365 results about "Adaptive computing" patented technology

The invention discloses an end-side cloud collaborative intelligent fire-fighting fire monitoring system based on image fusion, and relates to the technical field of intelligent fire-fighting, the system is composed of a plurality of functional modules, and the system comprises a multi-modal image fusion module which generates a dynamic scanning priority map based on prior data, distinguishes a natural heat source from an abnormal fire by using a dual-light fusion algorithm, and sends an image fusion result to a cloud server; a scanning area is divided according to the thermal risk grade, and the thermal imaging resolution is dynamically adjusted; the distributed edge computing module is used for carrying out space-time synchronization on cross-modal data through a multi-modal feature alignment network, and carrying out dynamic allocation on a CUDA core and CPU resources through adaptive computing scheduling; an improved artificial bee colony algorithm is adopted, the bandwidth of the multi-sensor data flow is dynamically allocated through a time-sharing multiplexing protocol, and three-dimensional path planning is carried out; and the end-side cloud collaborative decision module constructs a federated learning driven model sharing network, and each edge node trains a lightweight YOLOv5s pruning model based on local data.

The invention discloses a BIM-based traffic engineering construction progress and resource management optimization system, and the system comprises a three-dimensional cognitive interaction body which adjusts the construction progress and resource configuration in real time, and eliminates the physical difference between a virtual model and an actual environment; the self-adaptive calculation engine is used for realizing butt joint between the virtual model and an actual construction site by utilizing real-time sensing data and a cognitive correction mechanism; the progress prediction simulation module is used for realizing dynamic prediction and optimization of the construction progress through deep learning and a self-adaptive feedback mechanism; the virtual-reality synchronous perception grid is constructed based on the obtained progress dynamic prediction and optimization result, and a virtual model and a construction site are synchronized and dynamically corrected; and the self-optimization module is used for realizing self-adaptive optimization and long-period evolution through a genetic algorithm and multi-objective optimization by utilizing a self-optimization and ecological optimization mechanism. According to the method, the problem of virtual reality distortion is solved, and the difference between virtuality and reality in the construction process is accurately corrected.

The invention provides a self-adaptive computing power scheduling system for large model training, and relates to the field of electric digital data processing, the self-adaptive computing power scheduling system comprises a computing power management module, a task analysis module, a self-adaptive scheduling module and a data interaction module, the computing power management module is used for managing and monitoring overall computing power resources, and the task analysis module is used for analyzing the overall computing power resources; the task analysis module is used for analyzing requirements and characteristics of training tasks, the self-adaptive scheduling module performs computing power scheduling according to real-time resource conditions and task requirements, and the data interaction module is used for managing data interaction among different computing power nodes; according to the system, the training efficiency can be effectively improved by assigning the training tasks to different nodes and adjusting the allocated resources in the task execution process.

The invention discloses a passable map construction method for a leg-foot robot. The passable map construction method comprises the following steps: preprocessing acquired original point cloud information, inertia measurement information, latitude and longitude information and camera image information; performing error state Kalman filtering on the preprocessed point cloud information and inertial measurement information to obtain a rough pose of the robot; the precise pose of the robot is obtained through graph optimization; constructing a 2.5 D grid topographic map through extended Kalman filtering, and processing the 2.5 D grid topographic map to obtain a smooth and void-free grid topographic map; and extracting a normal vector in each unit grid in the grid topographic map, calculating the gradient, gradient change rate, roughness and passability in each unit grid, and maintaining a legged robot passable map in real time through extended Kalman filtering. According to the method, adaptive calculation of the passability of different terrains and materials in the real world is realized, so that the obtained passability graph is more credible.

The embodiment of the invention provides an information mapping method for intelligent perception information in a virtual space, perception data is acquired by deploying a multi-sensor array, data features are extracted based on a multi-scale feature pyramid and an attention mechanism, and a space mapping relation is established by using self-supervised learning and geometric constraints. Local data blocks are organized by adopting a hierarchical index tree based on a union search set, an adaptive calculation flow graph is constructed to realize efficient mapping, and data transmission is optimized in combination with predictive caching and a differential updating strategy. And meanwhile, parallel processing of incremental mapping is realized through a pipeline data channel, and the data synchronization efficiency is ensured by a dynamic scheduling mechanism based on load prediction. The method effectively solves the problem of performance bottleneck when a traditional mapping scheme processes large-scale multi-source sensing data.

