349results about How to "Reduce computational overhead" patented technology

The invention relates to an Internet of Vehicles multi-agent cooperative sensing method and system. The method comprises the following steps: constructing a collaborative sensing network, wherein the collaborative sensing network comprises a self-agent and a plurality of collaborative agents; acquiring and processing sensing data through the collaborative sensing network; performing feature extraction to obtain intermediate features; self-adaptive sparsification is carried out to obtain sparse features, and the sparse features are compressed and transmitted to a self-agent; performing time sequence feature enhancement on the features of all the agents at the self-agent end; fusing the features to obtain fused features; and constructing an Internet of Vehicles perception model, and realizing perception by the detection model according to the fused features. According to the method, the calculation complexity of traditional global attention is reduced from the square level to the linear level through an adaptive sparsification mechanism, the calculation overhead is remarkably reduced while the multi-agent feature interaction precision is kept, and the method is more suitable for real-time operation on the vehicle-mounted edge equipment with limited resources.

The invention discloses a lightweight multi-mode lower limb motion intention recognition method and system, and relates to the technical field of biomedicine. The method comprises the steps that multi-channel surface electromyogram signals sEMG of a target lower limb and joint angle signals and joint torque signals of lower limbs on the same side are synchronously collected; performing feature extraction by using a double-branch structure to obtain muscle-related deep time sequence features and joint-related high-level semantic features; performing feature fusion through a bidirectional cross attention mechanism to obtain cross attention fusion features; and performing flattening, nonlinear mapping, regularization and Softmax classification on the cross attention fusion features, and outputting the motion intention of the target lower limb. Through double-branch input, depth feature extraction and a bidirectional cross attention mechanism, on the premise of ensuring recognition precision, a lightweight attention module, a residual structure and a cross-modal interaction mechanism are introduced, and the parameter quantity and calculation overhead are reduced.

The invention belongs to the technical field of software engineering, particularly relates to an automatic driving system performance enhancement method based on an expert hybrid architecture, and aims to solve the core problems that an end-to-end automatic driving system is confronted with semantic fuzziness to cause unreliable decision, multi-task interference hinders optimization planning, too long reasoning delay increases driving risks and the like. According to the method, an ExpertAD framework is provided, task key features are amplified through a perception adapter (PA), and the relevance of scene context understanding is guaranteed; related driving tasks are dynamically activated through a sparse expert mixture (MoSE), and task interference is minimized; and in combination with a customized training loss function, collaborative optimization of planning effectiveness and reasoning efficiency is realized. Experiments show that compared with an existing method, the method has the advantages that the average collision rate is reduced by 20%, the reasoning delay is reduced by 25%, higher multi-skill planning capacity is achieved in rare scenes (such as accident handling and first-aid vehicle avoiding), and good generalization is achieved for unseen urban environments.

The invention discloses a two-stage space-time diagram network and multi-source information fusion time sequence probability prediction method and device, and relates to the technical field of artificial intelligence and big data analysis. The method comprises the following steps: acquiring historical wind speed sequence data and corresponding target variable sequence data; preprocessing the data, and constructing a training sample according to a preset time sliding window; constructing a time sequence probability prediction framework of the two-stage space-time diagram network and multi-source information fusion; based on the fluctuation correlation of the historical sequence data, constructing an adjacent matrix between nodes; inputting the training sample and the adjacent matrix between the nodes into a wind speed prediction model, training the model through a designed threshold perception loss function of a wind speed-power nonlinear relationship, and outputting a multi-node wind speed prediction probability of a first stage; and inputting the multi-node wind speed prediction probability and the multi-source environmental factor data into the gradient boosting tree model for second-stage prediction, and outputting a multi-node wind power prediction result. According to the invention, prediction errors can be reduced.

The invention discloses a cross-city traffic prediction method and system based on multi-modal fusion and spatial expert routing, relates to the technical field of traffic prediction, and provides an adaptive modal selection mechanism based on a signal-to-noise ratio for the problems of missing, noise and uneven quality of multi-modal data in different cities. And a low-quality mode is dynamically suppressed in combination with comparative learning, and robust multi-mode fusion is realized. In order to solve the problems of large space structure difference and weak generalization ability among cities, a multi-modal guided space expert routing architecture is designed: modal sharing experts and routing experts are activated by using a multi-modal context, and local space dependence is adaptively modeled for different functional regions. The method supports any modal combination input, city specific fine tuning is not needed, and the zero sample cross-city prediction performance is significantly improved.

