An Intelligent Approach for Asphalt Pavement Density Measurement Based on Artificial Neural Network and Principal Component Analysis

The coring method (CM) and the electrical-device-based method are usually applied to measure the density of the asphalt pavement. The CM is an accurate but inefficient method. A pavement quality indicator (PQI) is an electrical device that can measure the density efficiently but not accurately. The reason is that its accuracy is affected by the temperature and humidity of the asphalt pavement. In our previous work, ANN models have been proposed to improve the accuracy of a PQI. However, the effects of temperature and humidity have not yet been quantified. In this paper, a new approach, principal component analysis (PCA), is introduced to solve this problem. It can be used to quantify the principal component by calculating the covariance matrix and its eigenvalues. Additionally, the PCA model can also improve the accuracy of the artificial neural network (ANN) model in predicting the density of the asphalt pavement. Hence, a novel intelligent approach based on the combination of ANN and PCA is proposed in this paper. The input data include temperature, humidity, and the density measured by a PQI. They are first input into the PCA part of the model. Then the output data of the PCA part are combined with the mentioned input data and used to train the ANN part of the model. Finally, the accuracy of the trained ANN-PCA model is compared with that of the trained ANN model (without the PCA part). The results show that the ANN-PCA model performs better than the ANN model in terms of accuracy. In addition, the ANN-PCA model becomes more accurate if the model only contains a small number of neurons.