رئوس مطالب
- چکیده
- کلیدواژه ها
- 1. مقدمه
- 2. روششناسی
- 2.1. بررسی مدل
- 2.2. مدل پیشبینی مبتنی بر KNN
- 3. مطالعه موردی
- 3.1. جمعآوری دادهها
- 3.2. نتایج و تجزیهوتحلیل
- 4. نتیجهگیری
- منابع
Abstract
The objective of this research is to develop a dynamic model to predict bus dwell time at downstream stops. The research also intends to test the proposed model using real-world data. This model is based on k-Nearest Neighbour (KNN) algorithm using history and current data collected by GPS (Global Positon System) fixed on buses. In the research, the data of buses of No.B1 line of Changzhou in China is used. In the test with real-world data, the proposed bus dwell time prediction model performed effectively both on accuracy and calculating speed.
Conclusions
The proposed prediction model based on KNN can predict bus dwell time at downstream stations using history bus GPS data of same time. The proposed model can be used in practice without need of adjustment according to bus style, stop form and also without need of prediction the number of passengers will on and down.
The test results shown that compared with the predict method based on average dwell time and the method based on KNN using the total data rather than just using same type data, the proposed method performs better. The processing error of bus dwell time may influence the prediction accuracy. For example some bus stations are close to crossroads, so the time of waiting for traffic light may considered as dwell time. If sensors can be fixed at bus station in future to collect bus dwell time instead of GPS, then the proposed model will perform better.