54results about How to "Save memory space" patented technology

The invention provides a time-space dynamic graph construction method and system for an engineering construction project, and relates to the field of railway tunnel engineering management.The method comprises the steps that construction data entities are extracted, and an entity set is generated; taking entities in the entity set as nodes to be added into the graph, and configuring attributes for each node; adding a relation edge for each node according to a logic structure between the construction data; adding a time label and a space label to each node so as to form a data graph; associating each entity in the data atlas with a time chain and a space chain to form a data atlas attached to the time-space chain; combining a node merging strategy and a graph neural network to optimize a graph structure; and dynamically modeling dimensional changes of entities and relationships in time by means of a time sequence diagram neural network. According to the method, the dynamic change graph of the relationship between the construction data entities is constructed, so that the spatial-temporal information and the entity relationship of each process are accurately expressed and efficiently associated, and the retrieval efficiency and the management level are favorably improved.

The embodiment of the invention provides a data access method and device based on a last-stage cache, the method is applied to a distributed memory system, the distributed memory system comprises a plurality of cores, a memory subsystem of the distributed memory system is provided with a three-stage cache structure, and the three-stage cache structure comprises a first-stage cache, a second-stage cache and the last-stage cache. A label matrix corresponding to the last-stage cache at least comprises an exclusive core flag bit corresponding to a cache line and other core flag bits, and the method comprises the following steps: responding to a data access request of a target core; determining a target cache line corresponding to the data access request from the distributed memory system based on a target data access rule corresponding to the core type of the target core; and sending a response message to the target core according to the core occupation state corresponding to the target cache line. According to the data access method and device, the problem that when a distributed memory system is used for processing data access requests of different cores, data cannot be accurately accessed is solved, and then the effect of improving the data access accuracy is achieved.

The invention discloses a cache processing method of heterogeneous equipment, a heterogeneous system, a product, a device and a medium, and relates to the technical field of data storage. And according to the target virtual address access times corresponding to each device in the heterogeneous system, realizing a global data caching mechanism. And determining a file page corresponding to the target heterogeneous device based on the target virtual address access frequency, so that the file system layer has the capability of managing the memory cache of the heterogeneous device. And the predicted file page of the target heterogeneous device is verified, and the file page is moved under the condition that the verification is passed, so that the distribution accuracy of the target heterogeneous device is improved. The file page of the target heterogeneous device is mapped and moved to the memory of the host side, the target heterogeneous device does not need to be accessed, the file page stored in the memory of the processor is directly accessed, the increased access rate of the processor accessing the target heterogeneous device is avoided, the delay of accessing the memory of the target heterogeneous device is reduced, and the access efficiency is improved. And the data transmission rate and the system efficiency of the heterogeneous system are improved.

The invention discloses a DWA and APF path planning method based on an ocean digital vehicle, and is applied to the technical field of path planning. Comprising the following steps: initializing an ocean digital carrier, constructing a grid map and setting a target point and a starting point, planning a path by adopting an improved DWA algorithm, updating a current position to be a newly generated trajectory end point, determining a current local optimal plan by utilizing the DWA algorithm, and judging whether the current position is close to a final target point or not. And if yes, ending the algorithm and performing final path planning by using the APF, and if not, repeatedly adopting the improved DWA algorithm to plan a local path. According to the method, the improved DWA algorithm and the improved APF algorithm are fused, random unknown obstacles can be effectively avoided, and the algorithm is prevented from falling into a local minimum value in the final stage.

The invention relates to an intelligent integrated circuit and an intelligent chip in an artificial intelligence hardware platform. The invention relates to the field of deep neural networks in an artificial intelligence general technology, in particular to a data deployment system, a training method, an inference method and a chip of a deep neural network. The deep neural network data deployment system comprises a storage module; the matrix decomposition module is used for decomposing the original matrix into the sum of N sub-matrixes, and N is greater than or equal to 2; wherein one of the sub-matrices is further decomposed into a product of two or more low-rank matrices; the matrix expression processing module is used for expressing the sub-matrixes which are not decomposed at first precision and then storing the sub-matrixes; expressing the two low-rank matrixes with second precision and then storing the two low-rank matrixes; wherein the original matrix is a weight matrix and / or an activation value matrix, and the first precision is lower than the second precision. According to the method, on the premise that the performance of the model is basically not lost, the storage space required in the training and reasoning process is effectively reduced, and the calculation efficiency is improved.

