最新刊期
卷 34 ， 期 5 ， 2026
- “介绍了其在光学元件研磨检测领域的研究进展，相关专家提出了一种多测头非接触式面形检测方法，为解决传统检测方法难以兼顾精度与效率的问题提供了新的解决方案。”摘要：To address the limitation that conventional single-probe measurement methods used during the grinding stage of optical components are highly sensitive to machine tool guideway motion errors and struggle to balance measurement accuracy with in-situ measurement efficiency， a multi-probe non-contact surface form measurement method suitable for integration with conventional machine tools is proposed， enabling high-precision in-situ measurement during the grinding process of optical components. First， the differential measurement principle of a multi-probe system composed of three laser displacement sensors is introduced. By performing pairwise differencing among the three probes， the influence of machine tool guideway motion errors is effectively suppressed， and the displacement differences are transformed into local surface slope information. Subsequently， a surface reconstruction approach based on Zernike polynomial slope fitting is described， in which the full-aperture surface form is reconstructed from the measured slope field using a least-squares fitting method. A geometric simulation model of the measurement system is then established， and the effects of probe position and orientation errors on the reconstructed surface are analyzed using the Monte Carlo method. Furthermore， Sobol global sensitivity indices are employed to quantitatively evaluate the contributions of different error sources to the RMS error， providing guidance for system alignment tolerance allocation and error control strategies. Experimental validation is conducted on a Φ400 mm diameter single-crystal silicon plane mirror during the grinding stage. The measurement results obtained using the proposed multi-probe system are compared with those from a coordinate measuring machine （CMM）. The results indicate that the absolute deviation of the peak-to-valley （PV） value is 0.21 μm， corresponding to a relative deviation of 2.22%， while the absolute deviation of the root mean square （RMS） value is 0.14 μm， corresponding to a relative deviation of 7.78%. These results demonstrate that the proposed method maintains good measurement consistency and stability under machine tool operating conditions and provides a practical non-contact surface form measurement solution for high-precision in-situ measurement during the grinding stage of optical components.关键词：optical measurement;multi-probe;surface figure reconstruction;Monte Carlo analysis;Sobol sensitivity indices0|0|0更新时间：2026-03-12
- “研发了高灵敏度空间倾斜测量光纤光栅传感器，其采用凸台 - 凹槽组合结构实现悬臂式等强度梁的应变响应增强，通过两个正交布置的增敏型等强度梁同步感知空间二维倾角变化，为复杂工况下的结构安全监测提供高精度、高可靠性的倾斜数据支持。”摘要：To meet the technical requirements for multidirectional structural tilt measurement in practical engineering， a high-sensitivity circumferential fiber Bragg grating （FBG） tilt sensor was developed， accompanied by systematic theoretical analysis and laboratory validation. A combined lug-boss and groove structural design was adopted to enhance the strain response of cantilevered equal-strength beams. Two orthogonally arranged sensitized equal-strength beams enable synchronous sensing of two-dimensional spatial inclination. By applying differential signal processing to the central wavelength shifts of two FBGs mounted on opposite sides of the beam， temperature-induced effects are effectively eliminated while the measurement sensitivity is further improved. Experimental results indicate that the sensor exhibits sensitivities of 81.234 pm/（°） and 85.235 pm/（°） in the XOZ and YOZ planes， respectively， with linearity exceeding 0.999 79， a resolution better than 0.01°， measurement accuracies of 0.024 6% F.S. and 0.023 5% F.S.， and a measurement range of ±30°. Spatial performance tests further demonstrate that stable and reliable monitoring capability is maintained under two-dimensional tilt conditions. In addition， the sensor shows excellent repeatability （repeatability errors of 0.843% and 0.016%， respectively）， low hysteresis （0.158% and 0.945%， respectively）， and strong resistance to creep. These results indicate that the proposed sensor can provide high-precision and highly reliable tilt data for structural safety monitoring under complex working conditions， demonstrating significant potential for applications in engineering structural health diagnosis.关键词：fiber optic sensor;fiber Bragg grating （FBG）;inclination sensor;spatial tilt measurement;sensitized equal strength beam0|0|0更新时间：2026-03-12
