14results about How to "Fast fitting" patented technology

The invention discloses a data analysis and conclusion generation method and device based on multi-agent proxy interaction. The method comprises the following steps: constructing a multi-node distributed federated learning architecture; each node extracts communication protocol data; performing data feature extraction and dimension reduction on the communication protocol data to obtain low-dimensional network communication protocol features; generating local training data, then respectively inputting the local training data into local models of corresponding nodes for training, and uploading trained model parameters to a global model of a central server so as to generate a generative model based on a Transform network; and aggregating and updating the parameters by adopting an ecological balance optimization algorithm, and then issuing the updated parameters to each node to start a new round of iterative training until a preset iterative requirement is met. According to the method, the problem of data resource islands is solved, and more effective data can be aggregated and obtained; the data is used locally, and the model is trained locally, so that the data security is effectively ensured.

The invention discloses a power transmission line lightning classification method, which comprises the following steps: acquiring an original image, and cutting by using a lightning target cutting method to obtain a lightning target image; selecting different background images in the original image, splicing the lightning target image with the different background images, and performing first preprocessing on the spliced image to obtain a background image; modeling the background image by using a Gaussian mixture model to obtain a background model, and calculating the matching degree between each pixel of the background model and each Gaussian model distribution to obtain a foreground; judging the foreground through a threshold cutting method to obtain a lightning region, and performing second preprocessing on the lightning region to obtain a first lightning region; and inputting the first lightning region into a deep learning model for training and classification reasoning to obtain a confidence score, and marking the lightning type and the hazard level according to the confidence score to obtain a lightning classification result, thereby meeting the requirements of power transmission line monitoring safety guarantee.

The invention relates to the technical field of remote sensing image road extraction, and provides a remote sensing image road extraction method and device based on knowledge distillation, and a readable storage medium, and the method comprises the steps: constructing a teacher model and a student model, employing an encoder-decoder architecture to learn road features for neural networks of the teacher model and the student model, the number of layers of the student model neural network is smaller than that of the teacher model neural network, a middle layer is adopted for knowledge distillation, knowledge distillation of the middle layer is isomorphic distillation, and a result is output based on the model after target training. The problems that a traditional method is large in parameter quantity and high in calculation complexity are solved, real-time reasoning can be completed rapidly and accurately, and the requirement for real-time road extraction is met.

The invention discloses a method for obtaining the thickness and refractive index of a film, a test system, a storage medium and application. The method for obtaining the thickness and the refractive index of the thin film comprises the following steps: obtaining an intermediate frequency signal carrying transmission phase information by using two groups of nonlinear transmission line terahertz frequency comb generators and a broadband mixer, and obtaining an actually measured phase curve based on the intermediate frequency signal; constructing an equivalent multi-optical-path interference model of the thin film, performing coherent superposition on multiple reflection components of a measured wave signal in the thin film in a complex field to obtain a complex phase of a total transmission field, and obtaining a phase-frequency change spectrum based on the complex phase as a model phase curve; and an error function of the model phase curve and the actually measured phase curve is constructed, and a parameter combination enabling the error function to take the minimum value is solved by scanning the parameter combination of the refractive index n and the thickness d of the thin film. According to the method, thermal drift or time drift of the phase is avoided, and then calculation errors caused by phase jump are avoided.

The invention discloses a power line positioning method, device and equipment based on an airborne radar and a storage medium, the method simplifies an algorithm for a specific scene of low-altitude unmanned aerial vehicle obstacle avoidance, and the method can obtain the distance between an unmanned aerial vehicle and a power line without modeling the power line. And the operation speed and the positioning precision are relatively high. Based on a mathematical statistics method of a probability statistics histogram, a natural breakpoint method is used for automatically searching an optimal height threshold value for ground filtering, and excessive segmentation of a power line caused by a fixed height threshold value in a complex terrain is avoided. According to the feature of noise point distribution of the point cloud data, the DBSCAN algorithm is adopted to remove noise points, compared with a single filtering method, the DBSCAN algorithm has high adaptability, clustering can be carried out on the point cloud data, core points, boundary points and noise points in the point cloud data are classified, obvious outliers are removed, and the accuracy of a final calculation result is guaranteed.

