A method to optimize the flexible pivot mechanism of the electrostatic force balance （EFB） for expanding its stiffness range and enhance its force resolution is provided. A negative restoring torque device is introduced to adjust the vertical stiffness of the balance， and this adjustment can be achieved by changing the spring length in the device. A magnetic damping device is adopted to achieve stable control of the flexible pivot structure under different stiffness conditions， thereby reducing the convergence time of the balance after loading and unloading operations. Experimental measurements show that the vertical stiffness of the flexible pivot structure in the EFB can be adjusted within the range of 4.08×10^-2 ~ 1.01 N/m， and the convergence time after loading and unloading opertion could be reduced to 5 seconds. The designed flexible pivot structure can meet the requirements for stiffness adjustment range and stability of micro-newton force measurement devices.

The strain beam is an important tool for high-precision calibration of strain sensors. However， due to the dual influence of excitation attenuation and inaccurate response measurement， the load calibration model of strain beam under multi-source load is easy to deviation. For this reason， a multi-source load calibration error estimation method is proposed to error estimation. According to the mechanical characteristics of the cantilever of the strain beam， the load transfer relationship between the multi-source excitation and the measured section of the structural response is analyzed， and the stress model of the measured section is established. In order to verify the effectiveness of this method， a strain beam calibration system is built， and a multi-source load loading experiment is completed. Experimental results show that the accuracy of the proposed method is better than 0.99， and the relative estimation error is better than 1.3%.

A novel wing load calibration method based on high-order weighted least squares was proposed to address the issues of nonlinearity and heteroscedasticity in the strain bridge used for wing structural load measurement during ground load calibration tests. The spatial relationship between the loading point and the measurement section on the wing surface was considered， and a transfer model for multi-source loads on the wing surface was developed. Regression analysis between bending moment， shear force， and torque concerning the measurement section was carried out， establishing a higher-order overdetermined calibration model， which linked multi-source loads with multi-source responses. Parameter optimization of the higher-order overdetermined model was achieved by the introduction of error weighting factors for the parameters of the calibration model， which were combined with least-squares techniques， thus constructing the wing load calibration equation. Ground calibration experiments were performed on a lightweight wing to validate the effectiveness of the proposed method. The experimental results indicated that the proposed method was characterized by an average error of 1.84%， an average standard error of 3.02 N·m， and a goodness of fit of 0.988. In comparison， the traditional load calibration method was found to yield an average error of 2.56%， an average standard error of 4.03 N·m， and a goodness of fit of 0.999. These results demonstrated that the proposed method was superior to the traditional load calibration method in terms of accuracy and reliability.

Aiming at the problems of low accuracy， poor adaptability of device environment and high cost of traditional conveyor belt deviation and material location detection， a new method of conveyor belt deviation and material location detection based on the association of image and radar data is proposed. The method utilizes the Mask R-CNN model to segment the conveyor belt scene image by instances to fit the edge of the conveyor belt and to judge the deviation based on the area ratio of the rollers. At the same time， the radar data are preprocessed and the Bowyer_Watson algorithm is used to construct the Delaunay triangular section to generate the elevation image. Subsequently， K‑means clustering algorithm is used to simplify the elevation image and classify the material flow by gray scale mean filtering. Finally， the classification results are associated with the image information to show the location and state information of the material flow. Experimental results show that the method achieves more than 95% deviation detection rate and more than 80% material location detection accuracy in real scenarios. Compared with the traditional method， the solution has higher robustness and detection efficiency， and can realize the efficient and reliable detection of conveyor belt deviation and material location.

A multi-scale feature extraction and ResNet-Transformer algorithm （MSFRT） is proposed for fault diagnosis of pumping units. Firstly， the local feature extraction capability of the deep residual network ResNet-34 is utilized to capture the spatial details of dynamometer cards， and the context modeling capability of the Transformer encoder is utilized to obtain global features， constructing an end-to-end fault diagnosis framework for pumping units； Secondly， a multi-scale feature extraction module is introduced， which extracts feature information of different scales in parallel through 1×1， 3×3， and 5×5 convolutional kernels， enhancing the ability to perceive details in dynamometer cards； Finally， a feature fusion attention mechanism is designed to adaptively integrate multi-scale features and global semantic information. Experiments are conducted on a dynamometer cards dataset containing 7 typical operating conditions， and the results show that the algorithm achieved an accuracy of 94% in fault diagnosis tasks， verifying the effectiveness of the proposed algorithm.

