Four-electrode Based Nondestructive Test Technology For Cfrp And Sensor Optimization

The non-destructive testing using four-electrode methods for carbon fiber reinforced polymer(CFRP)that based on electrical impedance detection,is a new technology with several advantages of non-radiation,rapid-response,and low-cost.That brings it a huge development prospect as being able to detect a large area of carbon fiber composites quickly and accurately.Considering that changes in current which are caused by centre-rotating of measuring electrode or changing of carbon fiber laying direction,can fail the detection with the scanning.Thus,the testing processes need to be simulated and analyzed in order to ensure the accuracy and effectiveness of four-electrode sensor detection.The main contents include researches in the principle of four-electrode detection,optimization of sensor structure and excitation methods,as well as defect identification of neural network.The main works and results are as follow:1.Used the finite element simulation software to establish a four-electrode models on detecting carbon fiber composites.By obtaining the fitting function of the electrode rotated-angle and the output current after the electrode rotation,the sum of the absolute error is proposed as an indicator to evaluate whether the material is defective or not.In addition,the author built several damage models to verify the effectiveness of the proposed method.The results showed that this method was being able to effectively identify most of the damage,but the effect of deep layer damages recognition were not good.2.Based on the four-electrode CFRP detection equivalent circuit and finite element simulation calculations,sensors with different electrode radius structures and different excitation methods were designed,Optimized indicators for the sensor structure parameters were proposed,and the optimization results were obtained.3.The artificial neural network technology in pattern recognition technology was utilized to classify the collected simulation data,as well as the genetic algorithm was used to optimize the connection weights and thresholds of the neural network.Finally,the purpose of classifying material defects was achieved,and the recognition error was not greater than 8.81%.It has guided significance of the prediction of future detection resultsFinally,the author puts forward research directions and improvement suggestions for the future research on four-electrode sensors.