Metallic surface defect detection is an essential and necessary process to control the qualities of industrial products. However, due to the limited data scale and defect categories, existing defect datasets are generally unavailable for the deployment of the detection model. To address this problem, we contribute a new dataset called GC10-DET for large-scale metallic surface defect detection. The GC10-DET dataset has great challenges on defect categories, image number, and data scale. Besides, traditional detection approaches are poor in both efficiency and accuracy for the complex real-world environment. Thus, we also propose a novel end-to-end defect detection network (EDDN) based on the Single Shot MultiBox Detector. The EDDN model can deal with defects with different scales. Furthermore, a hard negative mining method is designed to alleviate the problem of data imbalance, while some data augmentation methods are adopted to enrich the training data for the expensive data collection problem. Finally, the extensive experiments on two datasets demonstrate that the proposed method is robust and can meet accuracy requirements for metallic defect detection.

Most of the current object detection approaches deliver competitive results with an assumption that a large number of labeled data are generally available and can be fed into a deep network at once. However, due to expensive labeling efforts, it is difficult to deploy the object detection systems into more complex and challenging real-world environments, especially for defect detection in real industries. In order to reduce the labeling efforts, this study proposes an active learning framework for defect detection. First, an Uncertainty Sampling is proposed to produce the candidate list for annotation. Uncertain images can provide more informative knowledge for the learning process. Then, an Average Margin method is designed to set the sampling scale for each defect category. In addition, an iterative pattern of training and selection is adopted to train an effective detection model. Extensive experiments demonstrate that the proposed method can render the required performance with fewer labeled data.

Rotating blade vibration measurements are very important for any turbomachinery research and development program. The blade tip timing (BTT) technique uses the time of arrival (ToA) of the blade tip passing the casing mounted probes to give the blade vibration. As a non-contact technique, BTT is necessary for rotating blade vibration measurements. The higher accuracy of amplitude and vibration frequency identification has been pursued since the development of BTT. An improved circumferential Fourier fit (ICFF) method is proposed. In this method, the ToA is not only dependent on the rotating speed and monitoring position, but also on blade vibration. Compared with the traditional circumferential Fourier fit (TCFF) method, this improvement is more consistent with reality. A 12-blade assembly simulator and experimental data were used to evaluate the ICFF performance. The simulated results showed that the ICFF performance is comparable to TCFF in terms of EO identification, except the lower PSR or more number probes that have a more negative effect on ICFF. Besides, the accuracy of amplitude identification is higher for ICFF than TCFF on all test conditions. Meanwhile, the higher accuracy of the reconstruction of ICFF was further verified in all measurement resonance analysis.

For ultrasonic systems, the resonance frequency tracking (RFT) is the most critical step. The rapid development in advanced material processing and microelectronics package has increased the demand of high speed RFT. Therefore, this paper proposes a fast RFT (FRFT) method according to the characteristics of piezoelectric transducers' (PT) admittance circle. In the proposed method, the PT is driven at two different frequencies, and the PT's admittance is collected and calibrated. Then, the PT's mechanical resonance frequency is derived using the admittance information after calibration. The proposed method is not affected by the parallel capacitor and the matching circuit. Additionally, the optimal initial values of the involved parameters are determined in order to improve the accuracy of the proposed method. Furthermore, an improved method based on multiple tracking is also provided. Simulations and experiments demonstrate that using the proposed FRFT method, the ultrasonic system can track the resonance frequency in a short time with high accuracy.

Blade tip timing is an effective method for blade vibration measurements in turbomachinery. This method is increasing in popularity because it is non-intrusive and has several advantages over the conventional strain gauge method. Different kinds of sensors have been developed for blade tip timing, including optical, eddy current and capacitance sensors. However, these sensors are unsuitable in environments with contaminants or high temperatures. Microwave sensors offer a promising potential solution to overcome these limitations. In this article, a microwave sensor-based blade tip timing measurement system is proposed. A patch antenna probe is used to transmit and receive the microwave signals. The signal model and process method is analyzed. Zero intermediate frequency structure is employed to maintain timing accuracy and dynamic performance, and the received signal can also be used to measure tip clearance. The timing method uses the rising and falling edges of the signal and an auto-gain control circuit to reduce the effect of tip clearance change. To validate the accuracy of the system, it is compared experimentally with a fiber optic tip timing system. The results show that the microwave tip timing system achieves good accuracy.

Motion control of unmanned surface vehicles (USVs) is a crucial issue in sailing performance and navigation costs. The actuators of USVs currently available are mostly a combination of thrusters and rudders. The modeling for USVs with rudderless double thrusters is rarely studied. In this paper, the three degrees of freedom (DOFs) dynamic model and propeller thrust model of this kind of USV were derived and combined. The unknown parameters of the propeller thrust model were reduced from six to two. In the three-DOF model, the propulsion of the USV was completely provided by the resultant force generated by double thrusters and the rotational moment was related to the differential thrust. It combined the propeller thrust model to represent the thrust in more detail. We performed a series of tests for a 1.5 m long, 50 kg USV, in order to obtain the model parameters through system identification. Then, the accuracy of the modeling and identification results was verified by experimental testing. Finally, based on the established model and the proportional derivative+line of sight (PD+LOS) control algorithm, the path-following control of the USV was achieved through simulations and experiments. All these demonstrated the validity and practical value of the established model.