This paper proposes a multi-task learning system for industrial defect detection, focusing on surface scratches on titanium gaskets as the detection target. Our model is built using a deep convolutional neural network (CNN). As defects are infrequent in actual production lines, we adopt the Faster-RCNN object detection network architecture and integrate it with a multi-task learning system. A classification task model is introduced into the backbone network to filter out a significant number of defect-free images. This step reduces the computation and data transmission time of the subsequent target detection task, accelerating the detection process. Experimental results show that our proposed method reduces inference time by 37.6% and 42.46% in actual production lines with defect rates of 12% and 5%, respectively, while maintaining 96% of the original model's performance.

Predictive maintenance has become increasingly important in modern industry as it can prevent unplanned production interruptions and reduce maintenance costs. This is especially important for companies working with expensive and complex machinery, such as those used in manufacturing, transportation, and energy. Predictive maintenance uses sensors and data analytics to monitor equipment performance and predict when a failure or breakdown might occur. This paper presents a preliminary study on the use of absorbed currents measurement of a three-phase induction motor aimed at electrical fault diagnostics. In detail, a 4-pole high efficiency induction motor with 3kW of rated power was simulated in the Ansys Electronic Desktop environment. The simulated operation is with direct mains power supply. The motor was simulated initially at no load, then load torque values of 20%, 40%, 60%, 80% and 100% of rated value were evaluated. The currents drawn on the three phases were evaluated with a computational step of 500 µs, corresponding to a sampling rate of 2 kHz. The same simulations were subsequently repeated by reproducing five different fault conditions in the motor. Therefore, an analysis of the waveforms of the absorbed currents is provided and a diagnostic system based on their analysis is proposed.

Arterial simulators are useful tools to reproduce the Pulse Wave Velocity (PWV) behavior depending on vessel characteristics. This quantity is related to the Pulse Transit Time (PTT), i.e., the time interval required for the pulse wave to travel between two sites in a vessel. In the literature, there is a lack of comparison of PTT evaluation methods from signals acquired through arterial simulators. In the present study, three PTT estimation methods (peak-to-peak, tangent-secant and cross-correlation) have been applied to two signals simulating the pressure wave traveling in an Arterial Surrogate (AS) over time. Tests have been repeated for different imposed delays between the generated waveforms. From the obtained results, the cross-correlation method showed the lowest discrepancy values between estimated and imposed time delay

New magnetic sensor systems like stray field compensating specimen demand new challenges for the design of respective permanent magnets and sensors themselves. The collaboration of FEM simulations with 3D-Hall sensor measurements provides new insights how to build and optimize such systems. The paper shows this by the example of two stray field compensating sensors for position detection of rotating systems. The first example provides different stages of the development of a new sensor. The second one shows steps during the improvement of permanent magnets for a commercially available sensor system. Both parts also supply results about the insensitivity of those sensors against external fields up to 20mT.

Pulse Wave Velocity (PWV) monitoring is a well-established method for both the diagnosis and prevention of cardiovascular diseases. Over the years, many arterial simulators have been made to study PWV dependency on the mechanical and geometrical characteristics of arterial surrogates; in particular, some of them are able to vary the tensional state of vessel walls to cause a change in PWV values. In this paper, a novel arterial simulator is presented, capable of varying the PWV of an arterial surrogate by acting independently on its inner and outer pressures. The simulator under assessment can manage pressures up to three times greater than the atmospheric one. This characteristic provides the possibility to explore many tensional states compared to its predecessors. The experimental outcomes confirm its working principle and an increasing trend in PWV is found. The result uncertainties are up to 10% of the corresponding PWV value. I