The application provides a logistics transportation management method and system based on artificial intelligence, relates to the field of logistics transportation, and solves the technical problems that the prior art is difficult to balance real-time performance, distribution feasibility and resource utilization efficiency in the sudden interruption scene of a logistics vehicle, resulting in slow scheduling response, frequent secondary conflicts or failure of high-value order fulfillment. The method comprises the following steps: obtaining a set of orders to be delivered by a target vehicle at present and real-time position information; based on a preset hard constraint condition, screening candidate vehicles and constructing a task migration feasibility graph; inputting the task migration feasibility graph into a pre-trained lightweight graph neural network model to output the compatibility scores of each candidate vehicle for each order to be delivered in the order set; based on the compatibility scores, evaluating the urgency of the order to be delivered; and based on the urgency of the order to be delivered, updating the delivery task scheme of each candidate vehicle. The application is used in the process of logistics transportation.

The application provides a cloud game evaluation method and medium based on space-frequency enhancement and sparse attention fusion, and belongs to the technical field of image processing. The method comprises the following steps: acquiring a cloud game video stream to be evaluated, and extracting key frames and key region image blocks from the cloud game video stream; inputting the key frames into a pre-trained global feature extraction network to obtain global features, global shallow layer features, global middle layer features and global deep layer features; and inputting the key region image blocks into a pre-trained local feature extraction network to obtain local features, local middle layer features and local deep layer features. The method of the application can precisely capture global scene distortion and local high-sensitive region distortion of a cloud game video by constructing a multi-stage sparse fusion network and a space-frequency multi-level fusion network, while reducing model parameter quantity and calculation cost, meeting real-time inference and terminal lightweight deployment requirements of a cloud game scene.

The application relates to a point cloud three-dimensional target detection method and system based on a local attention Mamba model, which comprises the following steps: obtaining a key point cloud subset from a point cloud; inputting the key point cloud subset into a local attention Mamba model composed of N hierarchical stacks, and sequentially performing the following steps on each level: for each point, obtaining a neighborhood point set through a hash table index, generating local attention weights based on element-by-element multiplication interaction, and weighting and aggregating neighborhood features to obtain local geometric features; then modeling global long-range dependencies of the local geometric features by using a bidirectional Mamba module, and outputting global features as the input of the next layer; and inputting the global features output by the last level into a bird's eye view backbone network to obtain a three-dimensional target detection result. The application forms a progressive feature learning mechanism of local perception and global enhancement, and significantly improves the expression ability and reasoning efficiency of a point cloud backbone network.

This invention discloses a video tag dynamic overlay anti-shake method and system based on AR, comprising: acquiring real-time video stream and device pose parameters at corresponding timestamps; constructing a dynamic perspective projection model based on the device pose parameters and calculating the initial pixel coordinates of the target AR tag in the current video frame; planning a local feature search window based on the initial pixel coordinates; calculating the average pixel displacement vector of image feature points in the local feature search window of the current and adjacent historical video frames; performing reverse displacement compensation on the average pixel displacement vector to obtain the target rendering coordinates, and rendering the target AR tag to the current video frame based on the target rendering coordinates. Compared with the prior art, this invention can overcome the shortcomings of large drift errors caused by simply relying on mechanical pose, high computational power consumption caused by global matching, and high-frequency jitter of tags caused by slight device vibrations, thereby improving tag mapping accuracy, reducing system computational power consumption, and improving the visual stability of video overlay.

The embodiment of the invention discloses a routing method and device for data transmission, equipment, a medium and a program product, which are used for simplifying the data routing complexity, reducing the calculation overhead and improving the data routing performance in a high dynamic environment. The method comprises the steps that a data transmission notification is received, the data transmission notification comprises a source node and a destination node for data transmission, one of the source node and the destination node is a ground network element, and the other one is a satellite-borne network element; according to pre-maintained satellite-ground topological information and inter-satellite topological information, routing information of each routing segment in a plurality of routing segments between the ground network element and the satellite-borne network element which are configured in advance is calculated, and the satellite-ground topological information comprises a connection relation between a gateway station and a satellite node; the inter-satellite topological information comprises a connection relationship among a plurality of satellite nodes; and performing routing configuration according to the routing information of each routing segment.