The invention provides an abnormal case detection method and device, and the method comprises the steps: storing one piece of personalized standard data and a plurality of abnormal cases in a smart card, and setting one abnormal feature and one piece of expected error information in one abnormal case; the method comprises the following steps: an intelligent card receives a transaction command sent by a cash register, and selects a target abnormal case to execute a transaction operation; the smart card obtains personalized standard data, organizes transaction data according to the target abnormal characteristics of the target abnormal case and the personalized standard data, and sends expected error information corresponding to the transaction data and the abnormal case data to the cash register; and after receiving the transaction data, the cash register detects the transaction data and judges whether the transaction data has errors, and if so, the cash register writes expected error information into an error detection log. According to the method, the personalized card generation time of each test case is shortened, and the card test interruption time is shortened by checking error detection logs in batches, so that the test time is shortened.

The invention belongs to the technical field of quantum circuit simulation, and relates to a quantum circuit simulation method and system. The method comprises the following steps: reading information of a quantum circuit, an input state and an output state, and converting the information into a tensor network; tensors in the tensor network are converted into a tensor decision diagram; shrinking the tensor decision diagram by using a shrinkage algorithm; and according to the tensor decision diagram after shrinkage is completed, single amplitude and full amplitude are read or sampling is carried out. Before the contraction is carried out, a tensor network optimization tool can be used for searching a better contraction sequence, and then the tensor decision diagram is contracted according to the contraction sequence. According to the invention, the memory space for simulation can be exponentially reduced on part of circuits, so that the scale of the circuit which can be simulated is greatly increased. Compared with simulation methods corresponding to other decision diagrams, the simulation efficiency can be improved on part of circuits, and the time required for simulation is shortened.

The invention discloses a screen sharing method and device, and the method comprises the steps: collecting a video stream presented in a screen of a mobile terminal through an extension program, so as to obtain original video stream data; the extension program performs YUV decoding on original image information included in a video frame in the original video stream data to obtain intermediate data; the extension program splits the intermediate data to obtain frame header data, basic information, a Y data block, a U data block and a V data block, and converts the frame header data, the basic information, the Y data block, the U data block and the V data block into binary data; the expansion program sends the binarized frame header data, the binarized basic information, the binarized Y data block, the binarized U data block and the binarized V data block to the main application; the main application determines that the transmission of the video frame is ended based on the received binarized frame header data; and the main application splices the received binarized Y data block, the binarized U data block and the binarized V data block to obtain a complete video frame. The memory space occupied by the mobile terminal by screen sharing can be reduced.

The invention discloses a genetic programming symbol regression method based on multi-population asynchronization and parallelism, which comprises the following steps of: constructing a sub-population of a symbol model based on a genetic algorithm, and taking the symbol model as an individual of the sub-population; applying for a shared memory for an elite group of the symbol model, and applying for a message queue used for collecting elite individuals in a sub-group of each symbol model; iteratively executing selection operation, crossover operation, mutation operation and fitness evaluation on each sub-population to optimize the sub-population; an independent thread is applied for executing a genetic programming algorithm for each sub-population, and a symbol model set is searched for the training data set of each sub-population; when each sub-population meets a sending condition, the sub-population sends an elite individual set in the population at the moment to a message queue; and after the termination condition of the genetic programming algorithm is met, performing primary sorting by adopting the root-mean-square error indexes of the elite individuals in the elite group, and taking the elite individual with the minimum root-mean-square error as a final output symbol model.

The invention provides a casting model mesh simplification method based on a Unity 3D platform, which reduces the rendering and calculation burden by simplifying the mesh structure of a casting model and reducing the number of vertexes and triangular surfaces of the model, solves the problem of higher rendering pressure caused by high complexity of the casting 3D model, and improves the rendering efficiency. And the accuracy of the 3D model is ensured while the simplification is realized.

This application relates to a data storage method, apparatus, vehicle, electronic device, and storage medium. The method includes: determining the computation order of N operators included in a neural network; determining M operators from the N operators based on the computation order; calculating the i-th operator among the N operators according to the computation order to obtain the output of the i-th operator; if the i-th operator is one of the M operators, then saving the output of the i-th operator to a first type of storage space, wherein the output of the i-th operator saved in the first type of storage space cannot be deleted after calculating the (i+1)-th operator but before calculating the k-th operator; if the i-th operator is not one of the M operators, then saving the output of the i-th operator to a second type of storage space, wherein the output of the i-th operator saved in the second type of storage space can be deleted after calculating the (i+1)-th operator. This method can solve the problem of excessive memory consumption during the computation of a neural network model.