To design and optimize tobacco filter particle filling equipment， it is essential to calibrate the contact parameters of the filter particles. The interaction parameters between the particles and the surrounding environment were calibrated through density measurements， angle of repose tests， coefficient of restitution calibration， static friction and rolling friction factor tests. The response surface method was used to optimize the parameters of the particle contact model， and the accuracy of these parameters was verified through discrete element simulation. The experimental results show that the calibrated particles have an average diameter of 0.98 mm， a density of 327.57 kg/m³， and an angle of repose of 5.39°. The coefficient of restitution， static friction factor， and rolling friction factor between the particles were 0.286， 0.7， and 0.202， respectively， while the coefficients of restitution， static friction factor， and rolling friction factor between the particles and stainless steel were 0.7， 0.094， and 0.04. Compared with actual experimental data， the simulation results show small errors， validating the effectiveness and reliability of the proposed calibration method.

The duration of sunshine is one of the key elements in surface meteorological observations. Since 2018， all 2 435 national ground meteorological observation stations in China have been upgraded from manual observation instruments to photoelectric digital sunshine meters for automatic observations. Aiming at the demand of sunshine hours measurement in provincial meteorological departments， an automatic verification device of sunshine hours sensor is developed. The device uses two wide spectrum solar radiometers as measuring standards， and has the functions of automatic north pointing， automatic level adjustment， automatic data acquisition and processing， which can realize the outdoor verification of sensors. The experimental results show that the direct irradiance measurement error of the standard is within ± 10%， and the consistency between the two tandards ranges between ± 5%； The measurement error of the measured sunshine hours by the sunshine meter meets the technical specifications of the China Meteorological Administration （±1 h/d or ±10%/d， whichever is larger）， and the expanded uncertainty U is 0.45 h （k=2）. In addition， the measurement results of clear weather have good consistency， with a standard deviation within 0.09 h； Cloudy weather has a significant impact on detection， and measurement errors can be higher than the average measurement error by 0.79 h.

For the existing Harris scale space combined algorithms with high complexity， low accuracy and poor real-time performance， an improved algorithm is proposed. Establish a scale space according to scale-invariant feature transform（SIFT） algorithm to detect Harris feature points， describe features using a 32 dimensional vector. Use vector similarity to match feature points. The Classic K-means algorithm is improved. It has not a fixed initial value， takes the candidate class center point with large distance and low correlation as the initial class center point and categorizes feature points into the class with the smallest distance. Three pairs of matching points were randomly selected from classes of feature points of two images to form a pair of triangles. The matching points are further filtered by triangles similarity.The improved RANSAC algorithm assigns values to all match points based on the absolute values of match point errors to jointly evaluate the transformation model. The experimental results show that the number of feature points extracted by this algorithm is about 22% less than that of SIFT and Harris algorithm， the matching accuracy is improved by about 13%，and the operation time is decreased by about 4.7%.

Thin films， due to their unique physical， chemical， and mechanical properties， play a crucial role in the development of modern high-tech industries， and the precise understanding of their thermal properties is becoming increasingly important. Thermal diffusivity refers to the ability of a material to approach a uniform temperature under a temperature gradient， reflecting the heat transfer process of an object under non-steady-state conditions. Accurate measurement of the thermal diffusivity of thin films is of great significance for material science and engineering applications. Unlike bulk materials， the measurement of thermal diffusivity for thin films is more difficult， and the requirements for measurement technology are more stringent， which necessitates the development of new methods or the improvement of existing technologies. The measurement techniques for the thermal diffusivity of thin films are mainly based on the photoacoustic and photothermal effects. Based on the methods for measuring the thermal diffusivity of thin films in recent years， the measurement techniques for the in-plane and cross-plane directions of thin films are reviewed， providing a reference for future research and development of thin film measurement techniques.

