Micro-hole drilling is an important application of micro electrical discharge machining (EDM). With the increase in the machining depth, bubbles and debris generated by discharges are difficult to exhaust from and accumulate in the narrow inter-electrode area. The by-product accumulation not only leads to the deterioration of the discharge environment and increase of abnormal discharges, but also make the observation of the machining area difficult. In this paper, a new observation method is proposed to observe the shape and distribution of bubbles generated by discharge at different machining depths. The depths can be simulated by the position between the electrodes and the transparent glass tube. This observation method is helpful to better study and analyze bubble movements and the flushing effect cause by the movement in EDM drilling of deep micro-holes.

SiC is a popular topic since it has better performance than conventional semi-conductor in electronic applications. However, the hardness of SiC approaches to that of diamond which results in a challenge for conventional abrasive wire slicing in EDM. Thus, arc plasma ignited on the SiC workpiece should be investigated. Besides, it is clear that current pulse shapes have influences on the discharge phenomena which can influence the machining efficiency, tool wear and thickness of damage layer etc. in EDM. However, the gap phenomenon of discharging ignited by different current pulse shapes have not be demonstrated clearly, which should be investigated carefully. In this case, a high speed video camera is used to observe the gap phenomena and take photos to get diameters of arc plasma, which is useful to understand the discharge process on the SiC when using different pulse generators.

In this study, aiming at processing marks on the metal workpiece surface, a stray-corrosion-free electrochemical machining (ECM) method was proposed. In this method, a kind of porous solid ball, called MFS-ball, is used as the electrolyte absorption material to limit the electrolyte distribution and eliminate stray-corrosion in ECM. Finally, examples of dimple English alphabets were etched on the workpiece surface.

Precision ECM in stationary electrolyte using stamp flushing method has been proposed in our previous study to avoid the electrolyte flow resulting in ununiformly distributed temperature and volume fraction of bubbles during ECM. However, the sludge and bubbles squeezed out from the gap between the tool electrode and the workpiece by stamp flushing cannot be flushed out quickly enough in stationary electrolyte, thus they may flow back into the gap after the stamping motion. Here, the electrolyte flowing along the axial direction is used to quickly flush out the bubbles and sludge squeezed out from the gap by stamp flushing. Since the electrolyte in the gap is not affected by the axial flow of electrolyte, ECM still occurs in stationary electrolyte. The influence of electrolyte flowing along the axial direction on the machining speed and accuracy were investigated.

Minimally invasive insertion of needles offers considerable advantages over the conventional sensing routes. Moreover sensing at transdermal region is easily accessible without any complications. Measurement of electrolyte abnormalities resulting from prolonged sweating would be easiest way to detect the heatstroke risk. In this regard, we proposed a low cost, facile preparation, easy to use sensor for the detection of ions in transdermal region. Microneedles was achieved by replicating structures from a laser engraved stainless steel master. And then PDMS elastomer was poured on stainless steel master, cured for 2 hours at 70oC and peeled off. The peeled off microneedle array structure is semi flexible and transparent. Then, Au/Ti electrodes were deposited on PDMS using shadow mask to make electrodes for sensing studies. Finally the electrodes were modified using Ag/AgCl and ZnO for sensing characterizations. Electrochemical studies were performed using the conventional three electrode system by employing electrochemical workstation (Bipotentiosrat 2325, ALS Japan). Cyclic voltammetric and amperometric techniques were used for electrochemical analysis of the bare and metal oxide functionalized microneedle electrodes in the range of 10-100 μM.