The invention belongs to the field of power grid control, particularly relates to an LLM Agent-based power system operation mode adaptive calculation method, and aims to solve the problem that load flow calculation of a power system operation mode is relatively difficult in the prior art. The method comprises the following steps: acquiring a task demand of a power system; the task demand is input into an LLM model, the LLM model splits the task demand into a plurality of sub-tasks, and Agents used for executing the sub-tasks are determined; determining the output of the LLM model, and allocating a plurality of subtasks to a load flow calculation Agent and a parameter adjustment Agent by calling a task management Agent; and calling the load flow calculation Agent and the parameter adjustment Agent, and executing each subtask allocated by the task management Agent. According to the method, the adaptability of the load flow calculation and adjustment strategy is improved by utilizing the LLM model, and meanwhile, the accuracy of load flow calculation is guaranteed through multi-Agent calling, so that the load flow calculation and adjustment effect on the power system in the prior art is improved.

The invention discloses a computing resource management scheduling method and system. The method comprises the following steps: carrying out resource demand prediction based on a machine learning algorithm; a main node main scheduler performs an allocation task by using the resource demand prediction result and the resource demand information provided by the execution node in combination with multiple computing resource allocation algorithms to realize multi-resource collaborative scheduling; the main scheduler realizes distributed scheduling of task, node and cluster levels based on a self-adaptive distributed multi-level scheduling architecture; and the main scheduler realizes monitoring data return through a feedback loop and carries out real-time distribution strategy adjustment. Aiming at the bottleneck problem in the resource management process, the invention provides a brand-new and dynamically adaptive computing resource management scheduling method and system, and the computing resource management scheduling performance is remarkably improved.

The invention provides a multifunctional inspection device detection system based on a distribution network mobile operation terminal, and relates to the technical field of electric power detection, and the multifunctional inspection device detection system comprises a portable main terminal of a dynamic collaborative architecture, a wearable terminal for augmented reality interaction and a distributed heterogeneous sensing cluster, which form a closed-loop data chain through an electric power dedicated low-delay wireless communication network; the portable main terminal is integrated with a heterogeneous computing unit with adaptive computing power distribution, can perform real-time fusion analysis on multi-dimensional sensing data, and outputs a visual result containing fault location and confidence; the wearable terminal superposes fault information to a real scene in a three-dimensional marking form through a virtual-real fusion positioning technology, and supports eye movement and voice collaborative interaction; multi-module collaborative acquisition of the distributed heterogeneous sensing cluster is combined with a nanosecond timestamp synchronization mechanism, so that the limitation of traditional single parameter detection is broken through, rich and synchronous basic data is provided for subsequent analysis, and the comprehensiveness of routing inspection is greatly improved.

The invention relates to the technical field of prediction algorithms, and discloses a high-precision active power filtering and prediction algorithm for a three-phase ammeter, which comprises the following steps: acquiring original signals of three-phase voltage and current for baseline correction, performing time-frequency analysis on non-stationary harmonic waves and noise of the original signals, and dynamically adjusting filtering parameters; voltage and current phase alignment is carried out through FIR phase shift and zero crossing point detection, a phase-locked loop is constructed by using GRU to track the frequency of a power grid, the sampling frequency is dynamically adjusted, and frequency mutation is detected; iapFFT transformation is optimized to suppress spectrum leakage, a phase error is corrected through a deep learning network, and weak harmonic detection is enhanced by using an attention mechanism; classifying harmonic waves, and adaptively calculating total active power; dimensionality reduction is carried out on historical power and environmental parameters through an auto-encoder, and future active power is predicted; low-power-consumption hardware is adapted, and training efficiency and data security are optimized; three-phase signals are processed in parallel, and FPGA storage and FFT / FIR cores are optimized for filtering processing; and dynamically adjusting parameters of the phase-locked loop.

The disclosure provides an approach for adaptive computing security. Embodiments include receiving, by a security rule adaptation system, alert information about a plurality of alerts generated based on one or more security rules. Embodiments include receiving, by the security rule adaptation system, user feedback with respect to a subset of the plurality of alerts. Embodiments include modifying, by the security rule adaptation system, a security rule based on the alert information and the user feedback to produce an adapted security rule. Embodiments include providing, by the security rule adaptation system, the adapted security rule to a security component, wherein the security component uses the adapted rule to detect computing security threats.