The utility model discloses a screen limiting device and glue laminating equipment, and belongs to the technical field of screen laminating processing, the screen limiting device comprises a lower limiting device, a monitoring assembly, a clamping device, a control device, a turnover device and an upper limiting device, an upper screen is mounted above the upper limiting device, a lower screen is mounted in the lower limiting device, and the lower screen is mounted in the lower limiting device. An upper limiting device is arranged on the right side of the control device, a lower limiting device is arranged above the monitoring assembly, the overturning camera is used for monitoring the glue attaching condition, the lower limiting devices are arranged on the left sides of the clamping devices in pairs and used for clamping and positioning, the lower limiting devices are arranged above the control device, and the upper limiting device is arranged above the overturning device. The problems that in the prior art, due to the fact that glue is low in viscosity, an upper screen, a lower screen and a glass plate are prone to sliding after being closed in the attaching process, the upper plate and the lower plate move, and then the product consistency is affected are solved. The method has the advantages of improving the product consistency, increasing the product percent of pass and improving the production efficiency.

The invention discloses a double-sided gluing machine for foam products. The double-sided gluing machine includes a foam jig, a gluing jig, a swing cylinder and a base; the upper and lower ends of the reference base surface of the foam jig are symmetrically connected to a plurality of lateral positioning pins, the rear side of the foam jig is connected to the swing cylinder, the swing cylinder is fixed in front of the vertical wall plate of the base, the gluing jig is a horizontal plate, and the upper and lower sides are respectively the gluing base surfaces, the two gluing base surfaces are symmetrically connected to a plurality of vertical positioning pins, the middle opening of the rear side of the gluing jig forms an accommodating cavity, the positioning feet on the left and right sides of the accommodating cavity are provided with magnetic elements, the front side of the left end of the gluing jig protrudes to form a rotating ear, and the bearing is fixed on the rotating ear; the rotating seat is fixed in front of the left end of the foam jig, the rotating cavity on the front side of the rotating seat fixes the vertical hinge shaft, and the bearing sleeve is in the vertical hinge shaft. The invention has the effects of fast laminating speed, high production efficiency, double-sided laminating position accuracy, and high yield rate.

The present invention relates to the technical field of road extraction from remote sensing images. The present invention provides a method, device and readable storage medium for extracting roads from remote sensing images based on knowledge distillation, which includes constructing a teacher model and a student model. The neural networks of the teacher model and the student model adopt an encoder-decoder architecture to learn road features. The number of layers of the student model neural network is less than that of the teacher model neural network. The knowledge distillation is performed in the middle layer, and the knowledge distillation in the middle layer is isomorphic distillation. The model outputs the result based on the target training. The problem of large number of parameters and high computational complexity of the traditional method is solved, and real-time reasoning can be completed quickly and accurately to meet the needs of real-time road extraction.

The invention discloses a felt opposite seam press-fit device, and relates to the technical field of automobile manufacturing, the felt opposite seam press-fit device comprises a pillar, the top of the pillar is fixedly connected with a top plate, the bottom of the pillar is fixedly connected with a bottom frame, the two sides of the pillar are fixedly connected with extension plates, the upper surface of the bottom frame is fixedly connected with a supporting rod, and the top end of the supporting rod is inserted with two press-fit tables; and a gap is formed between the two pressing tables. The felt cloth is pushed to move downwards through downward movement of the press-fit pad, so that the felt cloth is attached to the automobile roof plate on the press-fit table, the felt cloth can adapt to the shape of the automobile roof plate to be integrally attached in cooperation with a soft rubber material of the press-fit pad, the attaching speed is higher, and the attaching efficiency is improved. And meanwhile, the soft rubber material of the press-fit pad can adapt to the automobile ceiling plate in any shape, so that the pressure borne by the felt cloth is more uniform, the press-fit table corresponds to the automobile ceiling plate in shape, the automobile ceiling plate can be supported, and excessive press-fit of the automobile ceiling plate is avoided.

The invention relates to the technical field of physiotherapy instruments, and discloses a waveform output method and device for a pulse signal of a physiotherapy instrument and electronic equipment, and the method comprises the steps: obtaining the pulse signal of the physiotherapy instrument, splitting each continuous waveform in the pulse signal into a pulse signal segment, and analyzing each pulse signal segment, the method comprises the steps of obtaining a plurality of feature points of each pulse signal segment, fitting all the feature points of each pulse signal segment based on a functional expression in a preset function library to generate a fitting function of each pulse signal segment, and determining an envelope waveform of the functional expression corresponding to each fitting function as an envelope waveform of the corresponding pulse signal segment, and outputting the fitting function and the envelope waveform of each pulse signal segment according to the sequence of each pulse signal segment in the pulse signal. Through the mode, the waveform of the pulse signal of the physiotherapy instrument can be rapidly and accurately output, so that a user can conveniently research the physiotherapy instrument.