- “介绍了其在烟气监测领域的研究进展，相关专家提出了一种基于机器学习算法的H2S浓度预测方法，为解决H2S与CO2光谱混叠问题提供了切实可行的技术方案。”摘要：To address the spectral overlap between H2S and CO2 absorption lines and improve the accuracy of online H2S measurements in flue gas， an H2S concentration prediction method based on machine learning algorithms is proposed. A comprehensive training dataset was constructed by integrating 580 sets of demodulated data from direct absorption spectra generated using the HITRAN database with experimentally acquired second-harmonic signal data. Several models， including Gaussian Process Regression （GPR）， traditional linear regression， support vector machines， and neural networks， were employed to simultaneously predict the concentrations of H2S and CO2. The results indicate that the GPR model achieved the best performance， with mean relative errors of 0.816% for H2S and 0.673% for CO2， outperforming the other models. Long-term stability tests yielded root mean square errors of 14.181×10-6 for H2S and 0.101% for CO2， demonstrating excellent measurement stability. Overall， the proposed GPR-based machine learning approach effectively resolves the spectral interference between H2S and CO2， enabling high-precision and stable monitoring of H2S， a key indicator of high-temperature corrosion in coal-fired boilers. This method provides a practical and reliable technical solution for real-time flue gas component analysis in industrial environments.关键词：machine learning;overlapping spectra;hydrogen sulfide;tunable diode laser absorption spectroscopy0|0|0更新时间：2026-03-12
Modern Applied Optics
- “相关研究团队在微机电系统（MEMS）领域取得新进展，专家们提出并验证了一种基于硅-玻璃键合不等高工艺的小型化MEMS静电驱动微镜方案，为光通信系统中功耗与尺寸受限问题提供了高效技术路径。”摘要：Electrostatically actuated micromirrors based on silicon-on-insulator （SOI） technology commonly face challenges such as complex fabrication processes， high manufacturing costs， and limited design flexibility. To address these issues， a miniaturized electrostatically actuated MEMS micromirror based on a silicon-glass bonding （SOG） unequal-height process is proposed. A double-layer mask combined with stepwise etching is employed to form a vertical comb structure with a height difference， enabling effective vertical electrostatic actuation. The fabricated micromirror chip measures only 1.7 mm×1.7 mm， with a mirror diameter of 0.82 mm and a thickness of 450 μm， demonstrating a compact structure and good process compatibility. Experimental results indicate that a mechanical deflection angle of 0.52° can be achieved at a driving voltage of 10 V， with a response time of approximately 14 ms. These results demonstrate that the silicon-glass bonding unequal-height process can significantly simplify the fabrication procedure and reduce manufacturing costs while maintaining reliable device performance， providing a promising technical solution for optical communication systems with stringent requirements on power consumption and device size.关键词：electrostatic drive;MEMS micromirror;unequal height process;double mask0|0|0更新时间：2026-03-12
- “介绍了其在精密测量仪器或高端制造装备领域的研究进展，专家提出了一种基于多原理融合的精密旋转轴系六自由度误差同步测量方法，为精密旋转轴系误差的长期监测提供解决方案。”摘要：To investigate the influence of error variations in precision shaft systems on precision measuring instruments and high-end manufacturing equipment incorporating precision rotating shafts， a multi-principle fusion-based synchronous measurement method for six-degree-of-freedom errors of precision rotating shaft systems is proposed. The method integrates the principles of eccentric self-calibration for circular grating installation， self-collimation， and laser triangulation. First， based on the eccentric self-calibration principle for circular grating installation， a separation model for angular positioning error and radial error in precision rotating shaft systems is established. Second， a measurement module for axial and tilt errors is designed and calibrated by combining the principles of self-collimation and laser triangulation. Finally， experiments conducted on a direct-drive （DD） motor demonstrate that the angular positioning error is within ±3.5″， the radial error within ±0.5 μm， the X-axis tilt error within ±6″， the Y-axis tilt error within ±8.3″， and the axial error within ±6 μm. Compared with conventional measurement approaches for precision rotating shaft errors， the proposed method eliminates the need for instruments such as photoelectric autocollimators or commercial spindle error measurement systems. It enables simultaneous measurement of six-degree-of-freedom errors in precision rotating shafts and offers advantages including low cost and in-situ measurement capability. The method is therefore suitable for long-term monitoring of precision rotating shaft errors.关键词：laser triangulation principle;autocollimation principle;six-degrees-of-freedom error measurement;precision rotary axis;installation eccentricity self-calibration principle of circular grating0|0|0更新时间：2026-03-12
- “专家提出了一种结合线性跟踪微分器与迭代学习控制的双传感器切换像移补偿控制方法，为解决多模遥感观测设备在大速高比工况下的像移问题提供了有效解决方案。”