10results about How to "Loss mitigation" patented technology

The utility model relates to the technical field of pole lug welding, and discloses a soft package battery pole lug welding and pressing device which comprises an equipment base, and a driving motor, a driving shaft, a rotating motor, a strip-shaped rotating disc, a clamping plate, a metal sheet, a fixing frame and a top plate motor are inserted in the equipment base. The technical effects and advantages of the utility model are as follows: the structure is reasonable, the strip-shaped rotating disc is rotated by the rotating motor, the driving shaft is rotated by the driving motor, a metal sheet is inserted into the strip-shaped rotating disc, the strip-shaped rotating disc is rotated to align to the bottom of the pressing plate, the aerostatic press enables the air pressure column to descend, and the pressing plate abuts against the metal sheet; a top plate motor rotates a top plate gear, the top plate gear drives a top plate tooth row to move forwards, the metal sheet is extruded inwards, an extrusion motor rotates a pressing plate, the metal sheet is extruded again, and pseudo soldering is reduced through the structures such as the welding assembly, the extrusion motor and the plate bending column; and the labor cost is reduced.

The application belongs to the technical field of target detection, and discloses a UAV image target detection method based on an improved RT_DETR model, which introduces an FAPPA module in a backbone network, effectively alleviating the information loss problem of small target features in a deep network. The method also replaces the feedforward neural network of the AIFI module in the encoder with an EDFFN module, which selectively enhances high-frequency features such as edges and textures by using a patch_wise FFT strategy and learnable frequency domain weights. In addition, the method also uses a Zoom_cat module to realize adaptive alignment of multi-scale features, and uses the convolution_attention dual-path architecture of the CAFMFusion module to fuse local details and global semantics, thereby fully utilizing complementary information at different levels. The method can significantly improve the UAV image target detection capability and also identify small target features in a complex background.

The invention discloses a wake detection model construction method based on frequency domain multi-scale perception, which relates to the technical field of ocean observation, is used for capturing weak wake features to carry out sea surface internal wave wake detection, and comprises the steps of constructing a wake image data set, constructing a wake detection model by taking an RT-DETR model as a baseline model, and constructing a multi-scale texture information selection network. A backbone network of the baseline model is replaced, a frequency domain multi-scale attention perception module is constructed, an AIFI module of the baseline model is replaced, a golden cudgel convolution down-sampling module is adopted to replace a standard convolution down-sampling module of the baseline model, training and optimization are performed on the wake detection model based on the wake image data set, and a target detection result is obtained. According to the method, the wake detection model AUVRT-DETR is constructed, the detection precision is improved while the advantage of light weight is kept, the optimal balance is achieved between the detection precision and the calculation efficiency, and the model parameter quantity and the calculation overhead are remarkably reduced while the highest detection precision is achieved.

The present invention discloses a lumbar plexus nerve root segmentation method, medium and device based on magnetic resonance diffusion tensor imaging. First, lumbar plexus nerve DTI imaging DICOM data is obtained and stored in NRRD format; then, the region of interest is extracted from the DTI data, and its region of interest annotation data is converted into a grayscale image in PNG format for storage; finally, the DTI data and its region of interest annotation data are input into a deep learning segmentation model for training, and the lumbar plexus nerve root is segmented and predicted by the trained segmentation model. The present invention utilizes a deep learning segmentation model to learn richer image information, adopts a multi-head self-attention mechanism based on windows and moving windows to enhance the model's global feature learning ability at different resolution scales, and constructs a SwinDW fusion module by combining a convolution module and a wavelet feature extraction module to better fuse global and low-frequency features; the present invention also uses a multi-scale feature adaptive fusion module to fuse the feature map of the current encoding module with the feature map of the encoding module at different resolution scales, so as to learn the feature information of the encoding modules at different scales, alleviate the spatial information loss caused by encoder downsampling, and thus better improve the segmentation effect of the lumbar plexus nerve root.

This invention discloses an optimized scheduling method for a multi-source reservoir-canal system in an irrigation district based on the effective gravity-flow water level at the end of the canal. First, parameters of the irrigation district's reservoir group and main canal system are obtained, and the minimum effective water level required for gravity flow at the end of each canal is determined. The minimum water supply flow limit and pre-storage threshold are derived to form a baseline constraint. The pre-storage period and water supply period are divided, and a gravity-flow discriminant function is established. A mixed-integer linear programming model is constructed to solve the macroscopic discharge and water distribution process. This model is used as the upstream boundary of a one-dimensional unsteady flow model, and the hydraulic response is simulated using the Saint-Venant equations to calculate the net effective water supply. The deviation between this deviation and the target water demand is calculated, with the maximum deviation less than the tolerance used as the convergence criterion; otherwise, iterative optimization is performed after correction using a damping learning rate. This invention couples macroscopic scheduling with microscopic simulation, dynamically embedding the gravity-flow water level at the end of the canal into the model for optimization, alleviating ineffective water supply and tailwater loss, improving water resource utilization efficiency, and enhancing the feasibility of the solution.

