In this paper, a new four-electrode sensor for water conductivity measurements is presented. In addition to the sensor itself, all signal conditioning is implemented together with signal processing of the sensor outputs to determine the water conductivity. The sensor is designed for conductivity measurements in the range from 50 mS/m up to 5 S/m through the correct placement of the four electrodes inside the tube where the water flows. The implemented prototype is capable of supplying the sensor with the necessary current at the measurement frequency, acquiring the sine signals across the voltage electrodes of the sensor and across a sampling impedance to determine the current. A temperature sensor is also included in the system to measure the water temperature and, thus, compensate the water-conductivity temperature dependence. The main advantages of the proposed conductivity sensor include a wide measurement range, an intrinsic capability to minimize errors caused by fouling and polarization effects, and an automatic compensation of conductivity measurements caused by temperature variations.

This paper uses a linear transformer model to investigate the effect of the lift-off on the results of the thickness measurement of nonferromagnetic metallic plates. The transformer model previews that the time derivative of the magnetization curves obtained for different gaps between the excitation coil and the plate should intercept in a single point when low magnetic coupling factors are considered. To assess the validity of the model, results are compared with experimental data obtained with a giant magnetoresistive sensor probe. For comparison, the sensor output voltage time derivative must be performed as well. The similarity of the theoretical model results and those obtain experimentally with pulsed excitation, confirms the correctness of the transformer approach.

In this work, classification of cracks in a non-ferromagnetic material (aluminum 1050) is presented using two supervised machine learning techniques called support vector machines (SVM) and decision tree. The classification performances between the two classifiers were evaluated using confusion matrices and miss-classification errors. In order to improve the classification performance, a multi-level SVM model is proposed wherein each level implements SVM using different features and kernels. The experimental tests were performed in two aluminum plates containing 18 defects with different depths and lengths. The 18 defects were inspected in the surface and sub-surface locations, resulting in a study of 36 different crack cases. The experimental data used to extract features for supervised machine learning were obtained using an eddy current testing (ECT) probe that includes a planar coil and a tunnel magnetoresistance (TMR) sensor. The probe was excited by a multi-frequency signal with four testing frequencies to evaluate the dimensions and the locations of the defects. Due to the experimental noise existing in the measurement signals, it is important to pre-process the signals before feature extraction. Signal processing techniques (such as moving average filter, spline interpolation and symmetry corrections) were used before extracting the features for classification.