106 results about "Global distribution" patented technology

The invention relates to the field of supply chain scheduling decisions, in particular to a supply chain intelligent scheduling decision optimization method and system based on big data driving. The method comprises the following steps: collecting supply chain heterogeneous data streams, extracting abnormal disturbance events, carrying out nonlinear disturbance response mining, and constructing a supply chain disturbance response field; identifying dynamic state parameters of all nodes of the supply chain, analyzing correlation characteristics among the parameters, performing multi-node logistics state transfer evolution, and generating logistics state transfer characteristics of each node; and multi-node delay evaluation is carried out based on the supply chain heterogeneous data flow, global distribution mapping is carried out, and an upstream and downstream logistics state delay characteristic spectrum is constructed. According to the invention, adaptive optimization collaboration between multiple nodes and multiple targets is realized, so that the stability, flexibility and economic benefit of the whole supply chain system are effectively improved.

The invention provides a cross-park enterprise data collaborative analysis method based on federated learning, and relates to the technical field of distributed machine learning and data security, and the method comprises the steps that a central server distributes an initial global model and configuration parameters to each park node; the nodes execute local data feature alignment to generate standardized feature vectors; calculating dynamic collaborative factors of local data and global distribution; adjusting a training strategy based on the collaborative factors and updating model parameters; collecting model updating through an encrypted channel, and screening effective updating by adopting a dynamic aggregation offset threshold value; performing weighted aggregation to generate a new global model; and terminating the process when the cross-park convergence condition is met or the maximum round is reached. According to the method, heterogeneous data differences are eliminated through a dynamic feature alignment mechanism, dual-channel collaborative evaluation and adaptive security protection are combined, multi-park collaborative modeling efficiency and robustness are remarkably improved on the premise of guaranteeing data sovereignty, and the problems of feature space splitting, weak attack protection and node contribution imbalance are solved.

The invention discloses a pipeline defect magnetic flux leakage detection method and device based on multi-scale data driving deep learning, and relates to the technical field of pipeline defect detection. A multi-scale magnetic flux leakage signal data set containing defect global distribution and local details is generated, and multi-level primary features are automatically extracted by using a convolutional neural network; and the defects of poor generalization and easy key information omission of a manual method are overcome. And then dynamically enhancing and performing weighted fusion on multi-scale features by means of a multi-scale convolution branch and an attention mechanism, so that the model can learn global and local features of the defect at the same time, the problem that the global and local features of the defect are difficult to consider in traditional deep learning is solved, and through transverse connection and up-sampling fusion of a feature pyramid, a multi-scale feature is obtained. And the features have high semantic information and high spatial details. And finally, independently carrying out multi-task prediction by virtue of a task decoupling module, and calibrating result consistency by virtue of a task alignment module, so that the detection accuracy in a complex scene is improved, and more accurate and robust pipeline defect magnetic flux leakage detection is realized.

The invention discloses a target detection method and system based on dynamic memory enhancement, relates to the technical field of computer vision and target detection, and aims to solve the problems that a traditional method depends on local features of a single image, is difficult to detect a target, is low in precision and is weak in generalization ability. The method comprises the steps that after a to-be-detected image is preprocessed, multi-stage multi-scale features are extracted through a backbone network; an encoder outputs single image global features for feature interaction fusion, and a global distribution extraction module constructs and updates a data set level global feature distribution library; the decoder adopts a hierarchical structure, and optimizes the target query token layer by layer in combination with self-attention, cross attention and global distribution fusion sub-modules; and finally outputting a target category and bounding box coordinates. Through global context injection, small target and shielding target detection precision is improved, model training stability and generalization ability are enhanced, and the method is suitable for diversified detection scenes.