To address the issue of poor temperature field uniformity in the working area of existing high-temperature molten salt thermostatic baths， high-temperature molten salt is used as the fluid medium. The standard k-ε turbulence model is employed， and proportional integral derivative （PID） control is applied to the temperature regulation of the thermostatic bath using the UDF function in FLUENT. Computational fluid dynamics （CFD） simulations are performed to study the impact of the agitator and its layout on the flow field distribution and temperature field uniformity in the working area of the high-temperature molten salt thermostatic bath. The results show that the addition of an agitator creates a distinct circulation flow within the thermostatic bath， accelerating the mixing and heat transfer between the high and low-temperature molten salts. This improves the temperature field uniformity in the working area from 11.708 92 K to 0.082 33 K. Additionally， using an upper agitator structure effectively improves the uneven velocity distribution in the working area， enhancing the temperature field uniformity to 0.039 49 K.

Based on the laser raindrop spectrometer tests， a new method for testing the raindrop diameter and velocity errors and evaluating the accuracy of precipitation phenomenon recognition is proposed， and a raindrop simulation device that can simulate raindrops of different diameters and adjust the falling height is designed. Splash pattern was used to check the uniformity of the simulated raindrops and the results showed less than 5% error for the same type of raindrops. The relative errors in diameter and velocity measured by the raindrop spectrometer were 7.4% and 6.6%. A precipitation phenomenon simulation laboratory was established to simulate drizzle， rain， snow， and rain mixed with snow， and to achieve simulation of different rainfall intensities. Simulating different rainfall intensities， the measured values of the raindrop spectrometer were less than those obtained by manual measurement， with maximum deviations of 17.8% for rainfall and 14.3% for rainfall intensity. The calibration function was obtained based on the manual measurements， and the maximum deviation decreased to 8.5% after calibration. In addition， when simulating precipitation with different diameters， the raindrop spectrometer could draw the corresponding raindrop spectrum.

Based on the Drucker-Prager Cap model， numerical simulation studies on the densification process of Nd-Fe-B blocks were carried out using ABAQUS software. By constructing the stress field analysis model before and after powder sintering， the stress distribution characteristics and microstructure evolution laws at different sintering temperatures were systematically compared， and the results were analyzed in conjunction with scanning electron microscopy. The research shows that within a reasonable temperature range， as the sintering temperature increases， the uniformity of stress distribution and relative density improves， and the probability of crack initiation decreases. At the same time， the residual stress obtained from numerical simulation is in good agreement with the experimental test results. This study provides important theoretical basis and practical guidance for the optimization of the preparation process of Nd-Fe-B permanent magnetic materials and the improvement of material performance.

The remaining useful life （RUL） of lithium-ion batteries is an important parameter for battery health management. In the actual use process of batteries， the phenomenon of capacity regeneration will occur， and it is difficult to avoid noise interference in the process of battery data acquisition， which affects the quality of data. A battery RUL prediction model based on Transformer combined with pyramid convolutional（PyConv） network is proposed to address the above issues. Capacity is chosen as the health indicator， and feature information of the capacity sequence is extracted by the pyramid convolutional network with convolutional kernels of different sizes. The multi-head attention mechanism in Transformer is used to further learn the temporal features of the sequence. The weighted Huber loss function is used to improve the robustness of the model； Dropout technology is used to improve generalization ability of the model and prevent overfitting during training. The proposed prediction model is tested on the NASA and CALCE datasets and compared with other models. The experimental results show that the proposed model has higher prediction accuracy， on the NASA and CALCE datasets， the relative errors are 0.008 6、0.019 3 respectively； the mean absolute errors are 0.011 5 and 0.012 6 respectively； and root mean square errors are 0.017 3 and 0.018 9 respectively.

The short circuit failure caused by local squeeze of soft-pack batteries is a typical operating condition in a car crash， and it is also the most dangerous loading form. The experiment took a domestic commercial soft-pack battery as the research object， and conducted an extrusion experiment under the mechanical abuse experimental platform. By observing and recording the parameters of stress， temperature and voltage during the experiment， the characteristics of thermal runaway occur in the battery are analyzed. The results show that in the extrusion experiment， the process of thermal runaway in the battery can be divided into different stages at different angles， and each stage has its own characteristics； the state of charge has little impact on the mechanics， thermal and electrical aspects of the soft-pack battery under the extrusion conditions； with the increase of the loading speed， the battery limit load becomes smaller， the battery temperature changes faster， and the voltage short-circuit time becomes shorter； with the change of the loading angle of the head， the limit load and temperature both increase with the increase of the extrusion angle， and the battery short-circuit time is not much different； the larger the radius of the cylindrical head， the greater the battery limit load， and the temperature and the short-circuit time are not correlated with the size of the head.