This invention discloses a lightweight few-shot incremental learning method for power line defects. The method includes: first, constructing an inspection image dataset containing basic and incremental categories, and designing a lightweight feature extraction network; training the network using the basic category data to generate prototype vectors for each category, forming an initial prototype classification model; in the incremental stage, constructing incremental prototypes for a small number of samples of the new category, and introducing a prototype transfer mechanism based on inter-class tension, adaptively adjusting the new prototype based on the geometric similarity and discrimination conflict strength between the new and existing prototypes to alleviate category conflicts; simultaneously, introducing an adaptive feature fusion mechanism that perceives discrimination uncertainty, dynamically fusing the discrimination results of the main and auxiliary branches to improve recognition stability. This invention, employing the above-mentioned lightweight few-shot incremental learning method for power line defects, can achieve efficient and stable power line defect recognition in scenarios with few samples and continuously expanding categories.

The present application relates to the field of data processing, and particularly relates to a user travel behavior label identification method and device, the method comprising: obtaining target user information and converting into a user feature vector; determining a travel behavior label corresponding to the user feature vector based on a plurality of preset travel behavior labels respectively corresponding to a plurality of determination conditions and a plurality of travel behavior label prediction models; if the user feature vector and any determination condition have a preset association relationship, then the travel behavior label corresponding to the determination condition is taken as the result; otherwise, the user feature vector is input into the plurality of travel behavior label prediction models, and the result is determined according to the model output. The present application sets different key feature subsets as determination conditions for different travel behaviors, and combines the two-stage architecture of rule priority and model bottoming, thereby ensuring the identification accuracy, reducing the calling frequency of complex models, and improving the overall identification efficiency.

The application provides an X-band broadband high-gain radar metasurface antenna and a design method thereof. The metasurface antenna comprises, from top to bottom, a metasurface layer, an intermediate metal layer and a lower microstrip feed line layer. The metasurface layer comprises at least one metasurface unit. An asymmetric chamfer structure is designed on each metasurface unit. A center fork-shaped slot is arranged at the center of each metasurface unit, and four embedded parasitic branches are arranged at four side edges, respectively. Rectangular coupling slots corresponding to each metasurface unit are arranged on the intermediate metal layer. Feed lines corresponding to each metasurface unit are arranged on the lower microstrip feed line layer. The design method provided in the application can improve the design efficiency.

The embodiment of the invention provides an electromagnetic scattering calculation method based on a heterogeneous GPU cluster. The method is applied to the field of computational electromagnetism, and comprises the following steps: constructing a hierarchical bounding volume geometric acceleration structure of a target object and adding an edge index, and meanwhile, completely copying and distributing data of the acceleration structure to a video memory of each computational node in a heterogeneous GPU cluster; generating a random ray path starting from an emission source, modeling an electromagnetic scattering process as a ray path set in a probability space, and generating a ray direction through an importance sampling strategy; determining a static scheduling strategy for the ray batch based on the calculation cost of pre-sampling and geometric enhancement, and carrying out non-uniform distribution on task tiles according to the heterogeneous GPU calculation power; and monitoring the execution state of the heterogeneous GPU cluster in real time, and dynamically adjusting the distribution strategy of the residual ray tasks based on execution feedback. According to the method, the calculation overhead can be remarkably reduced while the calculation precision is ensured, and the efficiency and expandability of complex target electromagnetic scattering calculation are improved.

The application is suitable for the field of network communication technology, and provides a network congestion processing method and device, a terminal, a medium and a program product. The method comprises the following steps: acquiring a round-trip time sequence of a target network transmission link; performing differential processing on the round-trip time data at adjacent time instants in the round-trip time sequence to obtain a plurality of high-frequency components in the round-trip time sequence; updating the round-trip time data at even time instants in the round-trip time sequence according to the shift results of the plurality of high-frequency components to obtain a plurality of low-frequency components in the round-trip time sequence; identifying the network state of the target network transmission link according to the plurality of high-frequency components and the plurality of low-frequency components; and performing a first window reduction operation on the congestion window of the target network transmission link in the case that the network state is a network congestion state. The method can quickly and accurately distinguish the network jitter and congestion problems in a network environment with high bit error rate and high jitter, thereby improving the data transmission efficiency.