摘要：In high-speed， low-altitude remote sensing scenarios， the long exposure time of multimode observation systems often leads to significant image motion. Compensation methods relying solely on gyroscopes suffer from complex command profiles and view-axis deviation caused by accumulated gyro drift， making it difficult to satisfy the stringent requirements of remote sensing imaging for high overlap rates and high signal-to-noise ratios. To address these limitations， an image motion compensation control method based on dual-sensor switching and integrating a Linear Tracking Differentiator （LTD） with Iterative Learning Control （ILC） is proposed. Within a dual-sensor compensation framework combining a gyroscope and an encoder， the gyroscope is employed for closed-loop velocity control to achieve real-time compensation during the exposure phase. After exposure， the control system switches to a high-precision encoder for position closed-loop regulation， enabling rapid return-to-zero motion and eliminating drift accumulation. Meanwhile， LTD-generated commands are used to produce smooth velocity and position trajectories， thereby suppressing dynamic oscillations induced by step excitations. A PD-type ILC strategy is incorporated into the tracking of periodic velocity commands， where historical tracking errors are utilized to construct feedforward compensation， further improving trajectory tracking accuracy. Experiments conducted on a single-degree-of-freedom turntable platform demonstrate the effectiveness of the proposed method. Compared with conventional gyroscope-only compensation approaches， the proposed dual-sensor switching strategy combined with LTD-based command smoothing effectively eliminates view-axis offset， achieves rapid return control without overshoot， and shortens the imaging interval. During the exposure period， the velocity error band is reduced from ±0.241 0 （°）/s to ±0.086 7 （°）/s， representing a 64% reduction and a substantial improvement in compensation accuracy.关键词：image motion compensation;sensor switching;linear tracking differentiator;iterative learning0|0|0更新时间：2026-03-12
- “介绍了其在超精密加工测量领域的研究进展，相关专家构建了综合评定方法，验证了微纳米坐标测量机在圆度测量中的可靠性，为提升测量准确性与可信度提供了有效方案。”摘要：To verify the reliability of micro/nano coordinate measuring machines （CMMs） in roundness measurement and to accurately identify key sources of measurement uncertainty， an evaluation method is established to quantitatively analyze the uncertainty and the influence of various error components on the evaluation results. First， in accordance with the Chinese national standard GB/T 7235-2004， four center calculation methods-least squares， maximum inscribed circle， minimum circumscribed circle， and minimum zone-are employed， and the roundness measurement uncertainty is evaluated using the GUM-MCM method. Subsequently， a Random Forest regression model is introduced to establish the relationship between the errors associated with different uncertainty sources （repeatability， temperature variation， crosstalk force， and probe wear） and the resulting roundness error. By employing three evaluation metrics—out-of-bag error increase， feature elimination error increase， and stability weight—the influence of each error source on the roundness error is systematically quantified. Experimental results indicate that the best estimates of roundness obtained using the four methods are 0.629， 0.221， 0.616， and 0.608 μm， with corresponding standard uncertainties of 0.099， 0.138， 0.103， and 0.094 μm， respectively； among them， the minimum zone method yields the lowest uncertainty. Random Forest analysis further reveals that the error component associated with probe wear exerts the most significant influence in the least squares， minimum circumscribed circle， and minimum zone methods， with total scores of 0.74， 4.50， and 0.53， respectively. By integrating the classical GUM-MCM uncertainty evaluation framework with a Random Forest regression model， the proposed approach effectively verifies the reliability of roundness measurements performed by micro/nano CMMs while quantitatively resolving the contributions of individual uncertainty sources. These findings provide important engineering insights for improving the accuracy and reliability of measurements in ultra-precision manufacturing.关键词：micro/nano coordinate measuring machine;center calculation;roundness;uncertainty evaluation;random forest2|0|0更新时间：2026-03-12
- “相关专家提出了一种外延展螺旋形绝对式时栅角位移传感方法，为解决传统绝对式角位移传感器在设计和加工过程中存在的绝对编码和超精密光刻加工问题提供了新的解决方案。”摘要：To address the limitations associated with absolute coding and ultra‑precision photolithography in the design and fabrication of traditional absolute angular displacement sensors， a novel time‑grating‑based absolute angular displacement sensing method employing an extended spiral structure is proposed. Two sets of sinusoidal and cosine excitation windings with extended spiral geometries were arranged on the inner and outer rings of the stator. These windings are spatially orthogonal and possess opposite pole numbers. An alternating magnetic field was generated using time‑orthogonal excitation signals through time‑division multiplexing， effectively reducing electromagnetic crosstalk and establishing a mapping relationship between temporal signals and spatial angular displacement. Variations in the magnetic field were detected by induction windings to obtain displacement signals， and absolute angular displacement was determined using a combined “positioning + measurement” strategy. Experimental results indicate that an absolute measurement error of ±11.2″ was achieved within a full 360° range. Arcsecond‑level absolute position measurement was realized using sensing units with ten‑thousand‑arcsecond resolution， thereby eliminating the dependence of traditional absolute displacement sensors on absolute coding and ultra‑precision photolithography while significantly reducing manufacturing complexity. The proposed method is suitable for angular displacement measurement in harsh environments such as those contaminated by oil， demonstrating significant academic relevance and engineering application potential.关键词：absolute angular displacement;time grating;extrapolation spiral shape;time division multiplexing;coprime exponents0|0|0更新时间：2026-03-12
