9results about How to "Effective focus" patented technology

The invention relates to the technical field of medical instruments, in particular to an electromagnetic ultrasonic device for treating osteoporosis, which comprises a bed body, an electromagnetic ultrasonic probe, a data acquisition module, a data processing module and an artificial intelligence analysis module, an ultrasonic treatment assembly and an electromagnetic treatment assembly are integrated in the electromagnetic ultrasonic probe; the ultrasonic treatment assembly emits ultrasonic waves and transmits the ultrasonic waves to bones through soft tissues for ultrasonic treatment; the electromagnetic treatment assembly generates an alternating electromagnetic field to stimulate bone cells; physiological signal data are analyzed in real time through an artificial intelligence algorithm, physiological state changes are predicted, output parameters of PEMF and LIPUS are automatically adjusted to adapt to individual physiological differences, and personalized treatment is achieved.

The invention discloses a fault analysis and positioning method and system based on log data, and relates to the technical field of log data analysis and fault positioning, and the method comprises the steps: collecting first log data, and carrying out the preprocessing; identifying a log mode through a first identification method, and establishing a basic mode multi-dimensional vector of the abnormal log; and through a first log analysis tool, comparing the real-time log data with a historical baseline, identifying abnormity and carrying out fault positioning. According to the method, the log modes with similar characteristics are recognized and the multi-dimensional vector representation is established through the application of the first recognition method, so that the abnormal log modes can be accurately recognized, the accuracy of anomaly detection is improved, the false alarm rate is reduced through the comparison of angular distance calculation and the preset threshold value, and the accuracy of anomaly detection is improved. By using the first log analysis tool, the real-time log data can be compared with the historical base line, so that the abnormity is identified and the fault is positioned.

The present invention provides a novel polishing composition comprising a water-soluble polymer containing at least a vinyl alcohol resin having a side chain group having 3 or more carbon atoms.

The invention relates to a half-wave oscillator antenna device based on a reflection cavity structure and an adjusting method thereof, the device comprises a copper-clad plate reflection cavity, a half-wave oscillator antenna, a microstrip feeder line, a copper-clad plate, an SMA interface and a reflection cavity bottom surface, the copper-clad plate reflection cavity is connected with the peripheral edge of the reflection cavity bottom surface to form a semi-closed cavity with an opening facing one side; the bottom surface of the reflection cavity is provided with a through hole matched with the SMA interface, one side of the copper-clad plate is connected with the bottom surface of the reflection cavity, the half-wave oscillator antenna and the microstrip feeder line are both arranged on the surface of the copper-clad plate, and one end of the half-wave oscillator antenna is connected with the SMA interface through the microstrip feeder line. Compared with the prior art, the antenna has the advantages that the structure is simple, wave beams can be more concentrated under the condition that the bandwidth of the antenna is not affected, and therefore the high-gain effect of the antenna is achieved.

The application provides an open vocabulary target detection method and system, which comprises the following steps: integrating a learnable prompt anchor module in a multi-scale feature layer of a preset cross-modal feature fusion network; injecting each prompt anchor point vector into a corresponding feature map through residual connection to dynamically correct the multi-scale semantic bias of the feature map; in the model training process, using a frozen CLIP visual encoder as a teacher model and the preset cross-modal feature fusion network as a student model to train a corresponding small target region mask, and calculating the distillation loss of the dynamically corrected feature map according to the small target region mask; outputting the multi-scale features corresponding to the feature map according to the distillation loss, and performing boundary box regression and open vocabulary classification on the multi-scale features through a preset detection head to complete the corresponding target detection. The application can effectively improve the detection efficiency of small region targets.

The utility model provides a kind of hidden antiskid acrylic bathtub belongs to the field.It solves the inner wall of traditional bathtub is often too smooth, and user is difficult to find the position that can lend force to steady posture or buffer impact in the moment of slipping, in the sliding process without any obstruction, body will be uncontrollably directly impact on bathtub wall, and further easily lead to the problem of user being injured in different degrees.This hidden antiskid acrylic bathtub includes cylinder body and faucet for water discharge installed at the edge of cylinder body, a plurality of knobs for controlling the water outlet of faucet are provided on one side of faucet, water outlet hole for water outlet is provided in cylinder body, two symmetrical anti-skid grooves are provided on the inner wall of bathtub, and the anti-skid grooves are arc-shaped.The utility model has the advantages of safety and anti-skid.

This invention discloses an autofocus method, apparatus, electronic device, and storage medium, belonging to the field of autofocus technology. The autofocus method includes: adjusting the focal length within a first focusing range according to a first preset step size, acquiring a first image captured by the camera at each focal length, and determining an effective focusing area and a second focusing range based on the first image; adjusting the focal length within a second focusing range according to a second preset step size, acquiring a second image captured by the camera at each focal length, and determining the optimal focusing position based on the sharpness evaluation value of the effective focusing area in the second image; wherein the first preset step size is greater than the second preset step size. This invention performs focusing in two stages: the first stage uses a large step size to filter the effective focusing area and narrow the focusing range, and the second stage uses a small step size for focusing and uses fewer but more effective areas for focusing, thus achieving higher operating efficiency.

The invention relates to the technical field of radars, and particularly discloses a low-slow small target micro-motion feature extraction algorithm based on STFT-LSTM. The method comprises the following steps: converting an original signal into time-frequency domain representation through STFT, and constructing a feature matrix with time sequence relevance; a bidirectional LSTM network is introduced to carry out time sequence dependence modeling, adaptive fusion of a feature space is realized in combination with a multi-head attention mechanism, and the calculation efficiency is remarkably improved through a learnable weight distribution strategy. And designing a feature visualization analysis module to verify the performance of the model, quantitatively evaluating the mapping relation between the feature parameters and the classification accuracy, and finally forming a dynamic optimization closed loop. According to the technical scheme, a new technical path is provided for multi-mode radar echo signal processing, the precision and efficiency of feature extraction are effectively improved, and the method is suitable for a target recognition task in a broadband complex electromagnetic environment.

The present invention proposes a method and system for constructing a global call chain of a microservice cluster based on edge computing, and relates to the field of communication technology. The method includes: obtaining a microservice cluster topology map, locating edge nodes based on data sources; collecting call chain data in the microservice cluster topology map, and transmitting it to edge nodes; the edge nodes use a trained decision tree model to filter out high-priority data in the call chain data for edge computing, build local call chains, and upload them to the central node; the high-priority data includes abnormal requests and high-latency requests; the central node aggregates local call chains to form a global call chain. The present invention deploys a lightweight machine learning model on the edge node, and based on historical data analysis, filters the data that constitutes the local call chain, and then the central node receives and merges the local call chains from each edge node to form a complete global call chain view, effectively reducing the burden of data transmission.