The application discloses a multi-modal radio frequency authentication method based on a multi-scale signal representation, which comprises the following steps: firstly, pre-processing the original IQ signal of a target device received to obtain an instantaneous envelope signal; carrying out multi-scale decomposition and denoising processing on the instantaneous envelope signal to obtain a denoised envelope signal; constructing a multi-modal data set according to the original IQ signal and the denoised envelope signal; carrying out feature extraction and fusion on the multi-modal data set through a pre-trained target feature extraction network to obtain a fused feature representation; and carrying out classification on the fused feature representation through a pre-trained target classification network to output a classification result corresponding to the target device. The amplitude, frequency and time-frequency energy distribution information can be complementarily fused by constructing the multi-modal data set; the multi-modal features are extracted and fused through the target feature extraction network, so that the information loss is effectively avoided; and the target classification network is used for rapid classification, thereby reducing the calculation cost and ensuring the accuracy of identification.

This invention relates to the field of autonomous driving technology, providing a CUDA-based collision detection method, apparatus, electronic device, and storage medium. The CUDA-based collision detection method includes: in response to a planned candidate trajectory and collected obstacle information, abstracting the candidate trajectory into a sequence of bounding boxes and the obstacle information into a set of obstacle points; processing the bounding box sequence and the obstacle point set in parallel using CUDA to determine matching pairs with potential collision risks, where each matching pair consists of a bounding box and an obstacle point; and in response to the obtained matching pairs, detecting the collision risk of each matching pair in parallel using CUDA to obtain the collision detection result of the candidate trajectory. This invention utilizes the parallel computing power of CUDA to greatly improve detection efficiency, and further improves the speed and accuracy of collision detection by first coarsely screening matching pairs with potential collision risks and then finely detecting the collision risks of each matching pair, achieving efficient and accurate collision detection of candidate trajectories.

The invention relates to the technical field of image recognition, and discloses a wind power fan blade defect recognition method and system based on image recognition, and the method comprises the steps: carrying out the blocking cutting, illumination normalization and image enhancement processing of an original image, constructing a defect-free sample and a defect sample, and dividing the samples into a training set and a test set; the method comprises the following steps: taking a GANopen network as a basic framework, fusing the lightweight design of Mamba-YOLO, constructing a joint loss function by adversarial loss, reconstruction loss and coding loss based on a training set, carrying out unsupervised training, optimizing network parameters until convergence, and obtaining a defect identification model; inputting the preprocessed to-be-detected fan blade image into the trained defect recognition model, performing defect recognition and positioning, and outputting the defect type and position; based on an output result of the defect identification model, dynamically adjusting an early warning level and response measures through a self-adaptive early warning mechanism; according to the invention, the efficiency and accuracy of wind power fan blade defect identification are improved.

The invention discloses a reachable and rate optimization method based on stacked intelligent metasurfaces in a cellular network, which comprises the following steps of: configuring multiple layers of cascaded stacked intelligent metasurfaces at an access point transmitting end, and establishing an end-to-end wireless transmission link system; constructing a system equivalent channel matrix; the method comprises the following steps: deducing a closed expression of the reachable and rate of a system with respect to each layer of phase shift matrix of the stacked intelligent metasurface at all access points, taking the closed expression as a target function, taking maximization of the reachable and rate of the system as an optimization target, introducing a Riemannian manifold gradient algorithm, and utilizing geometric projection of an analytic gradient to optimize the phase shift matrix of each layer of the stacked intelligent metasurface. And the high-efficiency collaborative iteration of the total space element atom phase shift is realized while the phase shift unit mode length constraint is satisfied. The method can solve the problems that in the prior art, the calculation overhead is large, and the system reachability and rate improvement in the high-speed mobile cellular communication environment cannot be achieved.

