APPLICATION OF THE BAYESIAN INFERENCE METHOD TO SYNTHESIZE URBAN DRIVING CYCLE SPEED SCHEDULES USING MEASURED DATA

Standardized driving cycles are widely used to simulate real-world vehicle operating conditions. They are used for estimating vehicle ranges, evaluating emissions, and designing powertrain characteristics. Recently, due to the intensive electrification of urban public transport fleets, estimating electric energy consumption has gained significant importance. This process requires up-to-date, validated driving cycles developed from realworld operating condition measurements. This paper presents an original method for synthesizing the driving cycle based on a stochastic approach using Bayesian inference. It first determines probability distributions of vehicle speed for the entire line. Next, all trips for the line are divided into segments of uninterrupted movement. A probability distribution of speed is also constructed for each of these segments. Then, quality fit indicators are calculated for each segment. Segments that support the hypothesis of the total distribution with maximum likelihood are then selected using Bayesian inference. A driving cycle for the selected line is synthesized by combining several segments with the highest degree of fit. The condition for applying the method is having a large dataset of measurements obtained during actual trips. The method can be effectively implemented computationally, ensuring high accuracy in reflecting the driving conditions in a selected area. The method has low computational requirements and a high capacity to adapt to changing operational conditions. This paper presents the results of applying this method to three bus lines in Warsaw and uses publicly available traffic data. The proposed method can be used for various applications, including estimating energy consumption for route travel, assessing the impact of traffic management changes on driving smoothness, and determining the load on electric bus engines and batteries.