The waveguide method is a calibration method for microwave leakage energy meters provided in the IEEE Std 1309 international standard. Compared with the standard field method and the transfer comparison method， it has the characteristics of high accuracy， low cost， small power， and large field strength. To meet the calibration requirements of the 915 MHz microwave leakage energy detector， the theoretical calculation formula of the waveguide method was derived and refined using electromagnetic field theory. Based on the theoretical calculation formula， a waveguide-based calibration device was designed. The design process of the device was detailed， including the dimensions inside the waveguide， the waveguide-to-free-space conversion mechanism， and the design of the test aperture. The disturbance of the probe under calibration on the internal field and the uniformity of the internal field distribution in the waveguide were also investigated.The simulation results show that the center field strength of the device is basically consistent with the theoretical value. The measured results of the device show that its port voltage standing wave ratio is about 1.14 and insertion loss is about 0.13 dB. Compared with the TEM chamber， the relative deviation is only 0.29 dB. The experiment shows that the device can be well applied in the calibration of microwave leakage energy detectors.

The purpose of this study is to investigate the methodology for determining the chamber correction factor k _ch and to calculate absorbed dose to water in superficial X-rays radiotherapy accurately. Under superficial X-ray radiotherapy reference radiation qualities with tungsten alloy collimators of varying apertures， this study analyzed the ratio of readings measured in air versus on the surface of a PMMA phantom measured by plane-parallel ionization chambers， PTW23342 and PTW23344. Backscatter factors were determined through Monte Carlo simulations， and the variations of the chamber correction factor k _ch with half value layers （HVLs） and radiation beam sizes were measured. Under the four reference radiation qualities （SRT-50a， SRT-50b， SRT-70 and SRT-100）， with an SSD （source to skin distance） of 0.45 m and radiation beam sizes of 3.0~7.2 cm. For the PTW23342 plane-parallel ionization chamber： the ratio of readings measured in air to those on the PMMA phantom surface with ranges from 0.870 0 to 0.991 3. The measured results of correction factor k _ch are within 1.046~1.094. For the PTW23344 plane-parallel ionization chamber： the ratio of readings measured in air to those on the PMMA phantom surface ranges from 0.901 4 to 0.992 4. The measured correction factor k _ch​ are within 1.053~1.108.

International comparison of radionuclide’s activity is generally carried out by delivering samples to the participating laboratory， but which is difficult to execute comparisons among distant countries by this way for short-lived medical gamma nuclides， such as ^99mTc， ¹⁸F and ⁶⁴Cu. The Consultative Committee for Ionizing Radiation put forward a plan to send the international reference system transfer instrument to implement the international comparison of short-life medical gamma nuclide activity， which has been run successfully. The instrument， methods and main results of international comparison for short-life medical gamma nuclides were summarized， and the study progress of activity transfer instrument for short-lived radionuclide in NIM， China is introduced.

Based on lentiviral packaging system， the co-transfection technology of multi-plasmid was used to deliver the relevant genes into host cells containing specific nucleic acid sequences， and the human immunodeficiency virus （HIV-1） pseudovirus reference materials were fabricated. The corresponding spike protein was expressed on the outer shell. The standard value of copy number and expanded uncertainly of the reference materials was （300.12 ± 0.54） ×10⁴ copies/mL （k=2） by digital polymerase chain reaction（PCR） . Taking advantage of the characteristic that the pseudovirus infects by the recognition between the spike protein and the receptor of host cell， the binding could be interfered by adding neutralizing antibodies to inhibit the infection of the pseudovirus. By observing the inhibition degree of neutralizing antibodies at different concentrations on the expression of fluorescence， the Reed-Muench two-formula method was used to calculate the Z _PD50 and evaluated the potency of neutralizing antibodies.