8results about How to "Improve feature utilization" patented technology

The invention discloses a harmonic suppression and dynamic reactive compensation control method for an energy router, which is suitable for a micro-grid and a distributed energy access system. According to the method, three-phase current and voltage signals of an alternating current port of the energy router are collected, five-time and seven-time main harmonic components are extracted through fast Fourier transform, harmonic compensation current instructions equal in amplitude and opposite in phase are generated and injected into an inverter to be output, and active counteracting of main harmonics is achieved. Meanwhile, the bus voltage deviation and the change rate thereof are detected, a reactive current instruction is adjusted on line by adopting a fuzzy self-adaptive PI controller, rapid and smooth dynamic reactive compensation is realized, and the bus voltage is stabilized. Decoupling control of active / reactive and harmonic components is realized in combination with Park conversion, and virtual impedance is introduced to enhance system damping and harmonic suppression capabilities. Compared with the prior art, harmonic suppression and voltage support can be simultaneously realized under a single control framework, the response speed is high, the stability is high, and the method is suitable for high-performance grid-connected operation of the multi-port energy router.

A quantitative diagnosis method for freshwater fish diseases based on OpenCV weak supervision and ResNet50-ADCA dual-branch model is proposed. The method consists of the following steps: (1) Image acquisition and preprocessing: Based on the publicly available freshwater fish disease dataset, a balanced Salmon++ dataset is constructed by combining samples collected from aquaculture sites; (2) Weakly supervised lesion segmentation and labeling: (2.1) HSV color space targeted segmentation; (2.2) Iterative optimization mechanism; (2.3) Disease grading standard; (3) ResNet50-ADCA dual-branch model construction: (3.1) Backbone network and transfer learning; (3.2) SEA channel attention embedding; (3.3) Dual-branch multi-task output layer design; (3.4) Classification branch; (3.5) Grading branch; (3.6) Loss function design; (4) Model training and optimization. This invention balances labeling efficiency, diagnostic accuracy, and quantitative grading capability, and can effectively solve the problems of high labeling cost, poor adaptability to complex environments, lack of quantitative evaluation function, and imbalance between accuracy and speed in the existing technology.

The application provides a space target recognition method and device based on an attention mechanism adaptive enhancement. The method comprises the following steps: acquiring a space target image; inputting a tensor of the space target image into a pre-trained space target recognition model to obtain a recognition result of the space target. For any one processing stage submodel, a tensor processing module set unit and a convolution block attention module are included. In the tensor processing module set unit, a plurality of tensor processing modules are connected in sequence, the tensor processing module is used for convolution and channel transformation of the input tensor, and / or transformation processing of the tensor element value. According to the technical scheme of the application, the tensor processing module is used for convolution and channel transformation of the input tensor, and / or transformation processing of the tensor element value, the accuracy of the final recognition result can be significantly improved, the convolution block attention module realizes channel screening, is conducive to focusing on key features, and improves the feature utilization rate.

This invention discloses a modified children's dietary fiber based on microcapsule encapsulation and ultra-micro homogenization emulsification and its preparation method, which is prepared according to the following steps: (1) Raw material pretreatment: plant fiber is crushed; (2) Enzymatic modification: the fiber powder is prepared into a fiber suspension slurry, cellulase is added for enzymatic hydrolysis, and the modified fiber slurry is obtained after enzyme inactivation and cooling; (3) Composite wall material preparation: maltodextrin or whey protein and gum arabic are weighed, dissolved in deionized water, and natural antioxidants are added; (4) Homogenization emulsification and microcapsule encapsulation: the modified fiber slurry obtained in step (2) is dispersed in the wall material solution in step (3) as the core material, and homogenized using a two-stage gradient high pressure treatment. After homogenization, the product is obtained by spray drying or fluidized bed encapsulation and solidification. Through nanoscale reshaping and directional encapsulation, the rough taste of dietary fiber is completely eliminated, unpleasant flavors are masked, and natural antioxidants are used to improve the stability and intestinal bioavailability of the product.

