82 results about "Bandwidth optimization" patented technology

The invention relates to the technical field of data transmission and updating, in particular to a data transmission and updating system and method of a cloud mobile phone. According to the technical scheme, the system comprises an intelligent network monitoring module, a dynamic bandwidth allocation module, a server load balancing module, an intelligent data synchronization module and a cache management module, and the intelligent network monitoring module monitors network conditions, including network bandwidth, network congestion degree and network stability, between a user terminal and a cloud server in real time. Through the synergistic effect of the intelligent network monitoring module, the dynamic bandwidth allocation module, the server load balancing module, the intelligent data synchronization module, the cache management module and the like, the network condition is accurately monitored, the bandwidth is reasonably allocated, the server load is optimized, the data synchronization and completeness are ensured, and the cache efficiency is improved; stable and reliable network connection, preferential key data transmission and rapid data synchronization and loading are brought to users, and the use experience of the cloud mobile phone and the overall performance of the system are greatly improved.

The invention provides an abnormal traffic management and control method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of traffic processing. The management and control method comprises the following steps: on the basis of traffic characteristic data of network traffic, identifying abnormal traffic, an abnormal type of the abnormal traffic and a service type of the network traffic, the network traffic comprising the abnormal traffic and normal traffic; predicting a self-adaptive bandwidth occupation demand of the abnormal traffic based on the traffic feature data of the network traffic, the service type of the network traffic and the abnormal type of the abnormal traffic; and adjusting the current bandwidth occupation of the abnormal traffic based on the self-adaptive bandwidth occupation demand of the abnormal traffic so as to control the abnormal traffic. The invention aims to at least solve the problems that in the prior art, high-volatility network requirements cannot be fully met, flexible bandwidth optimization capability is lacked, and the aspect of refining flow management is still insufficient. The method is suitable for abnormal traffic identification and control scenes.

The invention relates to the technical field of wireless communication, and relates to a multi-terminal data processing method, a multi-terminal data processing device, communication equipment and a computer readable storage medium. The method comprises the following steps: an on-satellite UPF receives to-be-processed data sent by a terminal through a wireless access network, the to-be-processed data carries a joint task identifier, and the terminal is any one of a plurality of terminals corresponding to the joint task identifier; the satellite UPF sends at least part of data in the to-be-processed data to a ground UPF for processing to obtain ground processing data, and stores other data of the to-be-processed data; and the on-satellite UPF performs cooperative processing on the other data corresponding to the joint task identifier in the storage to obtain on-satellite processing data, and sends the on-satellite processing data to the ground UPF, so that the ground UPF performs forwarding based on the ground processing data and the on-satellite processing data. The data transmission amount can be reduced, the communication bandwidth can be saved, the utilization of resources can be optimized, and the data processing efficiency and timeliness can be improved.

The invention relates to the technical field of wireless communication, and discloses a wireless communication operation and maintenance data analysis method and system, and the method comprises the following steps: in all sub-regions of a wireless communication operation and maintenance region, carrying out the high-frequency similarity analysis of the business flow value data of each historical time period, and constructing the data analyzed in each time period to obtain a first set; performing distance analysis on the first set of each time period, and constructing a second set through a distance analysis result; and performing influence analysis on the sub-regions in the second set, and constructing a third set according to an influence analysis result. According to the method, the analysis sub-regions which influence each other and are key in the wireless communication network can be accurately identified, so that operation and maintenance personnel can focus on key regions, carry out key monitoring on the key analysis sub-regions, distribute network resources more reasonably, increase bandwidth in the key regions, optimize base station configuration and the like, and the resource utilization efficiency is improved.

Disclosed in the present invention is a URLLC multi-connectivity transmission and resource allocation method for industrial automation. The method comprises the following steps: S1, establishing a multi-connectivity technology-based URLLC system model; S2, setting constraint conditions of end-to-end transmission on the basis of quality of service requirements for an industrial automation application; S3, reporting the length of an information byte required for transmission, and determining a bandwidth optimization problem of multi-connectivity data transmission; and S4, executing a resource allocation algorithm to calculate an optimal cross-layer resource allocation solution for the multi-connectivity data transmission. According to the present invention, a URLLC multi-connectivity transmission and resource allocation method for industrial automation is established, to avoid the problem of cellular link failure caused by dynamic changes of channels in a traditional transmission mode; and under the constraints of end-to-end delay, reliability and maximum code length, an optimal cross-layer resource allocation method is provided, to minimize a bandwidth required for multi-connectivity transmission, thereby achieving the objective of saving spectrum resources.

