207results about How to "Reduce memory consumption" patented technology

The invention discloses an insulator defect detection method based on feature extraction and cross-scale fusion, and belongs to the technical field of insulator defect detection methods. According to the method, a defect detection model is obtained through training to find defects existing in a detected insulator image, and the method comprises the following steps: S1, inputting an image data set and a corresponding label data set; s2, designing a lightweight fast multi-scale extraction module to complete feature extraction; s3, designing an efficient feature fusion module to complete feature fusion; s4, selecting image features from the output of the cross-weighted bidirectional feature pyramid network through intersection-to-parallel ratio perception query selection; and finally, obtaining a confidence coefficient score and a bounding box of the target through a solution terminal, and further finishing decoding. The method effectively solves the problems that an existing detection method is insufficient in feature extraction capacity in the feature extraction stage and large in model parameter in the feature fusion stage.

The invention provides a GCN-based honey situation map analysis method, and relates to the technical field of neural networks. Node information and side information are obtained, a honey array subgraph is constructed, and a # imgabs0 #-order Laplacian matrix is obtained; updating and recurring a characteristic value of the # imgabs1 #-order Laplacian matrix based on the maximum characteristic value of the honey array subgraph to obtain a normalized Laplacian matrix; performing random feature masking on each node to obtain a plurality of updated node feature matrixes; and obtaining node outputs of the plurality of honey array sub-graphs on the plurality of levels based on the updated node feature matrix, obtaining a prediction value, carrying out iteration on parameters of the graph neural network based on the prediction value, obtaining an updated graph neural network, and generating a prediction result. According to the invention, the convolution calculation is carried out on the graph signal, and the Laplacian matrix and the Chebyshev polynomial of the graph are combined, so that an efficient graph convolution kernel is constructed, and the accuracy and the performance of network situation graph analysis are improved.

The invention relates to the technical field of computers, in particular to a fault report generation method based on the size of a dynamic window in an edge cloud scene, which comprises the following steps of: S1, performing data processing according to acquired data to obtain preprocessed data, and inputting the preprocessed data into a pre-training model for training to obtain an adjustment model; s2, multi-modal input data generates a multi-modal KV cache according to the adjustment model, encoding is carried out to obtain an encoding set, KV key value pairs are screened according to the encoding set, decoding is carried out after the KV key value pairs are combined, KV cache compression is carried out on the decoded data through Page Attelation to obtain matched compressed data, and a window is dynamically adjusted according to the compressed data; and S3, receiving second real-time operation state data of the edge equipment, generating a second historical fault report, and judging whether the second historical fault report meets the model requirements or not. Through LOOK-M and paged attention technologies, KV cache of a large model is reduced, memory use and calculation efficiency are optimized, model fine tuning is introduced, and it is ensured that a generated fault analysis report has high accuracy and real-time performance.

The invention relates to the technical field of quality evaluation, and discloses a lightweight beef quality evaluation method based on unsupervised feature learning, and the method comprises the steps: obtaining original hyperspectral data of beef as real data, and generating high-quality sample data based on a generative adversarial network model; preprocessing the high-quality sample data, mapping, reconstructing and fusing the preprocessed sample data, extracting spectral and spatial features of different levels, and fusing or splicing the spectral and spatial features with feature values of other auxiliary data; the fusion sample features are mapped and fused, sample data of specific tiny disturbance similar to real data are generated, the sample data and the real data are alternately input into a Transform model, and parameters of the Transform model are updated; the beef quality is evaluated on the basis of the Transform model with the updated parameters; according to the method, the accuracy, objectivity, rapidness, non-destructive property and comprehensiveness of beef quality evaluation are improved.

The invention discloses a document comparison method based on an improved LCS algorithm, and relates to the field of data processing, and the method comprises the steps: receiving N to-be-compared documents uploaded by a user; dividing the N documents into N groups, wherein each group comprises a main document and N-1 auxiliary documents; setting a minimum common subsequence length threshold value L0; for each group, constructing a virtual two-dimensional array, taking the character number of the main document as the row number, taking the character number of the auxiliary document as the column number, adopting a sparse matrix for storage, and only recording matched character positions; traversing the elements of the first row and the first column row by row from the elements of the last row and the last column of the virtual two-dimensional array along the main diagonal direction of the two-dimensional array, and recording the content of continuous matching characters and the starting positions and the ending positions of the continuous matching characters in the main document and the auxiliary document; in view of low comparison efficiency of tobacco purchasing files in the prior art, optimization strategies such as document grouping, multi-thread parallel comparison and clustering combination are introduced, so that the comparison efficiency is improved.