The invention provides an automatic history fitting method for gas drive numerical reservoir simulation, which comprises the following steps of: 1, establishing a gas drive numerical simulation mathematical model, and correcting the gas drive numerical simulation mathematical model by using a gas drive PVT phase state fitting experiment; 2, solving the corrected gas drive numerical simulation mathematical model obtained in the step 1, performing parameter analysis, and screening main control factors influencing accumulated oil production; 3, generating a numerical simulation training sample set based on an orthogonal experiment, and obtaining the main control factor parameter initial value constraint screened out in the step 2 through a machine learning algorithm; step 4, establishing a multi-parameter fitting objective function for realizing historical fitting; and step 5, using the main control factor parameter initial value constraint obtained in the step 3, performing iterative calculation through a projection gradient method, fitting the target function established in the step 4, and terminating when a convergence condition is satisfied. According to the method provided by the invention, the constraint parameter values can be adaptively calculated, automatic historical fitting is completed, and the definition of the fitting parameter range does not depend on empirical and experimental determination any more.

The invention provides a self-adaptive sampling method for fault diagnosis under data multi-class imbalance and related equipment, and effectively solves the problems of low sample generation quality, high parameter dependence, high calculation complexity, poor adaptability to complex working conditions and the like. The method comprises the steps of coping with different data feature scenes through a parameter adaptive calculation mechanism, then exploring a global optimal solution in multi-classification modeling accuracy model solution optimization by an evolution mechanism through employing a Newton-Raphson optimizer thought, and finally converting an optimal solution set into various fault samples by using a feature recombination mechanism. Therefore, high-quality sample equalization is realized, small sample multi-class imbalance fault diagnosis is carried out by combining MAESTE and multi-class LS-SVM, and the interpretability of the model is improved.

The invention discloses a dynamic calculation distribution and error control acceleration system and method for a diffusion model, and the system comprises a feature prediction module, a verification decision module, a sample complexity analysis module, and an interface and integration module. On the premise of ensuring that the generation effect is basically unchanged, the number of times of complete forward calculation is remarkably reduced, so that while the overall reasoning efficiency of the generative artificial intelligence task is improved, complete forward calculation is executed in a plurality of key time steps, and high-precision features are obtained to serve as a reference; and then a feature prediction module is used to combine feature difference information of the current time step and a plurality of previous time steps, and rapid prediction of features of a plurality of subsequent time steps is realized through a Taylor expansion formula. A prediction result is not directly adopted, but is submitted to a lightweight verification decision module for error detection and reliability evaluation; and the module judges whether to accept a prediction result or execute complete calculation in a back-off manner according to a verification result and a dynamically adjusted error threshold value, so that reasonable allocation of calculation resources among samples with different complexities is realized, and error accumulation is effectively inhibited while the reasoning speed is improved.

The invention discloses a storage and calculation separation method and device based on multi-level cache and intelligent scheduling and a server, and belongs to the technical field of data processing, and the method comprises the following steps: collecting operation behavior data of a user, carrying out distributed storage, and dynamically distributing and calculating node cache space; performing classification marking to form marked user data; performing intelligent scheduling layer analysis tasks on the marked user data, and screening and distributing adaptive computing nodes; the computing node receives the analyzed task to check the local cache, obtains corresponding data from the local or request remote storage according to the validity of the local corresponding data, computes the corresponding data, and caches the computed data to the local or remote storage according to the user access probability; and carrying out dynamic scheduling and hierarchical storage on the calculated data between local cache and remote storage according to access frequency and heat factors. According to the invention, the problems of low transmission efficiency and unbalanced resource utilization of the existing storage and calculation separation architecture are solved.

Aspects of embodiments pertain to a method for determining an Input / Output (I / O) device configuration for a music teaching system. The method may comprise receiving a plurality of I / O device configurations of a music teaching system; receiving, for a given I / O device of the music teaching system, an initial I / O device configuration; and determining an updated I / O device configuration for the given I / O Device, based on the initial I / O device configuration of the given I / O Device and at least one of the plurality of received I / O device configurations. The determining is performed such that the updated I / O device configuration has improved device performance compared to the initial I / O device configuration.