The invention discloses a silicon light modulator which comprises a silicon substrate, an insulating oxide layer, a top-layer silicon waveguide modulation region and a top-layer covering oxide layer which are stacked in sequence, and the top-layer silicon waveguide modulation region sequentially comprises an n-type heavily doped region, an n-type main doped region, a p-type main doped region and a p-type heavily doped region in the first direction. The invention further discloses a compensation doping method of the silicon light modulator. The silicon light modulator has the advantages that compensation doping is carried out on the flat plate region in an oblique ion implantation mode, and the effects of improving the bandwidth and the modulation efficiency of the silicon light modulator can be achieved at the same time. And compared with a single bandwidth optimization scheme and a single efficiency optimization scheme of the silicon optical modulator, the extra insertion loss introduced by the method is not obviously increased.

The invention relates to the technical field of unmanned aerial vehicle flight data scheduling, and discloses an unmanned aerial vehicle flight data scheduling method based on a distributed storage system, and the method comprises the following steps: S1, an unmanned aerial vehicle node collects flight data and stores the flight data in a local memory; s2, realizing consistency and synchronous management of the data through a block chain network, and verifying the validity of the uploaded data among a plurality of nodes; and S3, realizing the consistency and synchronous management of the data through the block chain network, and verifying the validity of the uploaded data among the plurality of nodes. By introducing a block chain technology, an intelligent contract and a bandwidth optimization algorithm, the problems of consistency, redundant copy management and data transmission path optimization of the unmanned aerial vehicle flight data in a distributed storage system are effectively solved, the integrity, security and high availability of the data are ensured, the bandwidth utilization rate and transmission efficiency are improved, and the system performance is improved. Therefore, the reliability and the intelligent scheduling capability of the system are enhanced.

The invention relates to the technical field of radiotherapy data analysis, provides a tumor radiotherapy abnormity early warning system based on a clustering algorithm, and aims to improve the abnormity detection capability and early warning precision of patient health data in a radiotherapy process. According to the system, by introducing a local outlier factor and an adaptive bandwidth optimization technology, density weighted calculation of a Mean Shift clustering algorithm is optimized, an internal mode in tumor patient data can be accurately identified, and the accuracy of preliminary clustering is improved; in the enhanced clustering process, an entropy regularization optimization P3D cutting method is adopted, strategy network updating of a PPO enhanced learning algorithm is further optimized, clustering features are dynamically adjusted, and the adaptability and accuracy of the system in a dynamic environment are enhanced; according to the system, reinforcement learning and clustering analysis technologies are combined, high-precision and high-real-time support is provided for abnormal early warning in the radiotherapy process, and the intelligent level and reliability of patient monitoring are remarkably improved.

The invention discloses a data transmission optimization method, system and device based on optical fiber communication and a medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining an optical fiber communication line and a monitoring time slot, and obtaining the number n for monitoring according to the optical fiber communication line; acquiring unit time slot average flows of the first one, two,..., n monitoring time slots at the current time point in real time, acquiring a time sequence weight corresponding to each unit time slot average flow based on a time sequence model, and acquiring a fluctuation index based on a fluctuation model, the n unit time slot average flows and the time sequence weight corresponding to each unit time slot average flow; judging whether the fluctuation index exceeds a preset fluctuation threshold value, if so, obtaining a bandwidth adjustment increment of the optical fiber communication line according to the fluctuation index, and obtaining an adjustment cost index according to the bandwidth adjustment increment; and completing bandwidth optimization according to the adjustment cost index and the bandwidth adjustment increment. The method has the advantages of fast dynamic response, accurate transmission bandwidth matching and collaborative optimization.

