7results about How to "Show validity" patented technology

The invention discloses a hypersonic flight vehicle adaptive attitude control method based on near-end strategy optimization, and belongs to the technical field of hypersonic flight vehicle control, and the method comprises the steps: building an air-breathing hypersonic flight vehicle longitudinal dynamic model considering power; performing reinforcement learning training on the obtained environment model by using a PPO algorithm to obtain an optimized control strategy; selecting a proper state representation and action space, and designing a strategy network and a value network for the processed model, so that an original complex attitude control problem is converted into a reinforcement learning problem which can be effectively learned; and training the reinforcement learning model by using an improved PPO algorithm to obtain a final control instruction. According to the method, the model is trained by using the PPO algorithm, so that the defects of complex controller design, strict requirement on model precision and the like of a traditional control method can be effectively overcome. A simulation experiment verifies that pitch angle tracking can be effectively realized, and the stability of an execution mechanism is guaranteed.

The present invention relates to a deep learning time-frequency representation method for radar micro-motion features, and belongs to the field of radar signal processing technology. The present invention first pre-processes radar echo data to obtain target information data containing a distance dimension; then designs a deep learning network model that represents the micro-motion features of radar targets, and the model includes four parts: a basis function selection module, a feature aggregation module, an energy concentration module, and an output module. A data set is constructed to train and verify the model; finally, the pre-processed measured target data is input to complete the time-frequency representation of the target micro-motion features based on a deep learning method. The present invention can achieve high-resolution micro-motion feature representation of radar targets. Compared with traditional time-frequency analysis methods, it not only shows the advantage of a high degree of energy aggregation, but also shows the details of micro-motion features that traditional time-frequency analysis methods fail to show, thereby improving the representation capability of the target's micro-motion features.

This invention discloses a feature denoising distillation method guided by Logit information, belonging to the field of model compression technology. The method includes: first, encoding the discriminative direction of the teacher classifier using a Logit discriminative encoder to ensure feature alignment along the semantic direction; then, constructing a feature denoising mask to emphasize the teacher's target attention region while suppressing redundant background responses, thereby achieving spatial-level denoising in the feature space. Finally, a feature denoising distillation framework guided by Logit information is designed to achieve model compression. This invention provides explicit semantic direction guidance through a Logit discriminative encoder, suppresses background and noise activation through a feature denoising mask, and integrates both into the intermediate feature distillation process between the teacher and student models. It systematically solves the problems of lack of Logit semantic guidance, severe background noise interference, and insufficient focus of the distillation target in existing technologies, achieving high-precision and highly compatible model compression and knowledge transfer.

This disclosure provides a method and system for enhancing low-exposure images based on embedded McLaurin and convolution, relating to the field of image processing technology. The method involves acquiring a low-exposure image to be enhanced, processing and decomposing the low-exposure image, inputting the low-exposure image into a backbone network to extract multi-scale features, introducing a dynamic adjustment factor into the backbone network, using the dynamic adjustment factor to learn a convolution weight calibration map, calibrating the convolution weights of the backbone network with the learned weight calibration map, performing convolution operations on the features using the calibrated convolution kernels to obtain the luminance component of the low-exposure image, and outputting a long-exposure enhanced image that varies with the dynamic adjustment factor using the luminance component.

This invention discloses a AIPL1 A gene mutation-related congenital amaurosis animal model, its construction method, and its application. The method includes: S1, constructing a humanized... AIPL1 S2, a homologous recombination vector with a point mutation of c.421 C>T (p.Q141X) knocked in; S3, Cas9 mRNA, gRNA1, gRNA2 and the homologous recombination vector are injected into animal zygotes to perform homologous recombination, resulting in transfected zygotes; S4, the transfected zygotes are transplanted into the uterus of pseudopregnant animals to reproduce offspring, and offspring are selected from the offspring. AIPL1 Homozygous animals with the c.421 C>T (p.Q141X) point mutation were obtained, thus yielding the aforementioned... AIPL1 A gene mutation-related congenital amaurosis animal model. The animal model of this invention exhibits the clinical phenotype of Leber congenital amaurosis type 4 (LCA4) from 12 days after birth; while heterozygous mice do not show a significant phenotype, consistent with... AIPL1 The clinical phenotype and inheritance pattern of LCA caused by the c.421 C>T (p.Q141X) gene mutation demonstrate the effectiveness and reliability of the animal model of this invention, providing a basis for understanding the genetic model of LCA. AIPL1 Animal models provide a reliable basis for studying the disease mechanisms and drug screening of mutation-related LCA.

This invention discloses a missing modality sentiment analysis method based on learnable category-aware prototype completion, belonging to the field of multimodal sentiment recognition technology. This method addresses the problem of modality missingness in multimodal sentiment analysis and employs a two-stage training strategy: In the first stage, under complete modality conditions, a single-modality discriminative representation is learned through a modality expert network, and a prototype memory is constructed using learnable category-aware prototypes; in the second stage, under modality missing conditions, a cross-modal joint retrieval mechanism is used, leveraging consensus information from known modalities as queries, to retrieve semantically aligned prototypes from the prototype memory to complete the missing modality. This invention achieves missing modality completion through category-aware prototypes, reducing the semantic gap problem in generative methods and significantly improving the robustness of the model in modality missing scenarios. Experimental results on benchmark datasets demonstrate that this invention achieves excellent performance under various modality missing conditions.

This invention discloses a multipath ghosting elimination method for indoor target localization, comprising the following steps: Step 1, initial target localization; Step 2, mirror target position calculation; Step 3, theoretical position calculation of single-path first reflection + single-path LOS path; Step 4, theoretical position calculation of single-path LOS path + single-path first reflection; Step 5, multipath position matching and elimination. The method disclosed in this invention utilizes millimeter-wave radar to perform multipath elimination and subsequent localization of targets in an indoor multipath environment. Based on the propagation phenomena of electromagnetic waves, the main propagation paths of electromagnetic waves in indoor multipath scenarios are analyzed, and the theoretical positions of combined paths and two-way first reflection paths are calculated. This removes the multipath positioning points from the final target calculation results, obtaining the true target positioning trace. Experimental results demonstrate the accuracy of the method in locating true targets and the effectiveness of multipath elimination.