The invention discloses a multi-GNSS satellite signal power abnormity real-time monitoring method. The method comprises the following steps: acquiring real-time observation data of an observation station network which is uniformly distributed globally; when the linear relation between the carrier-to-noise ratios of different frequency signals of each satellite in each observation station and the dynamic monitoring threshold value related to the elevation angle of each satellite are determined, one signal is used for predicting another signal by using the linear coefficient, and the deviation between an actual value and a predicted value is calculated; calculating a satellite elevation angle of the real-time observation data, and determining a monitoring threshold value of the current satellite at the current moment according to the satellite elevation angle; and counting an average value of the deviations of all satellites monitored by all observation stations at the current moment, and marking that the power of the current satellite at the current moment is abnormal when the absolute deviation of the current satellite is greater than a monitoring threshold value. According to the invention, on the basis of the carrier-to-noise ratio linear relation of different frequency signals and the dynamic monitoring threshold, real-time monitoring of various power abnormal scenes including natural interference, man-made interference, satellite active power adjustment and the like is realized.

The invention provides a multi-Alos robot cooperative sorting method for warehouse logistics and a related device, and relates to the technical field of warehouse logistics. Receiving real-time order data, and performing analysis according to the real-time order data to generate a task set including cargo positions, target sorting areas and priorities; dynamically calculating a task matching degree by combining state information of each robot through a distributed task allocation mechanism, and optimizing global task allocation by adopting a reinforcement learning model to generate a global task allocation result; modeling a storage environment into a space-time diagram structure on the basis of a global distribution strategy; generating a conflict-free path by adopting a path-finding algorithm in combination with the space-time diagram structure, and optimizing a trajectory in combination with a task time limit constraint; and according to the target position in the conflict-free path, goods are recognized through a visual system, and grabbing and placing are completed by combining with a mechanical arm. The technical effects that the sorting efficiency and adaptability are improved, and the operation and maintenance cost is reduced are achieved.

The invention relates to the technical field of logistics data processing and artificial intelligence, and discloses an intelligent warehouse management system based on logistics data processing, and the system comprises a multi-source heterogeneous data fusion preprocessing module which carries out the real-time collection of the data of a global distributed storage center, and outputs a time-space standardized multi-dimensional feature data set; the spatio-temporal correlation demand prediction module is used for constructing a spatio-temporal model and carrying out depth prediction by adopting a long-short-term memory network and an attention mechanism; the multi-constraint global inventory optimization module is used for constructing a multi-objective optimization model and solving an inventory configuration scheme through a decomposition coordination method; the reinforcement learning intelligent scheduling module is used for establishing an environment state space and performing strategy optimization by adopting a deep Q network; the digital twinborn visual decision support module is used for realizing real-time synchronization of physical storage and a digital model and outputting decision suggestions and risk early warning; according to the method, multi-source heterogeneous data can be effectively processed, intelligent prediction of cultural perception is realized, and global collaborative optimization is carried out.

The invention discloses a federated learning model optimization method and system based on meta-learning and regularization, and belongs to the technical field of artificial intelligence and data privacy. Aiming at the problems of slow model convergence and insufficient individuation caused by non-independent identically distributed data, the method adopts collaborative optimization of a meta-learning double-circulation mechanism and regularization: a central server distributes a global model to a client, and the client performs internal circulation gradient updating and external circulation meta-gradient calculation by dividing a support set and a query set; generating updating information adapted to local data; meanwhile, an L2 regular term is introduced to constrain the parameter difference between a local model and a global model, and a personalized model is optimized in combination with prediction loss. The system realizes cooperative training through a global distribution module, a local adaptation module, a personalized module and an aggregation module, and the server adopts data volume weighted aggregation client updating. According to the method, the model generalization ability under heterogeneous data is effectively improved, and the prediction difference between the clients is balanced.