The invention discloses a quantum photon hybrid interface adaptive calculation method, and relates to the technical field of quantum calculation and photon integration. Firstly, the current load state of the multi-modal data interface is monitored in real time, and an interface state feature matrix is obtained; then inputting the state matrix into a quantum photon hybrid calculation module, and optimizing an interface type, a transmission protocol and interface parameters in real time through a quantum state feedback and quantum variational optimization algorithm (QAOA) to obtain optimal adaptive configuration of the interface; and dynamically adjusting interface parameters according to an optimization result to realize that the random read delay is less than or equal to 40 microseconds, the sequential read delay is less than or equal to 25 microseconds, the write delay is less than or equal to 400 microseconds, the target identification response time is less than or equal to 0.5 ms, and the target tracking response time is 0.5-1 second. The dynamic intelligent optimization of the interface parameters is realized, the interface transmission efficiency and the real-time response capability are remarkably improved, and the method is suitable for the fields of artificial intelligence, intelligent transportation and the like.

The invention discloses a power distribution intelligent terminal-oriented multi-core heterogeneous chip task scheduling method and system, and belongs to the technical field of computers, the method comprises the following steps: receiving calculation tasks, storing the calculation tasks in a scheduling task queue, and obtaining task importance levels, task arrival time, task deadline and core resources required by the tasks of the calculation tasks; performing priority evaluation on the calculation task according to the data to obtain a priority value; performing scheduling arrangement on the scheduling task queue according to the task importance level and the priority value to obtain a current scheduling queue; in each scheduling period, performing task scheduling on each slave core in combination with the shared memory according to the current scheduling queue and the core resources required by the task; according to the method, index basis can be provided for scheduling of calculation tasks, the problem of chip-level task scheduling caused by different task importance and different calculation resources of the power distribution intelligent terminal is solved, the edge self-adaptive calculation capability of the power distribution terminal is improved, and the multi-service low-delay requirement of the power distribution intelligent terminal can be met.

The invention discloses a seamless switching method and system for a high-speed terminal of a ground-non-ground convergence network based on trajectory prediction and resource pre-reservation, and belongs to the technical field of space-ground convergence communication. The core of the method is that a switching area is accurately judged through bidirectional motion prediction of a terminal and an NTN node; a dynamic switching window is calculated in a self-adaptive mode by combining factors such as terminal speed and network time delay; intelligently selecting an optimal target link by using an AI enabled link scoring model; and cooperative pre-reservation and uplink and downlink synchronization of resources are completed before the window. The method has the advantages that the problem of failure of traditional RSRP judgment in an NTN dynamic environment is solved, the switching success rate and prediction accuracy in a high-speed moving scene are remarkably improved, the service interruption time and signaling overhead are greatly reduced, the continuity and reliability of communication are effectively guaranteed, and the method is suitable for large-scale popularization and application. The method is suitable for high-dynamic service scenes such as intelligent network connection vehicles and unmanned aerial vehicles in a 6G network.

The invention discloses an infrared segmentation double-platform histogram enhancement method based on redistribution, and the method comprises the following steps: S1, carrying out the statistics of a total histogram HIST of original infrared data DA, and carrying out the segmentation processing of the total histogram HIST; s2, carrying out redistribution on the total histogram HIST to obtain a redistributed histogram Dh; and S3, carrying out histogram equalization on the redistributed histogram Dh to obtain an equalization result DB. According to the method, specific upper and lower platform values do not need to be set, all self-adaptive calculation is achieved, the requirement for an infrared scene changing at any time can be met, and robustness is good.

The invention provides a building cooling and heating load adaptive calculation method, system and device and a medium, and belongs to the technical field of load prediction. The method comprises the following steps: collecting image data and power consumption data of a building, and extracting time sequence features from the power consumption data after preprocessing; extracting image features from the image data by using edge detection, feature point recognition and semantic segmentation technologies; generating a three-dimensional point cloud by using image features through a multi-view geometric technology and a surface reconstruction method, and converting a three-dimensional model into a BIM model by using Revit software; calculating solar radiation and an internal heat source based on the power utilization data and the power of the internal equipment of the building, and training a cooling and heating load prediction model by using a random forest and a neural network algorithm; and dynamically adjusting and optimizing the cooling and heating load prediction model by adopting an online learning and grid search technology, inputting the real-time power consumption data and the environmental parameters into the cooling and heating load prediction model, and generating a building cooling and heating load prediction result.

