Recently there have been advances in the development of a passive structural neural system (SNS) for damage detection. The SNS uses electronic logic circuits to mimic the signal processing in the biological neural system. The advantage of the SNS is that highly distributed continuous sensors provide high sensitivity to damage, and the biomimetic signal processing and passive sensing tremendously simplify the instrumentation and wiring of the monitoring system. Also, the SNS operates continuously during operation of the structure to detect ambient Lamb waves or bulk waves that are produced by cracking, delamination, bearing damage, rotor imbalance, flow instabilities, impacts, or other material failure modes.
In this paper, asymmetric Lamb wave propagation representing acoustic emissions (AE) is modelled based on a superposition of plate bending vibration modes. The simulation demonstrates that the SNS with four channels of data acquisition can localize damage within a grid of sensors irrespective of the number of sensors in the network. To experimentally validate the analysis results, a two-neuron prototype of the SNS was built and tested using a simulated AE source (a pencil lead break) on a riveted aluminium joint and on a composite plate. In both experiments, the SNS was able to localize simulated damages. These results indicate the feasibility of expanding the SNS to a large number of neurons.