A system (100) for optimizing data caching in content delivery networks (CDNs), comprising: a user behavior analysis module configured to capture and analyze real-time and historical user access patterns using artificial intelligence (AI) and machine learning (ML) algorithms to predict content demand; a predictive caching module operatively coupled to the user behavior analysis module and configured to cache frequently accessed content on edge servers based on predicted demand patterns; an adaptive cache replacement module configured to dynamically manage cache deletion and retention by taking into account content popularity, request frequency, recency, and network congestion; a multi-level hierarchical caching engine configured to organize content storage across edge servers, regional caches, and origin servers for optimized retrieval efficiency; a decentralized load balancer configured to distribute caching responsibilities among CDN nodes using blockchain-based or distributed ledger technology to prevent network congestion and ensure fault tolerance; a latency optimization module configured to determine the optimal placement of content based on user location, device type, and network speed using deep learning algorithms; a bandwidth optimization module configured to reduce redundant data transfers by using intelligent prefetching, compression, and deduplication techniques; a real-time monitoring and self-learning module configured to track cache performance and dynamically adjust caching strategies based on evolving network conditions and user behavior; a security and data integrity module configured to provide encrypted data caching, access authentication, and integrity checking across CDN nodes.

The invention provides a PCIe link energy-saving regulation and control method, device and equipment and a storage medium, and relates to the technical field of computers, and the method comprises the steps: collecting target parameters such as a physical layer signal quality parameter, a transmission layer flow parameter and an equipment state parameter; predicting the bandwidth demand in the future time period based on the target parameter; generating a link configuration instruction in combination with a quality of service (QoS) priority strategy and a thermal constraint condition; and finally, the width, the rate and the error correction mechanism enabling state of the PCIe link are adjusted according to the instruction. According to the method, through dynamic bandwidth demand prediction, QoS priority and thermal constraint integration and link parameter adaptive adjustment, 5ms-level burst load response, multi-device global bandwidth optimization and cooperative improvement of energy efficiency and link utilization rate in a high-noise / heat dissipation limited scene are realized.

The invention discloses an edge federated learning communication method combining dynamic sampling and bandwidth optimization, which relates to the technical field of edge computing, and comprises the following steps of: collecting the sample size, training time and network bandwidth of terminal equipment through a monitoring tool, classifying and extracting bandwidth labels after preprocessing, and optimizing by adopting dynamic sampling and asynchronous communication to obtain an edge federated learning result. Gradient compression based on bandwidth perception is combined, important gradients are preferentially transmitted, convergence speed is evaluated, resource collaboration is optimized, an MEC server asynchronously distributes parameters, a queue is managed and updated, and a model is aggregated. By combining dynamic sampling and bandwidth optimization, the edge federated learning communication efficiency is remarkably improved, the transmission cost is reduced, the sampling strategy is dynamically adjusted, the important gradient is preferentially transmitted, the model convergence is accelerated, and the equipment resource collaboration and asynchronous parameter distribution and queue management of the MEC server are optimized; the timeliness and accuracy of model updating are ensured, and the practicability and flexibility of edge learning are enhanced.

The invention discloses an MODBUS data transmission bandwidth optimization method, which comprises the following steps: acquiring basic information of each device from pre-stored device data item configuration information, and generating a data item feature matrix; calculating a data item priority vector based on the historical access frequency and the real-time requirement; constructing a data item incidence matrix through the equipment difference, the address overlapping degree and the priority difference; establishing a multi-dimensional constraint model of bandwidth consumption, real-time performance and reliability; data item grouping optimization is carried out on the basis of an improved Kernighan-Lin algorithm; and dynamic adjustment of a transmission strategy is realized through real-time performance monitoring and evaluation. According to the invention, the data transmission efficiency is improved, the real-time performance of important data is ensured, and the system reliability is improved.

The invention discloses a vehicle network bandwidth management method, device and equipment, a storage medium and a program, and is applied to the technical field of vehicle communication, and the method comprises the steps: collecting the network state information of a target vehicle; and updating the priority of the data flow in the target vehicle according to a preset bandwidth optimization strategy and the network state information so as to adjust the network bandwidth occupied by the data flow. According to the embodiment of the invention, the vehicle network bandwidth is dynamically adjusted, the utilization rate of the network bandwidth is improved, the stability of the vehicle network can be improved, and the vehicle network use experience of a user can be improved.

The invention provides chip-to-chip bandwidth optimization. A transceiver for transmitting and receiving data packets over a communication channel is provided. The transceiver receives a first request packet including a plurality of information fields having a first type of operation to be performed on the memory device and a first address. The transceiver stores the first type of operation and the first address in a memory associated with the transceiver and sends a first request packet having the first address to a target device. The transceiver then receives a second request packet, the second request packet including a second address in the memory device, and determines that the second request packet is part of a sequence of request packets to the target device based on the first type of operation, the first address, and the second address. The transceiver then deletes a portion of the second address in a header of the second request packet to form a third request packet and sends the third request packet to the target device.