A dynamic resonant frequency control method has been proposed for multimodal piezoelectric actuators. Resonant frequency matching is significant for piezoelectric actuators working in multiple vibration modes. In this case, a group of passive piezoelectric ceramics was used in the actuator design to overcome the resonant frequency ration shift caused by the actuator's machining error, abrasion, and boundary condition changes like load and temperature. A FET alternating current switch was employed to dynamic change the passive parts' electrical boundary condition between open and short, ensuing stiffness change and the continuous resonant frequency control. In this study, an RSIDM (resonant-type smooth impact drive mechanism) actuator prototype with passive PZT was fabricated and investigated. The first and third longitudinal modes' resonant frequency ratio was precisely adjusted to be 1: 2, and a quasi-sawtooth displacement was generated for liner inertia actuation.

Recently, due to rapid development of the photoelectric industry, the needs for non-axisymmetric aspheric optical components are increasingly presented, such as the lenses for short imagine distances head-up display and compensation lens for elliptical laser beam. The machining process for such die/mold requires an external feeding device mounted to a machine tool to achieve fast tool servo (FTS) function for fabricating a non-axisymmetric surface. In this speech, the non-singular terminal sliding mode controller (NTSMC) is introduced to precisely control the feeding device consisted of a piezoelectric (PZT) actuator. To examine the control performance based on the NTSMC, a machining example of non-axisymmetric aspheric surface is provided, details of controller design are described, and both the simulation work and control practice are presented. As the results of performance examinations, the peak-to-valley (PV) tracking errors of simulation and control practice are 0.01 nm and 0.96 μm, respectively. This large difference is found mainly due to a slow sampling rate of position commands. Finally, the effect of machining experiment based on the NTSMC is demonstrated. Future work to enhance machining performance through hardware improvement is discussed.

Ultrasonic motors are next-generation actuators with fast response, compact structure, and high torque at low speed.However, due to their non-linearity, their control is rather challenging. Under driving frequency control of the traveling waveultrasonic motor, the speed response experiences non-linearity, discontinuity, and speed hysteresis. A deep reinforcement learning(DRL) was proposed as a nonlinear, optimal, offline, and model-free controller. A DRL algorithm optimized a neural network speedcontroller through iterative trial and error. However, the obtained controller resulted in a non-zero steady-state error and instabilityaround the response region. In this work, we propose DRL for optimal gain scheduling of an integral controller to realize a stablenonlinear optimal zero-steady state error controller. To apply DRL to integral controller gain scheduling, an NN-based policy learnsthe mapping from an input state (A Markovian state includes the driving frequency, motor temperature, current speed, and targetspeed) to an optimal I-gain by maximizing some reward function(a function of the speed error and the control effort). The drivingfrequency step is obtained as a multiple of optimal I-gain and speed error. Using the Soft Actor-Critic (SAC) algorithm, optimalzero steady-state error speed tracking was realized both in simulation and experiment results.

Separating shapes and textures at different scales is useful in many image processing applications. Therefore, scale-aware image filters which remove structures smaller than the specified scale while preserving salient edges have attracted considerable attention. Conventional computational methods of such filters have some drawbacks including non-uniform convergence and salient edge sharpness. In this research, we propose a novel scale-aware image filter with more stable convergence and edge saliency for varying scales. Our filter is based on a recursive linear regression model which is adaptively and dynamically regularized for salient edge pixels. Our numerical experiments demonstrated the filtering results with uniform convergence and high accuracy for varying scales.

Flow rate control is essential for safety and time management in intravenous drip therapy. The main control method is to adjust the drop velocity. However, due to the influence of surfactants contained in some drugs, a drop may fall below the specified volume, and therefore, much effort and expense are required to improve the accuracy.In this paper, we report on the use of light to measure the volume of a droplet, and the successful miniaturization of a device and highly accurate flow rate measurement by devising an optical path. It is expected to contribute to the improvement of the performance of drip infusion systems.

I develop the device which uses lattice spacing scale and combined scanning tunneling microscope (STM) with highly oriented pyrolytic graphite (HOPG). Applying new interpolation method for the images of atoms obtained from this device, I aim to measure 2D displacement in Pico meter order. In addition, I perform calibration on optical interferometer using this device and confirm the effectiveness of the proposed interpolation method.