Container integrity and condition monitoring using RF vibration sensor tags

Global supply chain operations use hundreds of thousands of container trucks to transport valuable packages and items within and across continents. The recent developments in wireless sensing, including in the so-called RFID sensor tags, offer significant potential for continuous long-range integrity monitoring of containers at various phases of their shipment. In particular, the vibration patterns of a container and its contents can reveal significant information related to its operating environment and integrity during transport, handling and storage. The present work investigates the sensitivity of the patterns of container vibrations gathered as signals from wireless sensor tags to the following four major factors that define the container operating conditions: terrain type, speed of the vehicle, weight dimensions of the container items. Quantitative relationships are established between the vibration patterns and the salient factors through a series of experiments involving the use of a scaled model of a container truck and a wireless sensor that captures vibrations at 200Hz sampling rates. The idea is to classify the operating conditions by analyzing the complex dynamics underlying vibration signals. Using nonlinear dynamic characterization, we find that the Lyapunov exponents of dynamics underlying signals are positive (0.01-0.02 for Stage 1 and 0.005 for Stage 2 experiments). The statistical and nonlinear dynamic features together are successfully mapped using a neural network to classify between the different operating conditions. The neural networks were able to identify the correct operating conditions from the vibration sensor features about 90% of the times. In real life, the research results can be applied to accurately capture the environmental and operating conditions during container transport. This will help proactively address possible serious integrity losses.