The embodiment of the invention provides a terminal information management method and device, equipment and a medium, and relates to the technical field of computers, the method comprises the following steps: receiving and analyzing a scheduling request of a physical layer interface of a terminal, and extracting a radio network temporary identifier RNTI field value in the scheduling request; determining a physical identifier of the terminal from a radio network temporary identifier (RNTI) buffer area according to the RNTI field value; after decoding of the received signal is finished, updating a state value of the decoding state buffer area according to a decoding result, and judging whether a recovery condition of a storage space in the RNTI buffer area is met or not; the decoding state buffer area is pre-established and obtained after initialization processing; and when the recovery condition is met, obtaining the target storage position of the RNTI buffer area and carrying out recovery processing. According to the scheme, the physical identifier can be accurately determined in the RNTI buffer area, the storage space corresponding to the RNTI buffer area is released in time when the state value meets the recovery condition, memory waste is avoided, and the terminal information processing efficiency is improved.

The invention discloses a method and system for processing data in a model and electronic equipment, and relates to the field of model technologies and machine learning. The method comprises the steps that parameter information of a model is read from an off-chip storage space, the parameter information is stored in a blocking mode in an unstructured sparse matrix format, and the model is obtained based on machine learning training; the parameter information of the model is decompressed and stored in a corresponding shared memory, and the decompressed parameter information is stored in a structured dense matrix format; based on the decompressed parameter information, output data matched with the input data is determined, and the output data is used for representing a dialogue result generated when the model processes the input data. The technical problem that the model parameter storage consumes much space is solved.

This invention relates to the field of multimedia video processing and playback control technology, specifically a video segment-level content management and control method and system, comprising: a time-series discrete mapping step: obtaining a server-side rule set and mapping it to a time-series state index structure with video durations divided into discrete-time graticles; an asynchronous look-ahead detection step: calculating the look-ahead time based on the current playback time using an independent asynchronous thread and reading the corresponding graticle data from the bitmap; an instruction generation and caching step: generating instruction objects based on graticle control identifiers using preprocessing logic and pushing them sequentially into an instruction circular queue; and a frame-level synchronous application step: checking the queue during main playback thread rendering, and applying the instruction if the current time matches the instruction trigger time. This invention reduces playback CPU usage, eliminates micro-stuttering caused by main thread blocking, and ensures accurate triggering of control logic at the frame level.

This application relates to the field of communication technology and provides a number processing method, apparatus, device, and storage medium. The method includes: acquiring user number information to be stored that has the function of receiving smart SMS messages, wherein the user number information to be stored includes first number segment information to be stored and second number segment information to be stored, wherein the first number segment information to be stored is general information for each user number to be stored, and the second number segment information to be stored is specific information for each user number to be stored; deleting the first number segment information to be stored; and storing the user number information to be stored sequentially in a preset array according to the numerical order of the second number segment information to be stored. This solution saves memory space and improves query speed in subsequent queries by customizing and storing the user numbers to be stored.

The present application provides an information query method, device and electronic device, including: analyzing and determining the first bit vector information corresponding to the first data string to be queried in the preset bit vector, when the first bit vector information conflicts with the bit vector information of the second data string, transforming the first data string into a third data string with a greater difference from the second data string, analyzing and determining the second bit vector information corresponding to the third data string in the preset bit vector, and determining that the first stored data string is queried based on the conflict between the bit vector information of the non-data string and the second bit vector information of the third data string. In this scheme, the data string is stored using a bit vector, and only the bit vector information with a small amount of data needs to be stored, saving memory space. Moreover, the changed data string is used to expand the difference with other data strings, and bidirectional storage is performed based on the bit vectors corresponding to the data string itself and the changed data string, respectively, so that bidirectional judgment is realized during the query, reducing the misjudgment rate.

This application relates to a target data localization method, apparatus, and storage medium. The method includes: traversing a vector image matrix row by row based on matrix row indexes, determining the column index of the first vector data in each row based on preset conditions, and storing them sequentially; traversing the vector image matrix column by column based on matrix column indexes, determining the row index of the second vector data in each column based on the preset conditions, and storing them sequentially; determining the row and column coordinates of the target data based on each column index and its corresponding storage order, and each row index and its corresponding storage order, wherein the target data is determined based on the first vector data and the second vector data, thereby achieving target data localization and solving the problem of high memory resource consumption in target data point localization methods in related technologies.