The invention discloses an intelligent security and protection monitoring system and method based on computer vision, and belongs to the technical field of data processing.The method comprises the steps that firstly, a system performance state model, an efficiency state model and a value state model are constructed, and a system performance state coefficient, an efficiency state coefficient and a value state coefficient are output respectively; and the comprehensive operation state of the system is quantitatively evaluated. Furthermore, based on the effective alarm rate and the alarm response time under the performance and efficiency state coefficient, a quality-speed adaptation model is constructed, and the adaptation degree reflecting the balance degree of the system between the alarm quality and the response speed is output. And finally, combining the adaptation degree, the value state coefficient and the basic bandwidth occupancy peak value to construct a bandwidth optimization model, and dynamically outputting a target bandwidth occupancy peak value. According to the method, comprehensive evaluation of the multi-dimensional state of the security and protection system and self-adaptive optimal configuration of key resources are realized, and the intelligent level and the overall operation efficiency of the system are effectively improved.

The invention discloses a bandwidth optimization transmission method based on DICOM image information, and relates to the technical field of medical image information processing, and the method comprises the steps: employing a deep learning model to automatically recognize a key diagnosis information region and a non-key part in a DICOM image; according to an identification result, performing adaptive compression processing on the DICOM file to form an adjusted DICOM file; the multi-dimensional data characteristics in the DICOM file are deeply analyzed, and the multi-dimensional data compression algorithm is adopted to deeply optimize the characteristics of each dimension; the optimized DICOM file is uploaded to a cloud end for distributed storage; based on files stored in a cloud distributed mode, key DICOM file partitions requested by a user are rapidly positioned through an intelligent retrieval framework and a Bloom filter, and the transmission process is optimized. According to the method, the key region of the DICOM image is automatically identified through deep learning, adaptive compression is executed, the transmission efficiency is remarkably improved, key information is ensured to be complete, and the medical service quality is improved.

The present invention relates to the field of wireless communication technologies, and discloses a method and system for analyzing wireless communication operation and maintenance data, including the following steps: in all sub-regions of a wireless communication operation and maintenance area, perform high-frequency similarity analysis on the service traffic value data of each historical time period, and construct a first set from the data analyzed for each time period; perform distance analysis on the first set of each time period, and construct a second set based on the distance analysis results; perform influence analysis on the sub-regions in the second set, and construct a third set based on the influence analysis results. The present invention can accurately identify the mutually influential and key analysis sub-regions in a wireless communication network, so that operation and maintenance personnel can focus on key areas, conduct key monitoring on these key analysis sub-regions, allocate network resources more reasonably, increase bandwidth, optimize base station configuration, etc. in key areas, and improve resource utilization efficiency.

This invention discloses an optimization method for an online conferencing system, relating to the field of online conferencing technology. The method includes: optimization of image processing and transmission: optimizing image encoding and decoding algorithms; bandwidth optimization: automatically adjusting the image transmission bitrate using adaptive bitrate technology based on the user's device and network conditions; optimization of user interaction: adding functions such as liking, sharing, real-time voice calls, and real-time video calls; and optimization of device compatibility: automatically adjusting the image display resolution using a responsive design adapted to different devices and screen resolutions based on the user's device and network conditions. This invention's optimization method for the online conferencing system can effectively reduce image processing and transmission latency and improve image transmission quality, while reducing bandwidth requirements, enriching user interaction functions, and improving device compatibility.

This invention relates to the field of near-memory computing technology, and discloses a method and system for near-memory computing bandwidth optimization based on HBM. The method includes: decoding and parsing instructions for access tasks on the host side to obtain the storage address and the required operation type; constructing metadata for the access tasks to obtain a scheduling context, and pre-selecting target processing elements in high-bandwidth memory; performing data storage access on the high-bandwidth memory to obtain data blocks to be processed, and performing information dimensionality reduction mapping on the data blocks to obtain partial result data; performing multi-path fusion and reduction on the partial result data to obtain final result data, and returning the final result data to the host side; while performing multi-path fusion and reduction, pre-fetching scheduling is performed on the access tasks to be processed on the host side to establish a scheduling context for the access tasks to be processed; this invention can improve the efficiency of near-memory computing bandwidth optimization based on HBM.

