Desenvolvimento de algoritmos para estimação do deslocamento de transdutor ultrassônico em relação ao objeto inspecionado

Abstract

Through the application of Non-Destructive Evaluation (NDE), it becomes feasible to conduct inspections without causing damage to the object under assessment, thus preserving its physical integrity. Nowadays, the most widely used technique by the industry is ultrasonic inspection due to its ease of implementation, the ability to penetrate deep within the parts, and its efficiency in transmitting information about discontinuities in the inspected object. However, this technique has some disadvantages. To measure the transducer’s displacement during inspections, encoders are used, and since they are mechanical sensors, they can introduce measurement errors due to device slippage with the part. Furthermore, their operation requires cables to transmit the collected information, which can limit the movement of the transducers during a scan. Because of this, this work proposes the use of GME algorithms on reconstructed ultrasound image sequences. Three algorithms were implemented on different test specimens, and their performances were evaluated for transducer translation and rotation. Simulation data was acquired using CIVA software, and tests with real data were conducted using the M2M Panther system. The results obtained show that the Wavelets algorithm yields good results for both translation and rotation within certain displacement ranges. On the other hand, the Features and Keypoints algorithms did not perform well with ultrasound images for either type of motion.