This invention discloses a hardware acceleration method and accelerator for Transformer based on mixed-precision quantization and Huffman coding. The acceleration method includes: using a genetic algorithm to obtain several configuration schemes for mixed-precision quantization of Transformer network layers; performing mixed-precision quantization on each Transformer network layer based on each quantization configuration scheme to obtain a corresponding KL divergence; training a multilayer perceptron to obtain a quantization configuration prediction network using the quantization configuration scheme and the corresponding KL divergence as the output label and input feature, respectively; receiving a user-set target KL divergence value, using the quantization configuration prediction network to obtain the corresponding quantization configuration scheme, and performing mixed-precision quantization on each network layer based on the quantization scheme; and using Huffman coding to encode and compress all quantization weights before on-chip storage. This invention can reduce storage and computational overhead while maintaining model accuracy.

The application discloses a gateway-based CAN bus redundancy routing method and device, and is applied to a CAN network comprising a gateway controller, a first CAN bus link and a separately arranged second CAN bus link. The method comprises the following steps: firstly, obtaining a sending moment of a sending request issued by an application layer and an acknowledgement moment of a sending completion interrupt generated by a bottom layer hardware; based on the acknowledgement moment and the sending moment, obtaining a real-time load parameter of the link under a current message; if the real-time load parameter exceeds a preset load parameter threshold, determining that the first CAN bus link is in an abnormal state of high load, triggering a dynamic routing switching, and switching data transmission of a target node being the preset node to the second CAN bus link for transmission. The application can effectively monitor the bus state without occupying additional bus bandwidth and with low calculation power consumption, realizes dynamic targeted switching on key nodes, and guarantees the real-time performance and reliability of key service data of the whole vehicle.

The invention discloses a wireless communication spectrum space digital twinning method, which comprises the following steps of: generating basic static channel characteristics under a given environment model and a base station deployment condition; wherein the static channel characteristics comprise a path loss distribution characteristic, a shadow fading characteristic and a multipath propagation characteristic; associating the static channel feature with target vehicle information, and generating a single-frame channel feature when a vehicle exists; and capturing dynamic evolution information of channel characteristics in a vehicle movement process, and expanding the single-frame channel characteristics into sequential dynamic channel characteristics to obtain a digital twinborn model of a dynamic scene spectrum space. According to the method, the influence of a dynamic blocking body can be further considered on the basis of static channel generation, channel feature generation in a dynamic environment is realized, and the reality sense and application applicability of digital twinning are remarkably improved.

The application relates to the technical field of wireless sensing, and discloses a human posture estimation and non-contact vital sign monitoring method and system based on CSI. WiFi channel state information (CSI) is collected, and dynamic CSI feature vectors are obtained through preprocessing. Features are mapped to human key point feature space through Transformer cross attention, spatial reasoning is performed through graph convolutional neural network fusion of human skeleton topology constraints, and human posture estimation results are output. Respiratory rate and heart rate are extracted based on multi-band frequency domain analysis, and vital sign monitoring results are output after being optimized in combination with posture information. Without cameras and wearable devices, non-contact human perception that is privacy-safe and wall-penetrating can be realized, and the method has the advantages of high precision, strong generalization and low-cost deployment, and is suitable for smart home, medical monitoring, emergency rescue and other scenes.

The application provides a real-time online action detection technology based on a Mamba architecture, comprising: S1, training data acquisition, acquiring video streams of multiple perspectives in a target scene; S2, data preprocessing, extracting video features from the video streams based on a pre-trained two-dimensional convolutional neural network; S3, introducing a Mamba OAD framework, which is composed of an action detection module and a future prediction module; S4, constructing the action detection module, which is mapped to a feature space through a linear projection layer, and the linear projection layer converts input features to D dimensions through a learnable weight matrix; S5, constructing the future prediction module, which predicts future latent features based on short-term latent features output by the action detection module; and S6, model training and reasoning, in which related modules are jointly optimized in the training process. The mAP and mcAP indicators are used to measure the online action detection accuracy, and the FPS and GFLOPs indicators are used to measure the complexity of model inference. The application improves the accuracy and robustness of action detection.

The invention discloses an equipment fault detection method and related equipment, relates to the field of fault detection, and realizes a gated processing mechanism of first preliminary screening and then subdivision by synchronously acquiring long-time sequence vibration signals and noise signals of equipment and constructing a two-stage cascaded fault detection architecture. The fault preliminary screening model performs quick fault preliminary screening on all the to-be-detected fragments, and only the fragments judged to be faulty are sent to the fault type subdivision model for specific fault type identification, so that the unnecessary calculation overhead is remarkably reduced, and the system detection efficiency is improved. According to the method, bimodal information is fully utilized, and efficient and stable equipment fault detection is realized while high detection precision is ensured.

