This paper presents a review of the latest advances in the field of capacitive, inductive (eddy current), and magnetic sensors, for measurement of absolute displacement. The need for accurate displacement and position measurement in the micrometer, nanometer, and subnanometer scales has increased significantly over the last few years. Application examples can be found in high-tech industries, metrology, and space equipment. Besides measuring displacement as a primary quantity, absolute displacement sensors are also used when physical quantities such as pressure, acceleration, vibration, inertia, etc., have to be measured. A better understanding of the commonalities between capacitive, inductive, and magnetic displacement sensors, as well as the main performance differences and limitations, will help one make the best choice for a specific application. This review is based on both theoretical analysis and experimental results. The main performance criteria used are: sensitivity, resolution, compactness, long-term stability, thermal drift, and power efficiency.

Advances in Capacitive, Eddy Current, and Magnetic Displacement Sensors and Corresponding Interfaces

Learning from a few samples to automatically recognize the plant leaf diseases is an attractive and promising study to protect the agricultural yield and quality. The existing few-shot classification studies in agriculture are mainly based on supervised learning schemes, ignoring unlabeled data's helpful information. In this paper, we proposed a semi-supervised few-shot learning approach to solve the plant leaf diseases recognition. Specifically, the public PlantVillage dataset is used and split into the source domain and target domain. Extensive comparison experiments considering the domain split and few-shot parameters (N-way, k-shot) were carried out to validate the correctness and generalization of proposed semi-supervised few-shot methods. In terms of selecting pseudo-labeled samples in the semi-supervised process, we adopted the confidence interval to determine the number of unlabeled samples for pseudo-labelling adaptively. The average improvement by the single semi-supervised method is 2.8%, and that by the iterative semi-supervised method is 4.6%. The proposed methods can outperform other related works with fewer labeled training data.

/pdf/semi-supervised-few-shot-learning-approach-for-plant-32o95uxq4d.pdf

Semi-supervised few-shot learning approach for plant diseases recognition.

With the rise of artificial intelligence, deep learning is gradually applied to the field of agriculture and plant science. However, the excellent performance of deep learning needs to be established on massive numbers of samples. In the field of plant science and biology, it is not easy to obtain a large amount of labeled data. The emergence of few-shot learning solves this problem. It imitates the ability of humans' rapid learning and can learn a new task with only a small number of labeled samples, which greatly reduces the time cost and financial resources. At present, the advanced few-shot learning methods are mainly divided into four categories based on: data augmentation, metric learning, external memory, and parameter optimization, solving the over-fitting problem from different viewpoints. This review comprehensively expounds on few-shot learning in smart agriculture, introduces the definition of few-shot learning, four kinds of learning methods, the publicly available datasets for few-shot learning, various applications in smart agriculture, and the challenges in smart agriculture in future development. 

/pdf/a-survey-of-few-shot-learning-in-smart-agriculture-rutttvpg.pdf

A survey of few-shot learning in smart agriculture: developments, applications, and challenges

In this study, an eddy current (EC) detector is integrated in an additive/subtractive hybrid manufacturing (ASHM) process. The detector facilitates in-process inspection and repair operations through material deposition, defect detection, and removal processes layer by layer. A feasibility test is carried out on eddy current detection of subsurface defects in additively manufactured parts by using an EC detector. The study compares the results obtained from the EC detection with those by the X-ray computed tomography and the destructive methods. Experiments and simulations are conducted to investigate the effect of excitation frequency on intensity of the eddy current signal. The effects of residual heat of an additively manufactured specimen and lift-off distance of an EC probe on impedance changes are also investigated. In addition, the effect of defect width on EC signal is analyzed. The study shows that the EC method is capable of detecting subsurface defects in the ASHM parts. It is promising to integrate the EC detection and subtractive manufacturing into additive manufacturing to produce parts with improved quality and better performances.

Eddy current detection of subsurface defects for additive/subtractive hybrid manufacturing

Mechanical equipment is the heart of industry. For this reason, mechanical fault diagnosis has drawn considerable attention. In terms of the rich information hidden in fault vibration signals, the processing and analysis techniques of vibration signals have become a crucial research issue in the field of mechanical fault diagnosis. Based on the theory of sparse decomposition, Selesnick proposed a novel nonlinear signal processing method: resonance-based sparse signal decomposition (RSSD). Since being put forward, RSSD has become widely recognized, and many RSSD-based methods have been developed to guide mechanical fault diagnosis. This paper attempts to summarize and review the theoretical developments and application advances of RSSD in mechanical fault diagnosis, and to provide a more comprehensive reference for those interested in RSSD and mechanical fault diagnosis. Followed by a brief introduction of RSSD’s theoretical foundation, based on different optimization directions, applications of RSSD in mechanical fault diagnosis are categorized into five aspects: original RSSD, parameter optimized RSSD, subband optimized RSSD, integrated optimized RSSD, and RSSD combined with other methods. On this basis, outstanding issues in current RSSD study are also pointed out, as well as corresponding instructional solutions. We hope this review will provide an insightful reference for researchers and readers who are interested in RSSD and mechanical fault diagnosis.

https://www.mdpi.com/1424-8220/17/6/1279/pdf

Resonance-Based Sparse Signal Decomposition and its Application in Mechanical Fault Diagnosis: A Review.