The invention belongs to the technical field of unmanned aerial vehicle obstacle avoidance, and discloses a dynamic obstacle avoidance method for an unmanned aerial vehicle, and the specific technical scheme is as follows: according to the distance and speed of an obstacle relative to the unmanned aerial vehicle, dynamically adjusting and calculating a time step length through a fuzzy mathematical algorithm; the method comprises the following steps: firstly, integrating sensor information of an unmanned aerial vehicle to obtain the position, linear velocity and angular velocity of an obstacle, predicting the movement trend of the obstacle in a short time in the future, and taking the movement trend as input of a speed obstacle method to obtain an expanded speed obstacle area; selecting a speed which does not generate a collision risk with the obstacle in a short time as an optional obstacle avoidance speed; the optimal direction-adjusting obstacle-avoiding speed is preferentially selected; the method comprehensively considers the movement trend of the obstacle, is more predictable, obtains a safer and more reliable obstacle avoidance path, and reduces the cost of path, time and calculation resources caused by obstacle avoidance as much as possible by adopting adaptive calculation of time step length and improvement of a cost function.

Methods and systems are described for computation management of requests in edge cloud domains. A user equipment can send a computation management request. The network can determine the criteria for completing the computation management request, determine static and dynamic status of network resources, assess historical data on performance success and failure and use all of these factors to assign a network resource to respond to the computation management request.

The invention discloses an intelligent manufacturing workstation safety control system oriented to man-machine cooperation, and belongs to the technical field of intelligent manufacturing and man-machine cooperation. The system comprises a multi-modal sensor array, a personnel state sensing module, a robot state monitoring module, a dynamic safety area adaptive calculation module, an intention prediction and risk assessment module, a cooperative control decision module and an execution feedback module. The dynamic safety area self-adaptive calculation module constructs geometric representation of a man-machine interaction space based on a topological manifold theory, and calculates a dynamic safety boundary in real time; the intention prediction and risk assessment module is fused with multi-source information to predict a future trajectory of a person and assess a collision risk; the cooperative control decision module performs multi-objective optimization with safety and efficiency as double objectives to generate an optimal adjustment strategy, the system realizes prospective safety control based on intention prediction, the cooperative efficiency is improved by 25%-45% on the premise of guaranteeing safety, and the working safety accident rate is reduced to be close to zero.

The invention discloses a multi-source fusion carbon emission factor adaptive calculation system and method, and relates to the technical field of carbon emission management, and the system comprises a data input module, a sub-model module, a total model integration generation module, a model weight adjustment module and a carbon emission factor calculation system module. The sub-model module can generate a calculation model of the four sub-models according to a fuel type, a region, a time sequence and a process type, and the total model integration generation module integrates the four sub-models and generates a Stacking integrated prediction large model. According to the method, four sub-models can be generated according to the known fuel, region, time and process information, the four sub-models can be integrated into a large judgment model, the large judgment model can be applied to calculate the result according to the characteristics of different industries and different scenes through the large judgment model, and the prediction precision can be improved in a large range.

The invention discloses a short temporary rainfall forecasting method and system based on a dynamic neural network architecture. According to the method, training sample distribution is optimized through a space-time gradient driven resampling strategy, a time-frequency double-branch dynamic network architecture is adopted, a time domain branch extracts multi-scale space features through an encoder-decoder structure, a frequency domain branch adaptively activates an FFT calculation module through a content awareness dynamic network to capture multi-scale time features, and the multi-scale time features are obtained. And two key problems of meteorological data distribution imbalance and deep learning model optimization imbalance are solved in combination with a balance regression loss function. According to the method, adaptive computing resource allocation for meteorological events with different complexities is realized, the forecasting precision and computing efficiency of heavy rainfall events are remarkably improved, and an efficient and reliable technical solution is provided for short temporary rainfall business forecasting.