The invention discloses a thermal power generating unit primary frequency modulation capability online evaluation method, device, equipment and medium, and relates to the field of thermal power generating unit frequency modulation. A UMAP algorithm is adopted to carry out feature dimension reduction and reconstruction processing on a high-dimensional operation data set of a thermal power generating unit to be evaluated; the UMAP can capture the manifold structure of the data and meanwhile reserve local association and global distribution, the dimensionality of the features is greatly reduced, then the obtained unit output feature set is input into a pre-trained thermal power unit primary frequency modulation output prediction model, and online frequency modulation output prediction is achieved. A traditional framework which only focuses on frequency modulation performance is broken through in dimension evaluation, evaluation dimension quantification is carried out based on a unit output prediction result and a high-dimensional operation data set, and an evaluation factor set is constructed. And finally, an entropy weight TOPSIS method is adopted to carry out weight distribution calculation on the evaluation factor set, a primary frequency modulation capability comprehensive evaluation index is generated, the advantages and disadvantages of the frequency modulation capability of the unit are visually reflected, and the frequency modulation capability evaluation accuracy is improved.

The invention relates to the technical field of nanometer motion tables, in particular to a nanometer motion table design method, device and equipment and a storage medium. In the method, a structure sample library is composed of structure samples of a nanometer motion table, and a simulation result covers characteristics such as microscale stress and displacement; the trained parameter prediction model can accurately support the micro-scale performance prediction and optimization of the nano motion platform; a simulation result set of the structure sample library comprises multi-physics field data, a training data set is constructed based on a global simulation result, a proxy model can learn global distribution and coupling relations of physics fields, and optimization hidden dangers caused by single-point simplification are avoided; a proxy model is constructed and trained into a parameter prediction model, traditional finite element analysis is replaced for rapid prediction, and the calculation cost and time consumption in the optimization process are remarkably reduced; the multi-objective optimization algorithm and the parameter prediction model are combined, the optimization scheme is automatically generated, multi-objective optimization can be carried out at the same time, multiple performance indexes can be considered, and the comprehensiveness of the design scheme is improved.

The invention provides a three-dimensional object semantic automatic labeling method based on Gaussian splashing, and relates to the technical field of data processing, and the method comprises the steps: 1, collecting three-dimensional point cloud data of park building surfaces, road facilities, greening vegetation and dynamic objects, and according to the density of the three-dimensional point cloud data, the scene complexity and the characteristics of the dynamic objects, calculating the three-dimensional point cloud data of the park building surfaces, the road facilities, the greening vegetation and the dynamic objects; dynamically adjusting global distribution, the number of cycles and a neighborhood search range, and generating a three-dimensional space probability description; and step 2, inputting the three-dimensional space probability description into a multi-scale feature fusion network, extracting a multi-modal scene feature group including geometric curvature, material reflection and semantic association features, and monitoring feature quality in real time through a traceability diagnosis unit to generate a feature quality evaluation report. According to the method, the semantics of the three-dimensional object is adjusted and labeled through multi-dimensional data processing and a self-adaptive algorithm, the resource configuration is optimized, and the labeling accuracy and efficiency are improved.

The invention provides a photovoltaic output scene clustering method. High-frequency noise of an original photovoltaic output time sequence is suppressed through self-adaptive smoothing processing, fluctuation characteristics of the high-frequency noise are extracted, it is ensured that target time sequence data accurately reflect photovoltaic power generation working conditions, and high-quality input is provided for subsequent clustering; the target clustering number is determined based on joint optimization parameters (fusing local dynamic characteristics and global distribution characteristics), and scene division deviation caused by subjective setting of the clustering number is avoided; the clustering center is updated by iteratively optimizing the clustering objective function, and it is ensured that the typical scene center truly reflects the mainstream output mode. Therefore, scene representativeness is improved in the aspects of data preprocessing, clustering number decision making and central optimization of a whole link, and generalization ability of a power grid dispatching strategy is remarkably enhanced.

