Multi-mode Trip Information Recognition Based on Wavelet Transform

1 GPS-based travel survey is an emerging data collection method in transportation planning. 2 Its application in trip mode detection has been explored in many existing studies. 3 However, most existing research on GPS data based trip mode detection methods are 4 developed and tested with data collected from European and American countries. Their 5 methods cannot be easily adapted to Asian countries such as China, India, and Japan with 6 much higher population density, the complex road network, and the highly-mixed travel 7 modes during daily commuting. Furthermore, when conducting trip segment division in a 8 multi-mode travel, the existing algorithms use travel time and distance thresholds which 9 are highly dependent on the local travel behavior and lack universality across different 10 traffic environment. This paper proposes an innovative framework to detect trip modes 11 under complex urban environments. First, a smartphone application, named GPSurvey, is 12 developed to collect passive GPS trace data. Then the Wavelet Transform Modulus 13 Maximum (WTMM) algorithm is developed for trip segment division. WTMM has 14 outstanding capabilities of identifying singularity features of a signal which suits the task 15 of detecting mode changes in complex traffic environment. A neural network (NN) 16 module is further developed for mode detection based on cell phone GPS location and 17 acceleration data. Results indicate that the proposed method has a promising performance. 18 The average absolute detection error of mode transfer time is within one minute, and the 19 accuracy for detecting all modes is above 85%. 20 21