The invention discloses a laser radar positioning and three-dimensional mapping method based on an implicit neural field, and belongs to the technical field of positioning and mapping. The method comprises the following steps: for first frame data in radar point cloud data, converting an input sampling point from a sensor coordinate system to a world coordinate system by using a unit matrix or a constant matrix to initialize a map; the hierarchical potential features are optimized in a self-supervision mode, an octree data structure is used for storing the optimized potential features to generate a potential feature set, and features in the potential feature set are decoded into symbol distance values through a shallow multilayer perceptron (MLP); starting from a second frame of data of the radar point cloud data, obtaining a predicted pose through a matching mode from the frame to the implicit map every time a frame is input; converting a new input point into a world coordinate system by using the obtained predicted pose, and determining whether a new potential feature needs to be added or not according to a Morton value of the new input point; and completing the construction of the NSDF by repeating the operations of predicting the pose and expanding the NSDF.

The invention discloses a fan unmanned aerial vehicle autonomous inspection method and system, and the method comprises the steps: building a fan three-dimensional model based on the collection data of a laser radar for a fan, and enabling the fan three-dimensional model to be constructed through a recurrent neural network; planning an inspection route of the unmanned aerial vehicle; setting an intelligent inspection APP, and determining an autonomous flight path of the unmanned aerial vehicle; automatically identifying the to-be-detected image through the fan defect detection module, determining the fan defect type and positioning the fan defect position; and counting the defect positions of the fan and outputting an analysis inspection report to finish inspection. According to the routing inspection method, the laser radar, the edge calculation module and the like are integrated, the hardware requirement of fan routing inspection is met, and unmanned routing inspection work of the wind driven generator is greatly facilitated.

The invention relates to a hyperspectral image classification method based on a meta-pseudo-label multi-scale image convolution network, and the method comprises the steps: building a meta-pseudo-label multi-scale feature search image convolution network (MPL-MFSGCN), dividing the HSI into superpixel images of different scales through employing an SLIC superpixel segmentation technology, and converting the superpixel images into image structures; secondly, constructing a multi-scale feature search graph convolutional network MFSGCN architecture, optimizing a superpixel feature weight combination by using a neural network search algorithm, and enhancing the discrimination of feature expression; a semi-supervised teacher-student model sharing the same graph convolutional network is adopted to generate meta-pseudo tags, and a student network is finely adjusted to improve the robustness of the network; the key features can be automatically captured under the small sample condition, the classification precision higher than that of an advanced method is obtained compared with a common HSI data set, and the classification accuracy is effectively improved.

The invention relates to a GPU-based high-IO perceptibility long-sequence text injection method, medium and equipment, and the method comprises the steps: obtaining and coding a long-sequence text input, processing a coded matrix vector through a multi-head attention mechanism, and realizing text injection, the matrix vector comprises a query Q, a key K and a value V. The method is characterized in that the value V is equal to the query Q, the key K and the value V. In the multi-head attention mechanism, the similarity is calculated by using an L2 distance, the matrix vector is segmented into blocks to realize thread block parallel processing during forward propagation, and the similarity-based result matrix and the gradients of the query Q, the key K and the value V are recalculated by using a recalculation strategy during back propagation. Compared with the prior art, the method has the advantages of reducing the overhead of attention calculation memory access, improving the operation efficiency and the like.

The invention discloses a method, a system, equipment and a medium for analyzing biological characteristic correlation in biomics data, and belongs to the technical field of bioinformatics. In particular to a method, a system, equipment and a medium for analyzing correlation of biomics data. According to the method, the system, the equipment and the medium, a high-performance and parallel block matrix algorithm is used for processing large-scale correlation and clustering tasks, excellent performance is shown in the aspect of calculating a correlation matrix, and downstream analysis can exceed computer memory resources; according to the method, the approximate P value can be quickly and effectively estimated without repeated complex cumulative distribution function calculation, and the corresponding approximate correction P value can be quickly and effectively estimated based on the P value sampling subset without being based on all the P values; and a top-k algorithm is realized, and faster execution and lower memory consumption are realized by utilizing a partial sorting method based on heaps.