The invention discloses a switch intelligent scheduling management system, method and device and a storage medium, and relates to the technical field of network traffic scheduling, and the switch intelligent scheduling management system comprises an intelligent traffic management module and a distributed processing module; the intelligent traffic management module determines at least one transmission path of the data traffic in the network based on the initial switch node and the target switch node of the data traffic; determining a dynamic path weight of a transmission path based on switch node load information transmitted by the distributed processing module; and determining a target path based on the dynamic path weight of the transmission path. The dynamic path weight of the transmission path is determined based on the load of the switch node, so that intelligent path distribution of data traffic is realized, the burden of a certain path is reduced, the overall bandwidth utilization rate of the system is effectively optimized, overload of a single transmission path is avoided, and load balancing and bandwidth optimization of the system are realized.

A control panel delivery intelligent quality inspection and tracing system solves the problems that in the prior art, an automatic detection process is prone to being interrupted, data is dispersed and difficult to associate and trace, and dynamic risk early warning is lacked, and comprises an intelligent quality inspection subsystem and a delivery tracing subsystem which cooperate with each other. The quality inspection subsystem ensures safe and accurate detection through magnetic coupling isolation and golden finger connection, guarantees no loss of network disconnection data and network bandwidth optimization by using SQLite caching and a timed uploading mechanism, improves the maintenance and remodeling efficiency by adopting a physical separation detachable needle bed, and sets a fault simulation unit to comprehensively verify the reliability of a control panel. The traceability subsystem is in butt joint with cloud data through a Flask server, constructs an associated database to realize a full-process data link, supports macroscopic to microscopic data analysis by using a visual engine, performs anomaly identification and risk prediction in combination with machine learning, and generates an early warning report; according to the system, the delivery quality control level, the production efficiency and the risk prevention capability of the control panel are improved.

The invention relates to an aviation internet autonomous controllable platform flow analysis and management system. The method comprises a data acquisition unit, a data processing unit, a flow early warning traceability unit and an air-ground bandwidth optimization unit. Presetting aviation fleet satellite network acquisition nodes, and obtaining network flow data; five-dimensional standardized data sets of equipment, flights, applications, scenes and traffic are generated through processing, and the data sets are graded through an intelligent classification mechanism and stored according to grades; constructing an equipment dynamic digital file, generating an equipment flow diagram with a time axis, extracting total flow data of target equipment through multi-condition combination screening by combining flow diagram time sequence distribution characteristics and stage flow statistical data and utilizing a dynamic flow fluctuation threshold value and a grading early warning mechanism, and obtaining an evaluation report; and combining an air space application flow fingerprint spectrum technology and the five-dimensional standardized data set to identify flow air space application, and generating a bandwidth optimization strategy. Therefore, flow analysis and management of the aviation internet are realized.

The invention discloses a multi-channel high-speed data routing method based on an FPGA (Field Programmable Gate Array), and the method comprises the steps: monitoring the operation state of each data channel in FPGA hardware equipment in real time, generating a fluctuation coefficient Wf, carrying out the evaluation, triggering risk early warning, and effectively improving the reliability of data transmission; extraction, filtering and frame format verification operations are executed through a real-time data extraction mechanism and a DSP module, and data integrity is ensured. Calculating and evaluating the bandwidth utilization efficiency Bwe in real time to trigger a bandwidth optimization alarm, and dynamically adjusting a bandwidth allocation strategy based on the actual bandwidth usage amount Bac and the available bandwidth amount Bav by using a bandwidth allocation model to avoid resource waste or channel overload; a topology adjustment mechanism is triggered by calculating and evaluating a topology adaptability index Tad, and path selection is optimized in combination with a dynamic routing algorithm; the multi-channel optimization parameter Copt is utilized to dynamically adjust bottom layer extraction monitoring and preprocessing, efficient utilization and adaptability improvement of system resources are ensured, and finally the overall transmission efficiency and stability are ensured through routing performance evaluation.

The invention relates to the technical field of network communication equipment, in particular to a distributed network signal convergence concentrator, which comprises a main control module, and a power supply module supplies power to each module; the distributed signal acquisition unit is used for acquiring multi-path distributed network signals and transmitting the multi-path distributed network signals to the signal convergence processing module; the signal convergence processing module is used for carrying out integration, protocol conversion and bandwidth optimization processing on multiple paths of network signals; the network interface module is used for outputting processed convergence signals to superior network equipment; one side of the main control module is also provided with an extensible interface module; for some specific areas, the main control module is inconvenient to move or a connecting line is short, the mobile interface is arranged, so that butt joint with an external interface is facilitated, the whole equipment does not need to be moved, and a complex scene is adapted; and multiple types of wired and wireless signals are compatible, multi-protocol unified convergence is realized through protocol conversion, and the equipment compatibility is improved.

