Pavement damage detection system using big data analysis of multiple sensor

With the rise of intelligent cities, the artificial intelligence methods for road condition monitoring can benefit road information services of the smart city, making inspection and maintenance faster and the prediction of road condition more accurately. Damage to roads can lead to traffic congestion and even safety incidents, which can affect traffic management. In this paper, we present a new idea for monitoring road damage. We use the accelerometers in our mobile phones, and we use traditional accelerometers as well as cameras and sound recording equipment to capture different data. In this study, we attempted to use the STFT (Short-Time Fast Fourier Transform) algorithm commonly used in the field of audio recognition and the wavelet transform algorithm based on the Morlet parent function to process the acceleration signal to find the frequency of the corresponding damaged road feedback signal. In order to achieve effective detection of road surface cracks, we also explored the effect of window value on the results in the STFT algorithm. We use time and frequency domain features to detect road damages. The same method was used to process the audio signal, and preliminary results were obtained. Through signal analysis and processing with unfiltered signal data, the signal information of damage can still be found in the time-frequency diagram.