The invention discloses an aviation radiation dose global distribution real-time calculation system based on ground detection equipment, which belongs to the technical field of radiation detection and comprises a ground radiation detection module, a dose simulation module and a data correction module. The ground radiation detection module is used for continuously collecting ionizing radiation data of an atmospheric bottom layer in a calm state and a solar high-energy particle event so as to improve data precision and a signal-to-noise ratio under a weak dose, and the dose simulation module receives latitude and longitude, height and time parameters of an aviation flight path, and calculates a standard atmospheric model and a cosmic ray energy spectrum in combination with the standard atmospheric model and the cosmic ray energy spectrum. The data processing module is used for calculating theoretical dose rates and particle type distribution of different height levels based on a Monte Carlo method and outputting the theoretical dose rates and the particle type distribution, and the data correction module is used for dynamically correcting parameters and weights of simulation results by combining measured data based on a deep learning algorithm of deep learning, so that adaptive fusion of the simulation data and the measured data is realized; and the ground radiation detection module comprises a semiconductor radiation detector, a data acquisition and preprocessing unit, a remote control module and a network communication module. The system has the characteristics of high real-time performance and wide coverage range.

The invention belongs to the field of equipment fault diagnosis, and discloses an unsupervised cross-equipment fault diagnosis method fusing dual alignment and pseudo-label learning. Through collaborative optimization of three mechanisms of global domain adaptation, conditional domain confrontation and pseudo-label learning, depth feature alignment and refining of three levels of global-local-instance are realized, and the accuracy, robustness and generalization ability of unsupervised cross-equipment fault diagnosis are significantly improved. The core technical problems of confusion of different fault category features on a target domain and low diagnosis precision due to the fact that an existing domain adaptation method only pays attention to global distribution alignment and ignores a category structure are solved, and the technical bottleneck that when a traditional intelligent diagnosis model is applied to new equipment or new working conditions, the performance can be guaranteed only by depending on label data is solved.

The invention discloses an action sports scoring method and device based on big data analysis, and relates to the technical field of computer vision, and the method comprises the following steps: S1, constructing a human body posture tensor manifold; s2, generating aligned action track characteristics; s3, generating symmetrical positive definite manifold features; s4, generating deep manifold distribution parameters; s5, quantifying global distribution deviation characteristics; s6, calculating a residual vector based on the deep manifold distribution parameters and the standard action distribution parameters, inputting the residual vector into the improved AGCN model for processing, constructing an adaptive topology based on the residual vector, and performing aggregation analysis to obtain a joint-level physical angle error after multi-scale space-time convolution and manifold enhancement; and S7, outputting a comprehensive score vector. According to the method, the limitation that a traditional method only depends on local geometric features and ignores global topological structure constraints and statistical distribution priori is overcome, and an efficient solution is provided for intelligent scoring of sports actions.

The invention discloses a logistics unmanned aerial vehicle material distribution energy consumption and communication collaborative optimization method. The method comprises the steps of collecting distribution task data of an unmanned aerial vehicle and communication deployment data in a flight environment; establishing an unmanned aerial vehicle distribution system model based on the distribution task data and the communication deployment data, and constructing an optimization target to optimize the total energy consumption and the total communication interruption time of the unmanned aerial vehicle for completing the distribution task; based on the unmanned aerial vehicle distribution system model, solving a global distribution sequence with the lowest energy consumption; taking the global distribution sequence as input, converting an unmanned aerial vehicle trajectory planning problem into a Markov decision process, and solving a local distribution trajectory in the unmanned aerial vehicle distribution process; on the basis of the local distribution track, a Bayesian optimization model and a deep reinforcement learning network are adopted to obtain an actual flight action of the unmanned aerial vehicle at the next moment and execute the actual flight action; and based on the real-time environment information, generating an optimal flight path satisfying energy consumption and communication collaboration, and completing a distribution task. According to the invention, an efficient and safe distribution system is realized.

The invention relates to the technical field of cold chain management, and discloses a cold chain collaborative management method and system based on cloud edge collaboration, and the method comprises the steps: obtaining the temperature, humidity and inventory data of a multi-node sensor and the global distribution point path information; according to the temperature and inventory data, edge data aggregation, noise filtering and temperature threshold monitoring are carried out, and a local inventory adjustment demand is obtained; according to the local inventory adjustment demand and the global distribution point path information, performing global path optimization analysis, multi-path variation cross set generation and comprehensive evaluation sorting to obtain a global distribution path set; and performing node conflict detection, comprehensive decision and adaptive adjustment on the global distribution path set to obtain final optimal path distribution. According to the method, global efficient collaborative management of cold-chain logistics and dynamic optimization of a distribution path can be realized.