Micro\/Nano Technology and Fine Mechanics
- “专家提出基于单目视觉的插植手术机器人空间配准方法，建立手眼空间位姿关系，运用LM-PnP算法，实现手术目标精确配准，为提高手术机器人配准精度提供新方案。”摘要：To improve registration accuracy between the workspace of an implantation surgical robot and the surgical environment， a monocular vision-based spatial registration method is proposed. First， the spatial pose relationship between the camera coordinate system and the robot coordinate system is established through intrinsic calibration of the monocular camera and hand-eye calibration. Subsequently， a surgical target keypoint detection method based on the CenterNet neural network is developed using deep learning techniques. The Perspective-n-Point （PnP） algorithm is then employed to construct a mapping model between the surgical target coordinate system and the camera coordinate system. To further enhance the stability of spatial registration， the Levenberg-Marquardt （LM） algorithm is applied to perform nonlinear optimization of the mapping model. By integrating the hand-eye calibration results， an LM-PnP pose mapping model for the surgical robotic system is established. Finally， experiments are conducted to evaluate the stability and accuracy of the proposed method. The results indicate that the proposed LM-PnP algorithm achieves a detection error of less than 0.186 mm for the target circular hole. During robot pose transformation tests， the orientation and position deviations of the surgical target range from -1.5° to 1° and -1 mm to 1 mm， respectively. These results demonstrate that the proposed registration method provides high stability and accuracy. Compared with conventional binocular depth camera approaches， the proposed method requires only a monocular camera to achieve precise surgical target registration， operates under natural lighting conditions， and does not rely on infrared sensors or optical marker spheres.关键词：monocular vision;spatial registration;surgical robot;deep learning2|0|0更新时间：2026-03-12
- “介绍了其在机载成像系统领域的研究进展，相关专家提出了一种轻量化的基于光流法的视频后半帧实时预测方法，为降低自适应实时控制器计算延迟导致的误差提供解决方案。”摘要：The subsequent frame is utilized as the input for the current frame to mitigate errors caused by the computational latency of adaptive real-time controllers. This strategy effectively improves the correction accuracy of the fast steering mirror and has become an important approach in modern airborne imaging systems. The proposed method takes two consecutive frames as input， from which forward and backward optical flow fields are estimated through an optical flow estimation module. A temporal encoding factor t is introduced to linearly scale the predicted bidirectional optical flows along the temporal dimension. Subsequently， the latter-half frame is predicted via feature alignment based on bilinear transformation. To enhance computational efficiency， the optical flow estimation module adopts a multi-layer depthwise separable convolution architecture， significantly reducing the number of model parameters. In addition， a channel attention enhancement block is incorporated to enable adaptive recalibration across feature channels， thereby improving the extraction of fine-grained features. Experimental results demonstrate that， on the visible-light dataset， the method achieves a PSNR exceeding 27 dB and an SSIM above 0.93， while maintaining a latter-half frame prediction speed of no less than 116 frame/s， approximately 60 frame/s faster than typical algorithms. On the infrared dataset， a PSNR exceeding 41 dB and an SSIM above 0.98 are achieved， with a prediction speed reaching 123 frame/s， also about 60 frame/s higher than that of conventional methods. These results indicate that the proposed algorithm exhibits strong cross-band generalization capability and enables high-frame-rate， high-quality subsequent frame prediction.关键词：computer vision;video frame prediction;deep learning;optical flow method2|0|0更新时间：2026-03-12
-
Compensation and correction of segmented telescopes based on multilayer perceptron neural network AI导读
