2:
• determining classification indexes of the output values of the fuzzy logic model based on an actual requirement;
• step 4.3:
• determining a membership function for the input values, the output values, the levels of the input values and the levels of the input values, wherein the levels of the input values and the levels of the input values represent the different traffic states;
• step 4.4:
• designing a fuzzy rule between the input values and the output values;
• and
• step 4.5:
• converting a real-time input value into an output value by a preset fuzzy logic inference method, wherein the output value is an eigenvalue of the traffic state, and obtaining the traffic state of the approach region corresponding to the eigenvalue;
• and
• step 5:
• computing a traffic state of a remaining road segment based on real-time data preprocessed in step 1, wherein a method for computing the traffic state of the remaining road segment comprises the following steps:
• step 5.1:
• establishing a matching relationship between an upstream cross-section and a downstream cross-section, linking intersection information, and obtaining information of vehicles passing the upstream cross-section and the downstream cross-section successively;
• step 5.2:
• computing travel times of the vehicles meeting a requirement in step 5.1 within a signal cycle of a downstream intersection to constitute a set, and computing an average travel time;
• step 5.3:
• computing an approach-level traffic parameter and an intersection determination index based on computation results obtained in steps 1 and 3;
• step 5.4:
• based on computation results obtained in step 5.3, computing an average travel time of vehicles passing the approach region within the signal cycle of the downstream intersection;
• and
• step 5.5:
• based on a computation result obtained in step 5.4, computing an average travel time of vehicles passing the remaining road segment within the signal cycle of the downstream intersection, computing an average travel speed based on a length of the remaining road segment, and obtaining the traffic state corresponding to the remaining road segment.