The invention discloses an industrial Internet of Things federated learning method based on a federated increment decision tree, and the method comprises the steps: a cloud server deploys and initializes a federated learning global model and the federated increment decision tree, and sets a statistical histogram bucket boundary set of each feature value; in each federated learning iteration, the cloud server broadcasts a federated learning global model parameter, each industrial device adopts a local data set to train and count to obtain a local gradient histogram parameter, the edge server performs local aggregation on the local gradient histogram parameter and the federated learning model parameter, and the edge server performs local aggregation on the local gradient histogram parameter and the federated learning global model parameter; and the cloud server globally aggregates the local aggregation parameters of the gradient histogram and then incrementally trains the federated increment decision tree, simultaneously aggregates the parameters of a federated learning global model, and adaptively calculates an aggregation weight based on a second-order gradient value during local aggregation and global aggregation of the parameters of the federated learning model. The federal learning overhead can be effectively reduced, and the performance of the federal learning model is improved.

The invention discloses a tunnel oxygen supply control method based on multi-parameter coordination, and relates to the technical field of tunnel construction, and the method comprises the steps: constructing a multi-source environment monitoring data set, training a target ventilation quantity prediction model fused with a high-altitude correction coefficient based on the data set, and outputting an initial target ventilation quantity and an oxygen supply parameter; generating a pre-regulation instruction based on the initial target ventilation quantity and the oxygen supply parameters, simulating an instruction execution process through a digital twin virtual simulation model, and outputting oxygen supply evaluation parameters of oxygen content distribution, wind pressure loss and energy consumption of each block of the tunnel; constructing a tunnel oxygen supply evaluation model, and comparing oxygen supply evaluation parameters obtained by simulation with a safety threshold; and executing the formal adjustment instruction. Through multi-dimensional sensing treatment, high-altitude self-adaptive calculation, digital twinborn simulation verification, multi-scene parameter optimization, multi-device collaborative linkage and model dynamic evolution technologies, the breakthrough of tunnel oxygen supply from static experience control to dynamic precise intelligent regulation is realized.

The invention discloses an adversarial unsupervised time sequence anomaly detection method and system based on a KAN, and belongs to the field of industrial time sequence data anomaly detection. The method comprises the following steps: (1) acquiring a training set and a test set composed of label-free industrial time sequence window data; (2) training a shared encoder and double decoder model based on KANs by using the training set, and adaptively calculating an abnormal score threshold value of the model by using the test set; and (3) deploying the trained model on an actual industrial production line, calculating an abnormal score of industrial time sequence window data collected by an industrial sensor in real time, and if the abnormal score exceeds an abnormal score threshold, determining that the data in the window is abnormal. The method has the advantages of excellent high-dimensional data representation capability, strong anti-noise performance, high-quality anomaly detection capability and good portability without depending on priori knowledge under multiple industrial backgrounds.

The invention provides a service interface dependency relationship mining and analysis method based on AI and graph calculation, and relates to the technical field, and the method comprises the steps: constructing a dependency relationship graph, guiding random walk sampling through a meta path, adaptively calculating the radius of a dynamic sub-graph, and hierarchically detecting a framework. The detection accuracy of the abnormal dependency relationship is improved, the operation and maintenance cost of the micro-service system is reduced, and the stability and reliability of the system are enhanced.

The invention provides an unmanned aerial vehicle pose estimation method based on credibility weighted strong tracking filtering. According to the method, the posterior error covariance after actual strong tracking filtering iteration and the theoretical true posterior covariance of algorithm iteration are utilized to define a trust factor used for describing filtering deviation under noise uncertainty, the trust factor acts on self-adaptive calculation of a forgetting factor, then a time-varying fading factor is corrected through calculation of an innovation error covariance, and therefore the filtering deviation under noise uncertainty is corrected. And tracking estimation under the condition of noise uncertainty is realized. Compared with traditional strong tracking series filtering, the method can quantify the uncertainty of noise and relieve experience dependence of forgetting factors. Meanwhile, the time-varying fading factor acts on the process noise covariance, and compared with a traditional action mode, the method has better interpretability. When the residual error and the fading factor are calculated, compared with the traditional fixed forgetting factor and the trust factor, the self-adaptive adjustment of the forgetting factor can be realized, the performance error is reduced, and meanwhile, the artificial experience assignment is reduced.