The invention relates to an operation log recording method, device and equipment based on an SPI mechanism. The method comprises the steps that a loading operation log package is started, and log variable parameters are read; starting a log component to form log information, and determining a log dropping mode according to the number of logs to be processed; scanning, loading and defining an SPI function, and dropping the to-be-processed log into a disk according to the log dropping mode; a monitor is initialized, and whether the system runs normally or not is judged through the monitor. At least code readability and maintainability can be improved, memory consumption is reduced, performance is improved, and compatibility is improved.

The invention provides a heat sink preparation optimization method and device and a readable storage medium, and the method enables the process parameters of process control models corresponding to at least two adjacent deposition layers in the same heat sink to be different. The process parameters of the process control models corresponding to the deposition layers formed in the heat sink generation process and the performance parameters of the heat sink are collected to generate multiple sets of samples, and the models corresponding to the different deposition layers are trained through the samples and qualification judgment of the samples, so that the process control models are improved to optimize the performance of the deposition layers. And finally, the purpose of optimizing the comprehensive performance of the heat sink is achieved. The scheme provided by the invention is applicable to preparation of heat sinks with various thicknesses, the problems of defects and non-uniform performance in the production of the heat sinks can be effectively avoided, diversified and differentiated effects can be embodied on each deposition layer of the heat sinks, the different deposition layers can have the advantage of different directionality, and the production efficiency of the heat sinks is improved. And thus, the finally generated heat sink is ensured to have more excellent comprehensive performance.

The invention specifically discloses a low-overhead burst data stream detection algorithm, which filters and identifies an input data stream by using a constructed small stream filter and a burst stream detection table so as to output a real burst stream, and continuously filters a low-frequency stream by setting a time period so as to identify a potential burst stream. The detection precision of the subsequent burst flow is improved; meanwhile, a burst flow detection table is used for tracking potential burst flows, and whether the potential burst flows are real burst flows or not is recognized based on the frequency and the sudden increasing period of each potential burst flow in two adjacent time periods; moreover, a replacement strategy of the burst period and the data stream frequency probability is adopted to eliminate the data stream with lower burst, so that potential burst streams with higher possibility can be stored, and the memory consumption is further reduced. Therefore, the burst data stream detection algorithm provided by the invention has the characteristics of high accuracy, low overhead and high throughput.

The invention provides a method for extracting a velocity spectrum from DAS data, and the method comprises the following steps: dividing the DAS data of a whole well section into a plurality of data blocks according to a set depth interval, and obtaining a frequency-wave number diagram corresponding to each data block; performing grid division on each frequency-wave number diagram; obtaining a power spectrum of each grid; setting a plurality of slope lines with different slope values through the original point of the frequency-wave number diagram, wherein each slope value is within a preset slope value range; aiming at each frequency-wave number diagram, an integral value of each set slope line is calculated, and the integral value is a weighted average value of a length value of the slope line in each grid penetrated by the slope line and a power spectrum in the grid; the slope value of the slope line corresponding to the maximum integral value is used as a static speed value corresponding to the data block where the slope line is located; splicing the static velocity values corresponding to the data blocks to obtain a static velocity spectrum corresponding to the DAS data of the whole well section; and according to the static velocity spectrum and the fluid velocity in the wellbore, velocity spectrums of upgoing waves and downgoing waves in the wellbore are obtained.

The invention discloses a current transformer error prediction method and system, and relates to the technical field of online monitoring and error evaluation of current transformers, and the method comprises the steps: collecting signal data from equipment, and carrying out the preprocessing; carrying out dimension reduction processing on the preprocessed data features; and performing feature selection and prediction on the feature data after dimension reduction, and outputting a result. According to the invention, by combining the time domain, frequency domain and wavelet domain features of the current transformer monitoring signal, the multi-domain feature prediction model is constructed, and the capability of accurately identifying the operation condition of the current transformer is improved. And a CatBoost algorithm is used to evaluate the importance of the features, so that the feature interpretability of the model is improved, and the effectiveness of model training and the accuracy of prediction are ensured. And HistGB is selected as a regression model, so that the training speed is greatly improved, and the calculation cost and memory consumption are remarkably reduced when a large data volume is processed. According to the invention, quantitative analysis and timely prediction of the error of the high-voltage current transformer are realized.