The invention relates to the technical field of wireless communication, and discloses a double sampling and bandwidth optimization data processing method for a low-rate FSK (Frequency Shift Keying) signal, which comprises the following steps: receiving and processing an FSK radio frequency analog signal through a radio frequency receiving front-end module, setting a multiple symbol rate through a double sampling demodulation module, receiving and amplifying the FSK radio frequency analog signal through the multiple symbol rate, and outputting the amplified FSK radio frequency analog signal through a bandwidth optimization demodulation module. And the digital signal processing module processes the double sampling demodulation data by using an algorithm to restore the initial data. According to the invention, a high-density time domain sequence is generated by adopting a double sampling rate, so that the bandwidth of a digital filter of a receiver is increased, normal communication can still be maintained when frequency deviation exists between a transmitter and the receiver, and even if the FSK receiver has a high error rate problem caused by the frequency deviation problem at a low rate, the communication efficiency is improved. And low-rate and long-distance communication can still be realized.

The present invention discloses a silicon optical modulator, comprising a silicon substrate, an insulating oxide layer, a top silicon waveguide modulation region, and a top covering oxide layer stacked in sequence, wherein the top silicon waveguide modulation region comprises an n-type heavily doped region, an n-type main doped region, a p-type main doped region, and a p-type heavily doped region in sequence along a first direction. The present invention also discloses a compensation doping method for a silicon optical modulator. Advantageously, by compensating the doping of the flat plate region by oblique ion implantation, the effects of improving the bandwidth and modulation efficiency of the silicon optical modulator can be achieved simultaneously. Moreover, compared with the single bandwidth optimization scheme and the single efficiency optimization scheme of this type of silicon optical modulator, the additional insertion loss introduced by this method does not increase significantly.

The invention discloses an instantaneous frequency identification method based on short-time fractional Fourier multi-transform, and the method consists of four parts: time-varying component separation, short-time fractional Fourier transform and matching multiple synchronous extrusion transform, and specifically comprises the steps: firstly, carrying out the time-varying component separation through a binary image and a connected region mark; secondly, obtaining an estimated instantaneous frequency position through bandwidth optimization short-time fractional order Fourier transform, secondly, rearranging a time-frequency coefficient by matching a synchronous extrusion operator, and finally extracting an instantaneous frequency curve by using a local modulus maximum method. The method can effectively identify the instantaneous frequency of the time-varying structure.

This invention relates to the field of software update technology, specifically to a method and system for overall online upgrade of smart door locks, comprising the following steps: acquiring firmware and chip encoding to construct an upgrade package and digest value; allocating transmission bandwidth blocks by comparing task increment differences; sending the upgrade package; calculating attenuation rate parameters based on time offset rate and write latency difference; adjusting the packet sending interval and the upper limit of retransmission times; decrypting and verifying the firmware; distributing decrypted data through an interface; writing to a non-running partition to verify integrity; and generating an upgrade audit log. In this invention, bandwidth allocation based on task increment differences optimizes scheduling; attenuation rate is calculated using offset rate and latency difference; packet sending interval is flexibly adjusted to adapt to real-time networks; data security is ensured by combining algorithmic decryption and digest value verification; precise distribution is achieved based on encoding identification of the target interface; firmware is written to a non-running partition to achieve seamless upgrades; and data aggregation generates logs to improve collaborative update efficiency.

The invention discloses an optimization method and system for improving the bandwidth of a neural network accelerator, and relates to the technical field of neural network accelerators, and the optimization method for improving the bandwidth of the neural network accelerator comprises the following steps: carrying out the convolution processing of input data through a convolution pipeline; performing activation function processing on the data subjected to convolution processing by adopting a single data processor; an on-chip buffer is adopted to replace a traditional on-chip memory for caching, and pooling processing and local response normalization processing are carried out on the data after activation function processing through a plane data processor and a channel data processor; and performing data format conversion on the data subjected to the local response normalization processing through a data recombination engine. According to the invention, by optimizing the storage read-write process of the data when the accelerator performs acceleration operation, the utilization rate of memory bandwidth is effectively improved, the time consumed by single acceleration operation is reduced, and the performance of the accelerator is improved.