Machine Learning Algorithms to Detect Penetrations of PVDF

Polyvinylidene fluoride (PVDF) has widely used in detecting the interplanetary dust. In the case of penetration and non-penetration, the output signals of the PVDF are quite different. Detecting whether particles penetrate PVDF is a crucial issue. We create a set of experimental equipment for collecting the signals from the PVDF. The equipment consists of particle emitter, shield, conditioning circuits and data acquisition equipment. 600 experiments are conducted. Among 200 experiments, the particles penetrate PVDF. We successfully distinguish penetration of PVDF using four machine learning algorithms: Anomaly Detection (AD), Artificial Neural Network (ANN), K-Nearest-Neighbors (KNN), and Support Vector Machines (SVM). We propose a unique evaluation criteria OP to evaluate the performance of four classifiers including their accuracy and computational time. The results show that ANN is the best machine learning algorithm for our problem, and AD is not suitable for our problem.