Determination Of Moisture Content Of Subgrade Soil Using Artificial Neuro-electronic Control

Moisture content of soil is of utmost importance in the progression of road construction and is absolutely crucial for making decisions concerning design and construction of pavements. Pavement sustainability depends on the performance of its individual components, which require the assurance of qualities such as evaluation of the properties of soil and constant monitoring of some soil conditions (e.g. moisture content). For pavement design the most considerable engineering factor of soil is the strength that is practically accomplished by soil compaction and soil stabilization. However, to achieve the required soil strength, consideration of compaction effort or stabilization ingredients should be precise, and it is inevitably related to the moisture content of soil. Hence moisture content determination is elemental and must be performed frequently as necessary. The conventional method for its determination involves oven drying and this endeavour is time consuming (requires approximately 24 hours for drying), which may affect the subsequent undertakings. Some microwave oven based fast methods have been realized recently but these require continuous manual interventions. In this paper, a new approach will be proposed in a view to suppress the limitations of existing methods while maintaining the better accuracy. This innovation embeds an automatic electronic control as well as an artificial neural network (ANN) in the framework for time optimization. Artificial neural network and automatic electronic control both together can be termed as artificial neuro-electronic control. The artificial neural network has been optimized and trained by mapping the weights of soil samples at specific time steps to the respective final moisture contents. As a result, subsequently the system can be able to predict the final moisture content by analysing fewer data samples in the very beginning of moisture content determination tests. Validation of the predictive results has also been conducted in real time for soil samples suitable for subgrade layer of a pavement to ensure the system feasibility for laboratory and field uses. Experiments show that this fully automatic system can exhibit a significant accuracy and precision for the evaluation of moisture content in about 50% reduced time compared to the standard microwave based method.