The embodiment of the present application provides an audio data receiving and sending method, an audio receiving device and an audio sending device, and relates to the field of Bluetooth technology. The method includes: establishing a connection with each audio sending device in the audio sending device set, the audio sending device set includes n audio sending devices, and n is a positive integer; sending a synchronization link establishment request to each audio sending device, the synchronization link includes n synchronization sub-links, and the synchronization sub-links correspond to the audio sending device one by one; establishing a synchronization link; sending a first data packet to the target audio sending device corresponding to the target synchronization sub-link in the audio sending device set through the target synchronization sub-link; receiving a second data packet from the target audio sending device, and the second data packet is sent by the target audio sending device through the target synchronization sub-link. The embodiment of the present application can reuse the protocol stack, saving the memory space of the audio receiving device.

The invention provides an NTRU homomorphic outer product-based privacy information retrieval method, which comprises the following steps of: firstly, regarding a database as a v + 1-dimensional hypercube, generating a corresponding query tuple (i, j1, j2,..., jv) by a user according to a query index idx, carrying out compression coding and encryption on the query tuple to obtain an RLWE ciphertext cquery serving as a query, sending the ciphertext cquery to a server, and sending the RLWE ciphertext cquery to the server; the server expands the cquery into v + 1 groups of ciphertexts CTNtru, CTNtruGSW, 0, CTNtruGSW, 1,..., CTNtruGSW and v-1 based on a ciphertext inquiry expansion algorithm, NTRU homomorphic outer product calculation is adopted according to a hypercube form of a database, a ciphertext response result credit is generated and sent to the user, and the user recovers a result didx by adopting an NTRU decryption algorithm. According to the scheme, the hidden query of the database data records is realized, the complexity of a ciphertext query vector extension algorithm is reduced, and the throughput and query calculation efficiency of privacy information retrieval are improved.

The invention discloses a method for optimizing the use of RTPS protocol resources in DDS communication. When a PUB end / SUB end receives a data(p) message from a peer end, a builtin-proxy entity is created locally, and a builtin-proxy entity state table is created at the same time. For example, when the SUB end receives and processes the notification information of all writers from the PUB end, the SUB end sends an acknack message to the PUB end. At this time, the SUB end has completed the interaction of link building, and the memory occupied by the builtin-proxy publication writer entity can be released. At the same time, after the PUB end receives the acknack message from the SUB end, the SUB end has received and processed all the writer information of the PUB end. At this time, the PUB end has completed the interaction of link building, and the space occupied by the builtin-proxy publication reader entity can be released. The invention reduces the occupied memory space by releasing the memory space of the builtin-proxy entity, and the space saved in different scenarios is different.

The invention relates to the technical field of electronic equipment, and particularly provides a display control method and device, a display module and electronic equipment. The display control method comprises the steps of obtaining a target basic graph included in a to-be-displayed image and display parameters of the target basic graph, determining a target display function from display functions corresponding to pre-configured basic graphs, configuring the target display function based on the display parameters of the target basic graph, generating display data of the to-be-displayed image, and displaying the to-be-displayed image according to the display data. And controlling the display module to display the to-be-displayed image according to the display data. In the embodiment of the invention, the display image is combined through the pre-configured basic graph, so that the display module does not need to store full display image data in advance, the memory space of the display module is saved, more display images can be obtained through free combination by utilizing the basic graph, and the flexibility and richness of the display image are improved.

The invention relates to the technical field of tractor attitude detection, in particular to a tractor attitude detection method based on forward-forward backtracking, and the method comprises the steps: firstly building a tractor attitude detection hardware system which comprises an inertial measurement unit, a Beidou satellite antenna and a microcontroller; the method comprises the following steps: constructing a strapdown inertial navigation backtracking alignment error model based on an inertial system by utilizing high-frequency acceleration and angular velocity information originally output by an inertial measurement unit, and establishing a state equation of a tractor combined attitude detection system; real-time position information of a tractor is obtained by using a Beidou satellite, a displacement vector of the tractor under an initial navigation system is solved through position difference and matrix transformation, a tractor combined attitude detection system measurement equation is constructed for filtering, a filtering result is fed back to correct an error of a strapdown inertial navigation system, and the error is calculated. Therefore, the real-time attitude of the tractor can be quickly and accurately solved. The problems that a traditional tractor attitude detection method is low in convergence speed, poor in dynamic performance and the like in a complex field operation environment are solved.

