An Artificial Neural Receptor System for Structural Health Monitoring

The artificial neural receptor system is a new approach that greatly simplifies data acquisition, and hence may enable practical Structural Health Monitoring for large structures. The benefits of the system involve reducing data acquisition channels while maintaining the ability to extract substantial structural response information. The approach presented uses an array of piezoelectric sensors wired to mimic the basic receptor connectivity of the biological nervous system. The method of solving for array outputs from individual sensor strains is demonstrated using a 3-by-3 sensor array. For an N-by-N array of sensors, the number of channels of data acquisition is reduced from N ² in a conventional system to 2N in the artificial neural receptor system. The first few microseconds of signal output from the array rows and columns can determine the location of the excitation. An artificial neural network analysis can be used to extract excitation locations and individual sensor strains from the array. Acoustic emissions and dynamic strains are measured by the array.