The invention belongs to the technical field of network abnormal behavior detection, and particularly relates to an abnormal behavior detection method based on a TransGAN network. The method comprises the following steps: firstly, dynamically evaluating feature importance by utilizing a time sliding window mechanism in combination with a time sequence sensitive XGBoost algorithm, and screening a basic feature set and an attention key feature set through double thresholds; then, a TransGAN model in which a generator and a discriminator cooperatively work is constructed, the generator adopts a position sensing embedding technology to strengthen key feature focusing and introduces a gating mechanism to dynamically adjust basic feature weight, and the discriminator fuses depth separable convolution to extract local features and carries out global distribution comparison with a lightweight Transform decoder; and finally, through a dual-channel detection assembly line and a reconstruction channel, calculating residual features of input data and generating output, judging channel generation anomaly confidence, and fusing the two to form a comprehensive anomaly score to realize accurate judgment. The accuracy, robustness and real-time performance of anomaly detection are improved, and the method is suitable for various anomaly detection scenes.

The invention discloses a distributed chip security domain system based on an earth clone scale model, and the system comprises a modeling unit which is used for building a geographic grid according to the real latitude and longitude of the earth; the management unit is used for dividing the geographic grid into security domains corresponding to countries or regions one by one, each security domain is used for generating and storing a public and private key pair, a private key is embedded in the distributed chip security domain, and a public key is uploaded to a global distributed block chain network; and the authentication unit is used for carrying out cross-region security verification based on the security domain by utilizing a universe anchor point. The invention further provides a digital integrated authentication method and the eSIM mobile phone. According to the invention, independent COS examples are preset for each country in a distributed chip, and a total factor large Token and zero-knowledge proof technology are combined, so that identity authentication of a user at any place in the world is ensured, and integrated digital identity authentication with high safety, strong privacy and global mutual recognition is realized.

The invention discloses a missing modal learning method based on hierarchical consistency prompt distillation, relates to the field of computer vision, natural language processing and multi-modal information fusion, and aims to solve the problem that anchor drift and judgment boundary collapse are easy to occur in existing multi-modal learning under a modal missing condition. A modeling mechanism with cooperation of structure perception, anchor point guidance and layered distillation is provided, and stable alignment and high-quality prediction of the semantic space are achieved under the scene of any missing combination and high missing rate. The method comprises a structure perception representation modeling module, a double-layer semantic anchoring module, a hierarchical consistency prompt distillation module and a deletion perception classification and overall optimization module. According to the method, a three-layer semantic consistency framework is provided around anchor drifting, and modal-level consistency, fusion-level consistency and global distribution consistency are constrained in a shared semantic space at the same time, so that the class directions represented by modals and fusion are unified, boundary collapse is inhibited, and structural stability is kept and generalization robustness is improved in a modal missing scene.

The invention discloses an anomaly monitoring method for panoramic smart energy cloud edge collaboration, and the method comprises the following steps: S10, edge side credibility perception calculation: carrying out the real-time analysis of locally collected energy time sequence data through an edge side device, generating a preliminary anomaly probability Pe, and carrying out the calculation of the edge side credibility based on the probability Pe and the corresponding data; calculating an edge credibility score Re corresponding to the preliminary anomaly probability Pe by analyzing the difference with a historical normal mode and the uncertainty of local model fitting; and step S20, cloud credibility enhancement analysis: the cloud performs deep analysis according to the data uploaded by the edge side, generates a cloud anomaly probability Pc, and calculates a cloud credibility score Rc corresponding to the cloud anomaly probability Pc by analyzing the novelty of the data in global distribution and the stability of model prediction based on the probability Pc and the corresponding data. According to the invention, abnormity monitoring of panoramic smart energy cloud edge collaboration can be carried out more comprehensively.