The application relates to the computer technical field, in particular to a sensitive word storage and retrieval method and system, electronic equipment and storage medium, the method comprises the following steps: storing the sensitive word to be stored to a target sharding service through a target mapping table, wherein the target mapping table comprises a mapping relationship between the target sharding service and the first character of the sensitive word to be stored, the target sharding service is one of a plurality of sharding services in memory, and the target sharding service is a sharding service polled currently; based on the target mapping table and the first character of the word to be retrieved, the sharding service to be retrieved corresponding to the first character of the word to be retrieved is determined; and based on the word to be retrieved and the data stored in the sharding service to be retrieved, the sensitive word retrieval is carried out; the method can preferably improve the storage efficiency and retrieval efficiency of the sensitive word, and the cost is low.

The present invention provides an LBA full cache method and system based on append-write distributed storage, the method comprising: decomposing a shared block storage addressing space into multiple blocks and hashing them into a storage cluster; defining a data structure in the full cache as a linear table; storing an index of a chunk in a chunk table in an item item, and addressing an LBA address using the content identified by the item item; when addressing occurs overwriting, the original location pointing to the chunk is an idle location, when a chunk data is overwritten by LBA or integrated and completely moved away, the chunk is identified as an idle location in the chunk_table linear table, and a new chunk_id is stored when a new chunk is written; and the chunks are integrated so that the total size of the chunk_table linear table does not exceed a threshold. The present invention enables distributed shared storage to cache the LBA addressing information of all shared storage devices while consuming less storage space.

The present application relates to the field of computer vision language and image recognition, and particularly relates to a semantic description and question-answering system establishment method and system for remote sensing disaster detection, which generates text language description capable of accurately responding to task requirements image under the premise of adding text constraints by using image features, and establishes a mapping relationship between the pixel-level segmentation result and the question-answering system, and extracts high-level semantic information, and establishes a question-answering system, which can accurately output answers to disaster inquiry questions. The present application can complete three sub-tasks of semantic segmentation, sentence description and question-answering system by using a set of network models, thereby saving computer memory space and improving response speed, and providing technical support for improving the automation, intelligent level and rapid response capability of the remote sensing disaster detection system.

The invention provides a data processing method and device based on a distributed database and electronic equipment, and can be applied to the technical field of database data flash back. The method comprises the following steps: a main execution node sends an obtained operation request to a sub-execution node, wherein the operation request executes an operation for a target table; the sub-execution node sends an operation execution result to the main execution node, and the operation execution result is obtained by operating the target table according to the operation request; under the condition that the operation execution result represents that execution is successful, the main execution node obtains an operation version identifier corresponding to the operation execution result and sends the operation version identifier to a sub-execution node; the sub-execution node generates a bitmap record file according to the operation version identifier and operation record information related to the target table; and according to the bitmap record file and the operation version identifier, carrying out flash-back query and flash-back recovery operation on the data in the target table.

The present invention provides a patent novelty evaluation method, device, equipment and readable storage medium. The patent novelty evaluation method includes: determining the corresponding fruit fly counting bloom filter based on the category of the patent to be evaluated; inputting the text of the patent to be evaluated into the language model to obtain a first patent vector; reconstructing the first patent vector based on the fruit fly neural network algorithm to obtain a second patent vector; and calculating the novelty score of the patent to be evaluated based on the second patent vector and the fruit fly counting bloom filter. The algorithm for calculating the patent novelty score of the present invention has a fast execution speed and high time efficiency, and adopts a fruit fly counting bloom filter structure to store patent historical data information, and occupies a small memory space and has high space efficiency. The patent novelty score result calculated on this basis is also relatively accurate.

The invention provides an automatic control system for airborne equipment production and a control method thereof. The system comprises an airborne comprehensive production system, a server module, a workstation module, a bus module and a terminal module, according to the automatic control system for airborne equipment production and the control method thereof, data integrity and safety can be guaranteed, data analysis efficiency is improved, software system maintenance cost and workstation computer hardware cost are saved, compatibility is high, transplanting, splitting and fusion are easy, unified installation and maintenance are easy, and system troubleshooting is easy. And meanwhile, the purposes of saving the cost and improving the working efficiency are also achieved.

The embodiments of the present disclosure disclose a method, device, electronic device and readable medium for pushing virtual items. A specific implementation of the method includes: filtering a user rating information set; clustering the filtered user rating information set to obtain the filtered user rating information set after clustering as a user rating information group set; determining a user rating information group; generating a user similarity set; determining the similarity of each user that meets the neighbor user filtering condition as a neighbor user similarity set; determining a neighbor user rating information set based on the neighbor user similarity set; determining a target user rating information set; filtering out each target user rating information as a user rating information set to be recommended; determining a virtual item information set to be recommended; pushing the virtual item information set to be recommended to a client to present virtual items to the user corresponding to the client. This implementation can reduce the waste of CPU processor resources.