Research On The Damage Identification Of The Piezoelectric Smart Laminated Structures

Delamination of composite laminates caused by impact would easily lead to the decline ofmechanical properties, which results in the potential sudden and catastrophic danger. Therefore, basedon background of piezoelectric smart structural health monitoring using in fields of aerospace, theresearch on methods of damage identification and evaluation of composite laminates has importantvalue in engineering application. Paper using ANSYS establish the finite element damage model ofsmart piezoelectric laminates, which use the decrease of elastic modulus to simulate the structuraldamage, establish the single and multi-point damage from10%to70%degree.The methods of thedamage identification research of this paper divides into modal parameter identification based onmodal strain and pattern recognition based on BP neural network, the main research contents are asfollows:(1) Research the modal parameter damage identification methods of composite laminates. Bycomparing the modal frequency, modal displacement and modal strain mode of the damage and nodamage structures, paper draw out the conclusion: strain mode of piezoelectric smart laminates ismore sensitive and more reliable to damage identification. In order to study the damage location anddamage degree diagnosis, paper also compares the change of strain mode,change rate of strain modeand the differential of strain mode, the result conclude: both of change of strain mode and change rateof strain mode can identify well of damage location of a single point, but to multi-point damage, rateof change of strain mode identification is better, and with the higher frequency modes, identificationusing change of strain mode is more difficult; the differential of strain mode just can identify a singlepoint of damage without the no damage data, and the identification of multi-point damage is moredifficult.(2) Research pattern recognition based on BP neural network damage detection system to realizethe damage location and extent of online identification. As strain mode to be the input of network, andthe damage location and the degree damage directly to be the output of network, network training datafrom the same composite laminate of numerical simulation. The results show that: using strain modeas the input vector and LM optimization algorithm, BP neural network can be obtained for a betteraccuracy methoed of composite laminates damage identification, and it is very effective.