The invention discloses a super-multi-objective optimization method and system based on dual-file coevolution and dual-distance indexes, and the method comprises the steps: generating an initial population based on the parameters of a power system, initializing the parameters and a reference vector set, and storing the population into a convergence file and a diversity file; calculating an ideal point and a worst point based on the convergence file, and updating the convergence file in combination with a double-distance index; maintaining population global distribution through a reference vector association operation; evaluating a population distribution state in combination with information entropy to determine a local neighborhood size; calculating individual local neighborhood density based on the parallel distance and updating a diversity file; constructing a mating pool based on the double archives and generating a filial generation population; and iteratively executing until the non-dominated solution set in the diversity file is output. According to the method, the convergence and diversity retention capability of the algorithm to the complex leading edge is enhanced, and a power system scheduling scheme with multi-target balance of the power generation cost, the transmission loss, the line utilization rate and the power generation adjustment amount can be efficiently generated.

The present invention discloses an adaptive exponential decay method for decaying the offline model prediction value of the recommended task for the worker according to the real-time competition situation in the crowdsourcing scenario. Based on the distribution difference of the prediction value of different offline prediction models, the present invention designs an algorithm for adaptively adjusting the exponential decay parameter according to the local distribution and global distribution of the offline prediction value. In this model, the global distribution determines the upper limit of the exponential decay intensity, and the local distribution determines whether this upper limit can be reached, wherein the absolute distribution of the local distribution is used to determine the increase or decrease of the initial local distribution intensity, and the relative distribution determines the intensity of the increase or decrease. After adjusting the hyperparameters, the model has the ability to generalize between different offline models. This model stabilizes the problem of abnormal fluctuations in the indicators related to the recommendation list caused by the non-independence between workers in the crowdsourcing scenario.

The utility model discloses an insect sticking trap device and system, the device includes base body and sticking trap piece, and the base body is strip structure, and the base body includes first base plate and second base plate, and first base plate and second base plate are arranged in the included angle, and the inner surface of second base plate is equipped with sticking trap area and escape area, and escape area is equipped with the one side away from first base plate in sticking trap area, and sticking trap piece is equipped in sticking trap area, and sticking trap piece is equipped with the colloid for sticking insect. The utility model discloses compared with prior art plane formula sticking trap board single orientation's sticking trap surface, can utilize the included angle structure and enlarge effective sticking trap range, and adapt to the flight and crawl track of various insects, and improve the insect catching efficiency, and simultaneously, the inner surface of second base plate divides sticking trap area and escape area, and sticking trap area sets up the sticking trap piece with colloid, avoids the problem that the colloid of existing device global distribution is easy to stick non -target organism, and the problem of early failure due to accidental touch or dust adhesion, and can reduce the non -target organism's mistake through escape area and stick, and need not use the inducing agent, and reduce the cost.

The application discloses a tower crane detection method and system based on deep learning, and the method comprises image acquisition, initial image optimization, detail enhancement, aggregation correction, establishment of a tower crane detection model and tower crane detection. The application belongs to the field of tower crane detection, and specifically relates to a tower crane detection method and system based on deep learning. The application pertinently adjusts and attenuates channels, and improves the color authenticity of images. A global region extension and local distribution enhancement strategy is adopted, aggregation correction is based on, and false detection and missed detection caused by information loss are reduced. The feature dispersion degree in a group is quantified, an environment self-adaptive penalty is introduced to design and optimize a loss function, and the penalty for environmental interference is dynamically adjusted. A dispersion matrix and a shielding penalty coefficient are introduced to quantize the global distribution dispersion degree of tower crane features, process feature variation caused by tower crane shielding and component loss, avoid false judgment of shielding, and reduce the missed detection rate. Therefore, the detection effect is improved.