“专家提出基于MLP神经网络的拼接式望远镜补偿校正方法，验证元件偏心和倾斜可相互补偿，建立光学系统模型，训练子系统模型，实验表明该方法能有效预测调整补偿量，减少子镜调节维度，为高精度装调提供新途径。”摘要：To improve the efficiency of segmented telescope alignment， a compensation and correction method based on a multilayer perceptron （MLP） neural network is proposed. Based on nodal aberration theory， the coupling relationship between decenter and tilt in the optical system is first analyzed， and their capability for mutual compensation is demonstrated. An optical model of a segmented telescope is subsequently established by treating each segment and the secondary mirror as independent subsystems. For each misalignment condition， the first- to ninth-order Zernike coefficients of the wavefront at three field positions are extracted as model inputs， and datasets with different misalignment ranges are used to train 18 subsystem models. During simulated alignment， additional five-dimensional misalignments associated with segment decenter are introduced， while compensation is performed using only the three degrees of freedom-piston and tip/tilt-learned during training. The results indicate that the trained MLP models can reliably predict the misalignment values of the components， enabling the system to return to its nominal design state within two iterations， with a mean RMS of 0.010 7λ over 100 Monte Carlo simulations. When realistic measurement noise is included， effective correction is still achieved within two iterations， yielding a mean RMS of 0.026 9λ across 100 runs. The proposed method effectively reduces the alignment dimensionality of segmented telescopes and provides a practical approach for achieving high-precision assembly and alignment.关键词：segmented telescope;multilayer perceptron;misalignment calculation;nodal aberration;neural network0|0|0更新时间：2026-03-12 - “慧眼识“新”开放词汇目标检测方法，针对开放场景中模型前景判别能力弱、基类偏差大导致新类检测精度较低的问题，提出创新解决方案。”摘要：To address the low detection accuracy of novel classes in open-world scenarios-primarily caused by weak foreground discrimination and strong bias toward base classes-an open-vocabulary object detection framework named Sharp Eyes Spot the “Novel” in Open-Vocabulary Object Detection （SSN-OVD） is proposed. First， a Foreground Feature Discrimination （FFD） module is introduced， in which a foreground estimator is employed to model potential novel-class regions and generate high-quality pseudo-labels， enabling more precise foreground-background separation and enhancing the discriminability of foreground features. Second， a Bidirectional Feature Alignment （BFA） module is designed to leverage bidirectional cross-modal alignment together with confidence calibration， thereby mitigating base-class bias during training and strengthening the model's capability to learn robust representations of novel classes. Third， a Contrastive Denoising Training （CDT） module is developed， incorporating noisy visual–text pairs into the contrastive learning process to further improve feature discrimination and generalization for novel categories. Experimental results demonstrate that the proposed approach achieves state-of-the-art performance， yielding novel-class detection accuracies of 44.9% on COCO and 37.4% on the more challenging fine-grained LVIS dataset. These results indicate that the method effectively enhances novel-class detection in open-world environments.关键词：object detection;open-vocabulary;foreground feature discrimination;bidirectional feature alignment;contrastive denoising training2|0|0更新时间：2026-03-12
- “精密自动移液技术是现代实验室自动化系统的核心。专家们针对移液器吸头微小缺陷检测难题，建立了TIP-YOLO模型体系，为解决移液器吸头缺陷检测问题提供了有效方案，推动实验智能化发展。”摘要：Quality defects in pipette tips directly affect the accuracy of liquid aspiration and dispensing， leading to volume deviations and reduced pipetting precision. However， pipette tips are typically manufactured from translucent polypropylene， and their defects are characterized by small size， random distribution， and shallow-layer structures. These properties result in extremely low optical contrast with the background， producing weak defect features and blurred boundaries that are difficult to identify accurately using conventional manual visual inspection methods. To address these challenges， a tiny defect detection model for pipette tips， termed TIP-YOLO， is proposed and applied to automated pipetting devices. First， the GrabCut algorithm is employed to achieve precise segmentation of tip defects， after which Deep Image Blending is utilized to integrate the segmented defect features with background images， thereby constructing a high-quality multi-type defect dataset. Furthermore， a dynamic upsampling module is introduced into the neck network of YOLOv8n to enhance the reconstruction of edge details of tiny defects through adaptive learning of pixel offsets. In addition， a four-layer detection head incorporating adaptive spatial feature fusion is designed， in which a dedicated detection branch for tiny defects is added to improve multi-scale feature fusion efficiency and increase sensitivity to spatial feature representation. Experimental results demonstrate that the proposed TIP-YOLO model achieves a precision （P） of 0.969， recall （R） of 0.971， and mAP@0.5 of 0.986， representing improvements of 3.0%， 5.2%， and 2.8%， respectively， compared with the baseline model. These results indicate that the proposed method provides an effective solution for pipette tip defect detection and offers strong support for advancing intelligent laboratory automation.关键词：YOLO model;defect detection;pipette tip;small object detection;automatic pipetting0|0|0更新时间：2026-03-12
Information Sciences
- Address：No.3888 Dong Nanhu Road, Changchun, Jilin, China Postal code：130033
- Tel：0431-86176855 Email：gxjmgc@ciomp.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 吉ICP备11002662号-17
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0