This invention discloses a method for preparing a long-lasting stain-resistant fabric, relating to the field of fabric technology. The long-lasting stain-resistant fabric prepared by this invention comprises, from the inside out, Lycra fabric, foamed rubber, and a stain-resistant film. The stain-resistant film is prepared by reacting hyperbranched polyglycidyl ether with modified cyclodextrin. The modified cyclodextrin is obtained by grafting a hydroxylamine polymer onto mercapto-carboxymethyl-cyclodextrin, which enhances stability and introduces a large number of hydroxyl groups, enhancing hydrophilicity and giving the fabric stain resistance. It also increases the density of the stain-resistant film, thereby enhancing the fabric's tear resistance. The foamed rubber is prepared by vulcanizing and foaming a blend of chloroprene rubber, polyurethane rubber, and fluorinated polyether nitrile. The addition of polyurethane rubber accelerates the vulcanization speed and improves the tear resistance of the foamed rubber. It also tightly bonds the Lycra fabric, foamed rubber, and stain-resistant film, enhancing the fabric's peel strength and extending its service life.

This invention relates to the field of defect detection technology, specifically to a method for improving the YOLOv8 network and its application in strip steel surface defect detection. Based on YOLOv8, this invention first proposes an improved coordinate attention mechanism, the three-channel coordinate attention mechanism TCCA. Addressing the overly simplistic offset mask generation method in DCNv2, which leads to insufficient deformation modeling capabilities, this invention proposes deeply embedding TCCA into DCNv2. Furthermore, this invention introduces a global attention mechanism to improve the model's ability to extract global features. The invention also introduces a BiFPN structure and dynamic serpentine convolution to enhance the model's multi-scale feature fusion capabilities. MDPioU replaces the original loss function of YOLOv8, solving the problem of loss of effectiveness due to identical aspect ratios in predicted bounding boxes, while increasing convergence speed and localization ability. Extensive comparative and ablation experiments on the NEU-DET dataset demonstrate that the improved algorithm of this invention achieves higher defect detection accuracy and faster speed.

The utility model discloses an electronic equipment film pasting assistor which comprises a positioning base, a plastic uptake positioning frame is arranged on the positioning base, a gasket is further arranged on the top face of the positioning base, a positioning block is arranged at one end of the positioning base, and a film drawing reserved opening is formed in the inner wall of the end, away from the positioning block, of the plastic uptake positioning frame. A receding groove corresponding to the film-drawing reserved opening is formed in the section, close to the film-drawing reserved opening, of the positioning base, the plastic uptake positioning frame is integrally arranged on the positioning base, a screen film and electronic equipment needing film pasting can be aligned, positioned and fixed, and in the film pasting process, the film pasting efficiency is greatly improved. The film drawing reserved opening is formed in one end of the positioning base, so that a protective film of the screen film can be conveniently drawn out from the screen film, the screen film and the electronic equipment screen are attached more tightly through the cross-shaped gasket arranged on the top face of the positioning base, and meanwhile the effect of increasing the attaching speed of the screen film and the electronic equipment screen is achieved through the gasket.

The invention belongs to the technical field of aeronautical manufacturing, and provides an electric fitting processing intelligent manufacturing production efficiency evaluation method, which comprises the following steps of: according to production efficiency attributes, refining production quality, production cycle and production cost into production efficiency sub-characteristics; taking the production efficiency sub-characteristics as metric elements to form a metric element set; respectively establishing production efficiency sub-characteristic evaluation models for production quality, production cycle and production cost by using the pigeon inspired improved RBF network; and evaluating the intelligent manufacturing production line of the airborne unit based on the evaluation model, and predicting the production efficiency after the intelligent reconstruction of the production line. According to the electric fitting processing intelligent manufacturing production efficiency evaluation method provided by the invention, the production efficiency of the production line can be predicted before the construction of the production line is successful, so that the design of the scheme is prevented from being inconsistent with the expectation.