The embodiment of the invention provides a virtual processor context switching method and device, a storage medium and electronic device.The method comprises the steps that under the condition that a current virtual processor receives an interrupt signal, the affinity attribute of a current task executed on the current virtual processor is determined; under the condition that the affinity attribute represents that the virtual processor associated with the current task is the first virtual processor, predicting the switching probability of the current task; under the condition that the switching probability is greater than or equal to a preset probability threshold value, pre-caching context information of the first virtual processor in a cache region; when a context switching request is received, the current task is interrupted, the context information of the first virtual processor is extracted from the cache region and loaded to the first virtual processor, and the interrupted current task continues to be executed on the first virtual processor. And the problem of low virtualization real-time performance is solved.

The invention discloses a data leakage prevention method and device, computer equipment and a storage medium, and belongs to the technical field of big data. According to the application, a pre-constructed log cloud platform is deployed at a cloud end, and the log cloud platform is accessed to a service system; deploying a request interceptor in the service system, and connecting the request interceptor with each interface; monitoring by using a docking interface, and when a user request is monitored, obtaining the request; obtaining log data matched with the request, and intercepting a log field corresponding to the identity information and the access address information from the log data; synchronizing the log field to a log cloud platform, and performing log anomaly detection on the log field by using a log anomaly detection model to obtain a detection result; and determining the data query permission of the user according to the detection result. The invention further relates to the technical field of block chains, and the log data are stored on the block chain nodes. According to the invention, the data leakage prevention capability of the system is improved, and the memory consumption of the system is reduced.

The invention discloses a novel power system output regulation and control strategy generation method and device and a control system, and belongs to the technical field of power system regulation and control. The output regulation and control strategy generation method comprises the following steps: representing net load historical data corresponding to net load climbing demand uncertainty characteristics by adopting a Gaussian mixture model; the uncertainty of the net load climbing requirement is reduced; inputting the net load day-ahead prediction data corresponding to the net load historical data and the net load uncertainty scene set into a two-stage risk avoidance model based on the lowest operation cost and the worst condition risk probability assessment theory; the two-stage risk avoidance model divides the whole strategy generation process into day-ahead optimization scheduling plan generation and intra-day real-time scheduling plan generation; the day-ahead optimal scheduling plan is optimized by using the obtained intra-day real-time scheduling plan to obtain a safe and economic output scheduling strategy; and furthermore, the novel power system is controlled in different scenes to realize efficient, stable and economical operation.

The invention discloses a traveling salesman problem solving method based on a quantum self-attention neural network, and belongs to the field of quantum computing. A classic encoder-decoder structure is adopted, a quantum self-attention layer is designed in the encoder and comprises a quantum multi-head attention layer and a quantum full-connection feed-forward layer, and a variable component sub-circuit is used in a sub-layer to replace a full-connection layer of a classic neural network. Firstly, an encoder receives an input data set, node features are embedded into a quantum self-attention layer, and a complex relation between nodes is extracted through a quantum multi-head attention mechanism. The decoder then stepwise outputs the sequence of nodes to form an optimal path according to the embedding generated by the encoder. In the model training stage, a REINFORCE deep reinforcement learning algorithm is utilized, the total path length serves as an optimization target, and model parameters are updated through strategy gradients. Compared with the existing algorithm, the method has the advantage that the performance is well improved.