The embodiment of the specification provides a multi-node distribution method and device for a large model. The large model corresponds to a plurality of model files. A target model file is pre-divided into a plurality of file blocks, and each file block is pre-divided into a plurality of shards. The method comprises the following steps: a target node sends a download request to a scheduler, wherein the target model file is indicated; the scheduler selects a plurality of candidate nodes to join a candidate parent node set from candidate nodes in a cluster based on at least index values of state indicators of the candidate nodes and global distribution information of the shards on the candidate nodes, the candidate nodes having downloaded at least one shard, and the candidate nodes and the target node at least satisfying topology connectivity and not forming a loop; the scheduler determines a target file block to be preferentially downloaded by the target node from the plurality of file blocks; the scheduler sends the candidate parent node set and a block identifier of the target file block to the target node; and the target node preferentially downloads shards in the target file block from nodes in the candidate parent node set.

The invention discloses an unsupervised multi-modal equipment instance alignment method based on double-space embedding, and the method comprises the steps: employing a pseudo seed generation module, calculating the similarity of visual and text modal features, automatically constructing a preliminary pre-alignment instance pair, continuously expanding a high-confidence seed set in an iteration process, and carrying out the alignment of an unsupervised multi-modal equipment instance. The method comprises the following steps: firstly, carrying out data labeling on equipment instances, so as to effectively reduce the dependence on manual data labeling, then respectively carrying out embedding learning in Euclidean space and hyperbolic space by adopting a double-space embedding strategy, capturing a local structure relationship between the equipment instances, and carrying out accurate modeling on a hierarchical structure and global distribution characteristics of the equipment instances, and then, by constructing an adaptive feature fusion module, carrying out dynamic weighted integration on the embedded representations of the two spaces, and by introducing an iterative constraint mechanism and utilizing a loss function, ensuring that the representations of the same instances in the double spaces are as close as possible and the distance between different instances keeps enough discrimination.

The invention discloses a distributed storage deployment method and device based on cloud computing and a medium, and the method comprises the steps: determining the number of storage servers, determining a distribution model according to the number of storage servers, determining a plurality of rack servers according to the distribution model, and determining the storage number of the rack servers, loading the servers corresponding to the storage number into a rack server; a data deployment mode is determined according to the distribution model, the data deployment mode comprises global distribution and local distribution, and storage nodes are determined in the multiple rack servers according to the data deployment mode; and sending the written data to a storage node so as to store the data through the storage node. Through accurate resource configuration, a flexible data deployment mode, redundancy guarantee of data copies and dynamic resource adjustment, the performance and reliability of the storage system are remarkably improved.

The invention discloses a deep learning-based tower crane detection method and system. The method comprises the steps of image acquisition, primary image optimization, detail enhancement, aggregation correction, establishment of a tower crane detection model and tower crane detection. The invention belongs to the field of tower crane detection, and particularly relates to a tower crane detection method and system based on deep learning. According to the scheme, an attenuation channel is adjusted in a targeted mode, and the image color authenticity is improved; a global regional extension and local distribution enhancement strategy is adopted, and based on aggregation correction, false detection and missing detection caused by information missing are reduced; quantifying intra-group feature dispersion, introducing an environment adaptive penalty design optimization loss function, and dynamically adjusting environment interference penalty; a divergence matrix and a shielding penalty coefficient are introduced, the global distribution divergence of tower crane features is quantified, feature variation caused by tower crane shielding and part missing is processed, misjudgment shielding is avoided, and the omission ratio is reduced; and the detection effect is improved.

The application relates to a global distributed computing power resource access method, device, program product and medium, and relates to the technical field of distributed computing. The method comprises the following steps: acquiring application layer response data of a computing power service endpoint, determining a target endpoint and a computing load evaluation value based on the data, and constructing a resource state table in combination with a network quality index, so that double evaluation of endpoint processing load and network transmission quality is realized; further, a candidate endpoint set is screened from the resource state table based on a service type identifier, and a comprehensive score is calculated by comprehensively considering the load evaluation value, the network quality index and a resource charging unit price, so that multidimensional trade-off of service quality and cost factors in the endpoint selection process is realized; finally, a flow type mark of a computing power request is identified, and a transmission strategy is determined according to the mark, so that differentiated transmission processing of different types of computing power requests is realized, and end-to-end response time delay is reduced.