This invention discloses a method and apparatus for constructing a microdroplet self-Fenton system. The construction method includes the following steps: mixing an iron salt with an acidic solution, and subjecting the resulting mixed solution to ultrasonic atomization to form micron-sized droplets, thereby obtaining the microdroplet self-Fenton system. The construction apparatus includes an atomizing reactor for storing the mixed solution and an ultrasonic generator for providing ultrasonic waves. The microdroplet self-Fenton system constructed by this invention has the following advantages: good adaptability, enabling the acquisition of a self-Fenton system under conditions of no light, no catalyst, and no sacrificial agent; it helps to increase the generation rate and yield of ·OH in the system, giving the system stronger redox capabilities and enabling efficient degradation of organic pollutants, providing important guidance for promoting the effective treatment of organic pollutant wastewater; the process is simple, easy to operate, low in cost, environmentally friendly, suitable for industrial applications, and facilitates the effective treatment of organic pollutant wastewater.

The application claims a micro-expression recognition method based on spatio-temporal feature representation learning, aiming to solve the problems of spatial detail loss, temporal feature ambiguity and small sample overfitting in micro-expression recognition, belonging to the technical field of computer vision and pattern recognition. The method comprises the following steps: step 1. The application proposes an upsampling and position perception module, which restores high-resolution features and performs orthogonal decomposition on them to accurately locate the coordinates of high-frequency action units, so as to alleviate the loss of spatial details. Step 2. The application designs a motion difference fusion module, which highlights weak motion changes through explicit time difference and adaptive gating mechanism, and prevents them from being covered by deep semantic features. Step 3. The application designs a two-stage adaptive prototype contrast learning strategy, which compresses the intra-class variance and expands the inter-class boundary through focal loss preheating and orthogonal prototype alignment based on memory bank, to cope with the problems of small sample and class imbalance.

The invention provides micron silicon slurry, micron silicon carbon negative electrode slurry, a preparation method of the micron silicon carbon negative electrode slurry, a negative electrode plate, a lithium ion battery and electric equipment, and relates to the field of lithium ion batteries. The micron silicon slurry comprises the following raw materials in parts by weight: 3-7 parts of a lithium supplementing binder, 0.3-1 part of a dispersing agent, 4 parts of micron silicon powder and 88.7-95 parts of a solvent, the lithium-supplementing binder comprises a polyacrylic acid-based polymer containing carboxyl functional groups and lithium. Different from the powder form of the traditional silicon-based negative electrode, the invention directly provides the micron silicon slurry, the micron silicon slurry can be directly applied to the current negative electrode slurry mixing process, and the dispersing agent and the lithium-supplementing binder are added, so that the lithium-supplementing effect is improved. When the micron silicon slurry is used, the problems of dispersion difficulty and bonding strength of micron silicon do not need to be considered too much.

The invention discloses a perovskite solar cell laminated assembly, a preparation method thereof and a photovoltaic system, and relates to the technical field of solar cell laminated assemblies. The perovskite solar cell laminated assembly comprises TCO glass, a first functional layer, a first top electrode, a second functional layer and a second top electrode which are laminated in sequence. One or more groups of P1a grooves / P3a grooves, P2a grooves, P2b grooves and P3b grooves which are arranged in sequence are formed in the transverse direction of the TCO glass; the P2b groove and the P2a groove serve as conductive channels and are used for enabling the second top electrode to be electrically connected with the transparent conductive oxide thin film on the surface of the TCO glass. According to the technical scheme, the grooves are reasonably arranged, the thermal influence generated by laser etching of the grooves can be greatly reduced, the overall mechanical stability of the perovskite solar cell laminated assembly is improved, the electric leakage loss and the short circuit risk generated by the tunneling layer can be effectively relieved, and the photoelectric conversion efficiency of the perovskite solar cell laminated assembly is improved.

The invention belongs to the technical field of human body posture estimation, and discloses a lightweight posture estimation method based on a frequency domain attention mechanism, and the method comprises the steps: compressing input feature information through grouping convolution, and then restoring the compressed feature information through three attention modules. Firstly, the frequency attention module uses the filtering module to extract different frequency components, important frequency components are selected through softmax normalization key values, and then the channel and space attention module better captures global information and accurately positions key features. The calculation complexity of the frequency domain attention mechanism module provided by the invention is linearly related to the image resolution, the calculation complexity is smaller, the problem that the calculation amount is rapidly expanded along with the increase of the resolution can be solved, and meanwhile, the module can reduce information loss caused by packet convolution feature compression, so that the accuracy of the algorithm is improved. Therefore, the human body posture estimation accuracy in the lightweight scene is further improved.