The invention provides a memory allocation method and device for an NPU heterogeneous platform, and belongs to the technical field of neural networks, the memory allocation method for the NPU heterogeneous platform comprises the following steps: acquiring a first target operator in the NPU heterogeneous platform, and performing memory allocation on an input tensor and an output tensor of the first target operator and the first target tensor; a second target operator in the NPU heterogeneous platform is obtained, the memory of the output tensor of the second target operator multiplexes the memory of the input tensor of the second target operator, and the second target operator is an operator supporting in-situ operation; and performing memory allocation on each tensor of each third target operator one by one according to a sequence of operator widths in operator configuration information of the third target operators in the NPU heterogeneous platform from large to small. According to the memory allocation method for the NPU heterogeneous platform, static memory allocation is carried out based on combination of time and space information of the NPU heterogeneous platform, the processing efficiency can be effectively improved, memory reuse is achieved, and a large amount of memory consumption is reduced.

The invention discloses a flow field data order reduction processing method and system for atmospheric pollution prediction, and belongs to the technical field of atmospheric pollution prediction. An existing air pollution data processing scheme needs to process a large amount of data, the calculation time is generally long, the memory consumption is high, and the analogue simulation cost is high. According to the flow field data order reduction processing method for atmospheric pollution prediction, large-scale original flow field data is decomposed into discrete blocks, namely different sub-matrixes, and each sub-matrix can be subjected to decomposition calculation, so that large-scale simulation data can be processed under limited hardware resources, and the calculation efficiency is improved. And the decomposition calculation speed is increased. And meanwhile, singular value solution is performed on each sub-matrix to obtain a flow characteristic primary function, so that original flow field data can be accurately described by using a small number of primary functions, the purpose of data order reduction is realized, the requirements on computing resources and equipment are further reduced, and the analogue simulation cost is reduced.

The invention relates to the crossing field of agricultural biotechnology and information technology, and discloses a smart breeding system for hybrid rice, comprising: a smart breeding data acquisition and processing system for collecting and processing genotype data, environmental factors and seed phenotype data of rice parents and offspring; the intelligent breeding model parallel optimization system is used for optimizing the parallel calculation process, and the invention further discloses a hybrid rice breeding method of the intelligent breeding system, and the method comprises the following steps: collecting rice genotype, environmental factor and seed phenotype data; processing and analyzing data by using the intelligent breeding model, and training a prediction model; predicting hybrid rice characters through a prediction system, and generating a report; and formulating a breeding scheme according to the report and performing field verification. By integrating an advanced information technology, comprehensive intelligentization and high efficiency of hybrid rice breeding are realized, the breeding period is shortened, the cost is reduced, and the system performance is continuously improved through a closed-loop iterative optimization mechanism.

The invention discloses a data association relationship determination method, system and device, a medium and a program product applied to an artificial intelligence technology, and the method comprises the steps: taking to-be-associated data of a data association task and all source data in associated data information as node features, and inputting the node features into an association relationship recognition model; performing graph sampling on the incidence relation identification model, aggregating feature information of each sampling node and adjacent neighbor nodes based on a target polynomial, and determining an incidence relation identification result between the sampling nodes corresponding to the current graph sampling according to the current aggregation feature; and determining target data with an association relationship in the association data information for the to-be-associated data according to each association relationship identification result. The problem that task precision and task execution efficiency cannot be considered at the same time in the prior art can be solved, and on the basis of improving the task execution efficiency determined by the data association relationship, it is ensured that a high-precision data association result is output.

The invention discloses a method and equipment for reconstructing a three-dimensional scene, and relates to the field of computer graphics. The method comprises the following steps: simulating the process of light from the optical center of a camera to a three-dimensional scene through a light advancing module, and determining space coordinates of a plurality of sampling points in the three-dimensional scene; decomposing the color feature space and the density feature space, and determining a matrix and a vector pair corresponding to the space coordinates of each sampling point; interpolating each matrix and vector pair, and determining a color feature vector and density of each sampling point; determining a spherical harmonic feature vector of each sampling point according to the color feature vector of each sampling point and a spherical harmonic code of surface modeling; determining an anti-aliasing feature of each sampling point according to the color feature vector and the cone code of each sampling point; splicing the spherical harmonic feature vectors and the anti-aliasing features, and inputting a predicted color of each sampling point output by a target multi-layer perceptron; and carrying out volume rendering on the predicted color and density to determine the reconstructed three-dimensional scene, and reconstructing the three-dimensional scene with high efficiency and high quality.