For displacement measurement, to explore the influence of tilt angle on sensor’s measured values, the relationship between coil equivalent inductance and lift-off under different tilt angles and the relationship between coil equivalent inductance and tilt angle under different lift-offs were investigated by the finite element method in this article, respectively. Comparing the two variation laws, the increase of tilt angle will lead to smaller sensor’s measured values. Moreover, the applicability of the transformer-equivalent circuit model has nothing to do with the tilt angle. Besides, the experiments also testify that the measuring errors can be compensated well by the Gaussian function.

Influence of tilt angle on eddy current displacement measurement

Mirror milling has been regarded as an effective way for large monolithic thin-walled parts machining. However, the supporting force not only improves the stiffness of the cutting point but also affects the dynamic behavior of the machining system. Essentially, the influence mechanism of supporting force on the mirror milling stability should be analyzed. Firstly, a 3-DOF dynamic system model has been developed considering the supporting head-workpiece interaction during mirror milling. And then, the evolution of modal parameters of thin-walled parts under different supporting forces is investigated with a series of impulse harmer tests, which will supply the dynamic parameters for the mirror milling stability lobes prediction using the full-discretization method. On this basis, the optimum supporting force can be determined according to the relationship of the limited cutting depth and the supporting force. At last, the analysis resulted was verified with a series of mirror milling slot test. From the comparison, the optimum supporting force could maintain the stable mirror milling and avoid excess deformation simultaneously.

The influence of supporting force on machining stability during mirror milling of thin-walled parts

Influence of Coupling Interference on Arrayed Eddy Current Displacement Measurement

Due to the synthetic effect of hydraulic force and magnetic force, chains in magnetorheological fluid (MRF) in squeeze mode appear bending even fracture, especially the latter of which has a potential influence on the macroscopic yield of MRF. Thus fracture mechanism of chains is investigated. As an important factor leading to fractrue, the magnetic force between particles is modelled. Based on the magnetic force model, a fracture criterion is established that when the horizontal component of magnetic force turns to zero, the fractrue phenomenon happens. Based on this criterion, fracture angle and fracture position of chains are analyzed. At last, fracture phenomenon was observed experimentally using an innovative device. It is found that the fracture was the result of interaction of hydraulic force and magnetic force. Driving effect of hydraulic force makes chains bend. Restoring effect of magnetic force prevents chains breaking up. When chains are bent to the extent that the fracture criterion is met, fracture phenomenon happens.

https://iopscience.iop.org/article/10.1088/1361-665X/ab39f9/pdf

Aggregated chain morphological variation analysis of magnetorheological fluid (MRF) in squeeze mode

The thickness measurement of large thin-walled part has become increasingly important to inspect the machining quality and compensate the machining errors. So far, commercial hand-held ultrasonic thickness gauges have been widely used in the wall thickness inspection of large thin-walled parts. However, the manual operation has some disadvantages such as insufficient measurement efficiency, intensive labor, and limited accuracy. To tackle these problems, an automatic thickness measuring method is supposed to be an effective approach. In this article, an ultrasonic on-machine scanning technique for thickness measurement of large thin-walled parts is proposed. First, a multi-function integrated ultrasonic thickness on-machine measuring device is innovatively designed, which enables the posture of an ultrasonic sensor to adapt to the measured surface orientation automatically. Second, the on-machine measurement model of the thickness distribution is established, which includes a measuring point coordinate extraction model and a precise thickness assessing model. Then, a scanning motion control strategy for contact ultrasonic measurement is proposed to maintain a constant pressure between the ultrasonic transducer and the workpiece during the scanning process. Finally, taking the arched part, inclined plate, and S-shaped thin-walled part as typical objects, the experimental study has been conducted using the proposed method. The experimental results show that the ultrasonic thickness on-machine measurement accuracy can be controlled within 0.04 mm, which validates the feasibility of the ultrasonic thickness on-machine scanning technique on thickness measurement of complex surface thin-walled parts.

Meng Lian

Papers

Influence of tilt angle on eddy current displacement measurement

The influence of supporting force on machining stability during mirror milling of thin-walled parts

Influence of Coupling Interference on Arrayed Eddy Current Displacement Measurement

Aggregated chain morphological variation analysis of magnetorheological fluid (MRF) in squeeze mode

Ultrasonic on-machine scanning for thickness measurement of thin-walled parts: modeling and experiments