The invention discloses a vehicle target detection method based on improved RT-DETR, and the method comprises the steps: firstly obtaining a single-frame image from a road monitoring video stream, and carrying out the preprocessing and feature extraction; secondly, performing multi-scale linear attention modeling based on the extracted features to obtain a context feature map; then, based on the extracted features and the context feature maps, in combination with a top-down semantic injection module and a bottom-up space injection module, cross-scale feature fusion is carried out on the multiple feature maps, and cross-scale features are obtained; and finally, object query initialization is carried out based on an uncertainty minimization query selection mechanism, and the category and bounding box prediction of the vehicle target is completed by a decoder and a detection head. According to the method, the global context relationship of the vehicle target is effectively modeled, the boundary positioning precision and the detection robustness are improved in a complex scene in which vehicles are densely distributed or severely shielded, and the overall calculation overhead is effectively reduced.

A beamforming design method for intelligent metasurface arrays based on orthogonal time-frequency-space technology includes: aggregating multipath channels under direct illumination into cascaded channels with near-field direction vectors; randomly initializing the phase shift of intelligent metasurface units; establishing an effective channel matrix based on the cascaded channels; then establishing a multi-user and rate objective function; maximizing the objective function through phase shift optimization using a hierarchical search strategy; repeatedly executing the hierarchical search strategy to search and optimize subsequent blocks until the number of intelligent metasurface units in each block reaches a preset value; then performing an exhaustive search in the final block, thus completing one iteration; repeating the iteration until the maximum number of iterations is reached; outputting the final optimal phase shift; and implementing beamforming design based on the final optimal phase shift. This invention can reduce the complexity of beamforming design and ensure reliable transmission of high capacity during high-speed movement.

This invention relates to the technical field and discloses a method for real-time aerial data fusion and secure transmission of low-altitude unmanned aerial vehicles (UAVs) based on state constraints. The method utilizes a 5G-A integrated sensing base station to simultaneously perform communication and radar sensing, extracting micro-Doppler features from the echo and determining the health level. Health characteristics such as UAV rotor speed fluctuations and body vibration are used as hard constraints. Downlink control information from the physical layer directly overwrites the UAV's uplink transmission parameters. The UAV's MAC layer uses a forced-mode gated video encoder output and layered encrypted transmission. Based on the UAV's state-aware forced constraint communication, secure communication transmission of UAV aerial data is achieved. Furthermore, multi-source spatiotemporal alignment and edge fusion at the base station edge nodes enable secure, real-time, and intelligent management of low-altitude data, improving the security, real-time performance, and reliability of low-altitude UAV data transmission.

This invention belongs to the field of image recognition technology, specifically relating to a continuous learning target recognition method based on spatiotemporal information fusion using a spiking neural network. It comprises two learning modules: a fast learner and a slow learner. The slow learner acquires general features that are invariant to input changes but sensitive to semantic representation through self-supervised learning. In the fast learner, whenever a new task arrives, the processor freezes the trained feature extraction modules and adds new feature extraction modules to learn features of the new category. These feature extraction modules gradually aggregate to form a joint feature representation. By fusing general representations with task-specific features across spatial and temporal scales, this method effectively mitigates task confusion and catastrophic forgetting, improving target recognition accuracy.

This invention discloses a robust fusion method for large models based on semantically aligned fuzzy clustering. The method includes: first, obtaining a sequence of probability distribution vectors from the outputs of multiple heterogeneous large-scale pre-trained models for the samples to be processed; then, introducing non-negative reliability weights to weight this sequence to obtain an aggregated probability matrix, and constructing a graph Laplacian matrix accordingly; second, approximating the aggregated probability matrix into a fuzzy membership matrix and a semantic prototype matrix, constructing a cost objective function by combining the graph Laplacian matrix, and iteratively updating each matrix and weight until convergence; finally, solving for the maximum value of the converged fuzzy membership matrix to obtain the sample prediction category result. This invention can effectively suppress the influence of inferior models in unsupervised environments, significantly improving the accuracy and robustness of fusion prediction.