The invention relates to a memory allocation method and device, computer equipment and a storage medium. The method comprises the steps that a target model is acquired, tensor information of each tensor generated in a topological structure based on the target model is determined, the tensor information comprises a use cycle, and the use cycle indicates an input operator and an output operator of the tensor; according to the tensor information of each tensor and the topological structure of the target model, operator information of operators contained in the topological structure is determined, and each operator is matched with at least one tensor; sorting the operators according to the operator information, and determining the priority of each tensor according to the sorting result of the operators and the tensor matched with the operators; and in response to the priority of the tensor, determining that the tensor which is not subjected to memory allocation exists in the tensor, and performing memory allocation on the tensor which is not subjected to memory allocation at least according to the input operator and the output operator indicated by the use cycle of the tensor which is not subjected to memory allocation. By adopting the method, memory resources can be saved.

The invention discloses a GPU (Graphics Processing Unit) parallel acceleration method for subdivision volume parameter evaluation, which comprises the following steps of: S1, storing and traversing a volume grid through a data structure containing a half side and a half surface, and judging the type of each volume unit; s2, performing CC body subdivision under any topology once, and isolating irregular vertexes if a plurality of irregular edges in a body unit must converge at an irregular point; s3, aiming at different topologies, numbering control grid vertexes by using a data structure containing a half surface and a half side, obtaining a subdivision matrix according to a numbering sequence and a CC subdivision rule, and pre-calculating a subdivision matrix feature structure; and S4, performing parallel parameter evaluation on different types of volume grid units. According to the method, the optimized parallel algorithm is designed, the multi-core architecture and the floating-point operation capability of the GPU are fully utilized, and the parameter evaluation speed of the complex body subdivision model is effectively improved.

The application discloses a low-depth sequencing population genotype filling calculation memory optimization method, which comprises the following steps: carrying out block division on a single nucleotide polymorphism site (SNP), and setting a checkpoint value according to the block division; and carrying out calculation on a forward auxiliary variable and a backward auxiliary variable according to the set checkpoint value, and calculating a haplotype observation conditional probability according to the block division. The application has the advantages of simple principle, effective reduction of calculation memory consumption, improvement of genotype filling efficiency and the like.

The invention discloses a lightweight fire hazard multi-target detection method in an electric power environment. The method comprises the following steps: S1, constructing a lightweight fire hazard multi-target detection data set in the electric power environment; s2, a preset RT-DETR model is improved, and an improved RT-DETR model is obtained; s3, training the improved RT-DETR model by using the data set, and optimizing model parameters through an improved loss function to obtain a lightweight fire hazard multi-target detection model in the electric power environment; and S4, performing multi-target detection by using the lightweight fire hazard multi-target detection model in the electric power environment. According to the invention, high-precision disaster and hidden danger multi-target detection is realized.

The embodiment of the invention provides a data storage method, a data retrieval method and related devices. The method comprises the following steps: storing multiple paths of storage data in a first database, and generating an index according to a storage position of the storage data in the first database; generating integrated data according to the storage data and the index; storing the integrated data into a second database to realize data storage; the method comprises the steps of extracting first data corresponding to storage data from to-be-retrieved data; according to the matching degree of the data in the first database and the first data, searching in the first database to obtain second data, and determining a target index corresponding to the second data; searching in a second database according to the target index to obtain target data so as to realize data retrieval; by combining the first database and the second database, all data contents can be quickly accessed, and the memory consumption of the system is reduced; quick vector data retrieval matching can be carried out, so that the retrieval speed is improved under the condition of mass data.

The invention discloses a method and a system for calling local functions among different Web applications under the same Tomcat instance, and relates to the technical field of computer software, a plurality of Web applications are deployed under the same Tomcat instance, and the applications needing to provide local function calling are set to be started preferentially; in the Web application which is started preferentially, a new class is generated, and an object instance is created; the new class and the object instance are loaded to a system class loader, so that other Web applications can globally access the new class and the object instance; when other Web applications in the Tomcat are started, obtaining an object instance from the system class loader, and distributing the object instance to a target instance variable in the proxy class; other Web applications call the object instance through the proxy class; and the proxy class returns the calling result of the target instance variable, so that network delay and performance overhead caused by traditional remote procedure calling are avoided, and the calling efficiency between applications is greatly improved.