The application discloses a kind of based on multi-view fusion's occlusion target detection method and system, belong to computer vision technical field, it is applicable to intelligent monitoring, automatic driving etc. Scene.For the problems of inaccurate cross-view alignment, inefficient fusion, fuzzy occlusion recognition and poor real-time performance in the prior art, the present application synchronously collects images and calibrates parameters through at least 2 devices;After adaptive filtering and distortion correction preprocessing, enhanced features are extracted using improved ResNet50 containing CBAM;Based on Kalman filtering dynamic alignment cross-view features, identify the occlusion area by combining feature missing rate, similarity and multi-dimensional index of depth difference;Optimize features using hierarchical fusion strategy and input improved YOLOv8 containing occlusion branch for detection.The alignment error of the present application is not more than 2 pixels, the occlusion recognition accuracy is not less than 92%, the detection accuracy is not less than 95% and the frame rate is not less than 20fps, effectively solving the bottleneck of occlusion scene detection, adapting to latency-sensitive applications.

This invention relates to the field of medical assistance technology, specifically to a pneumonia medical assistance system based on an improved VMamba model. The system includes: dividing the lung image to be diagnosed into raster cells based on a raster cell partitioning algorithm; calculating the lesion representation coefficient of each raster cell within each management time period; calculating the lesion representation index of the lesion trend path within the management period based on the raster cells along the lesion trend path; determining whether the raster cells covered by the lesion trend path are lesion regions based on the lesion representation index; inputting the lung field feature map and semantic feature vector into the improved VMamba model after channel-level concatenation, and outputting the pneumonia classification diagnosis result; and linking the pneumonia classification diagnosis result with a pneumonia knowledge graph to generate a visual auxiliary diagnosis report, significantly improving the accuracy and stability of lung medical image processing and achieving full-process automation of intelligent image analysis.

This invention provides a method and related equipment for measuring the temperature inside a top-blown tin smelting furnace. The method involves reorganizing one-dimensional time-series data from a source domain dataset to obtain a two-dimensional dynamic matrix. A multi-scale parallel convolutional architecture is then pre-trained to obtain a first furnace temperature prediction model. Knowledge transfer is performed on this first model using a target domain dataset to obtain a second furnace temperature prediction model. Real-time collected operating data of the top-blown tin smelting furnace is compared with historical operating data in the target domain dataset to obtain a fine-tuned dataset. This fine-tuned dataset is then used to fine-tune the second furnace temperature prediction model to obtain a third furnace temperature prediction model. Historical operating data of the target top-blown tin smelting furnace is input into the third furnace temperature prediction model for prediction, yielding a future furnace temperature prediction. This method can improve temperature measurement accuracy even when labeled data in the target domain is extremely scarce.

The invention discloses a low-illumination image enhancement method based on gradient perception and intermediate frequency guidance, which mainly constructs a double-branch encoder-decoder network model in an HVI color space so as to effectively recover image texture details, suppress noise and improve color fidelity under a low-illumination condition. Firstly, an RGB image is converted into an HVI color space, and two branches of HV and I are separated; secondly, in the encoding process of an encoding layer, a gradient guide channel attention module is adopted to enhance detail perception, and meanwhile, deformable cross-branch fusion attention is adopted between middle-layer features and in a decoding layer to achieve cross-branch dynamic information interaction; in a bottleneck layer, an intermediate frequency guided multi-scale frequency domain enhancement module is adopted to perform intermediate frequency guided amplitude feature enhancement and phase maintenance in a frequency domain; and finally, reconstructing the enhanced HVI image through a symmetric decoder, carrying out residual connection on the HVI image and the initial HVI image, and inversely transforming the HVI image back to the RGB space to obtain a recovered image. Compared with the prior art, the method has better image visual quality.