17results about How to "Low latency requirement" patented technology

The invention discloses a speech recognition streaming feature extraction and batch reasoning method for a high-concurrency artificial intelligence real-time dialogue system. According to the method, a heterogeneous computing architecture in which a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit) cooperatively work is adopted, operations such as preprocessing, feature extraction and normalization of voice data are executed through the CPU, and batch feature reasoning is executed by utilizing the GPU; the method specifically comprises the following steps: after segmenting and preprocessing voice data at a CPU (Central Processing Unit) end, executing Mel filter bank feature extraction and cepstrum mean variance normalization to obtain a standardized acoustic feature vector; extracting feature vectors from the feature buffer area in batches at the GPU end, filling and aligning the feature vectors, and sending the feature vectors to a neural network model for parallel reasoning calculation; according to the method, streaming processing and batch reasoning in the speech recognition process are realized, and the concurrent processing capability and the real-time performance of the system are effectively improved.

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

The invention relates to the technical field of Bluetooth sound boxes, in particular to a Bluetooth sound box synchronous playing method and system. The method comprises the following steps: performing master device election on a Bluetooth loudspeaker box group to obtain a loudspeaker box group master device identifier; carrying out distributed time synchronization network construction on the Bluetooth sound box group based on the sound box group main equipment identifier to obtain a sound box group time synchronization network; performing environment signal scanning on the Bluetooth sound box group to obtain environment frequency band signal data; performing dynamic frequency band allocation on the Bluetooth loudspeaker box group based on the environment frequency band signal data to obtain a loudspeaker box group frequency band allocation scheme; performing network performance parameter monitoring on the Bluetooth sound box group to obtain a sound box group network performance parameter set; and according to the sound box group network performance parameter set, performing audio buffer area determination on the Bluetooth sound box group to obtain an audio buffer area distribution scheme. According to the invention, the accuracy and stability of audio synchronization are improved, the user experience is optimized, and the application scene of the Bluetooth sound box is widened.

The present disclosure relates to a text rewriting method and device, electronic equipment and computer program product, belonging to the technical field of natural language processing. The method comprises: obtaining first text data input by a user, and obtaining first input data according to the first text data and key data related to a role-playing model, wherein the key data contains role setting data and effective historical interaction data of the role-playing model; inputting the first input data into a rewriting model to obtain second text data after rewriting, wherein the training data of the rewriting model is obtained by screening through a scoring model; inputting the second text data into the role-playing model to enable the role-playing model to interact according to the second text data. The present disclosure can improve the quality of the rewritten text by rewriting the ambiguous text input by the user, make the role-playing dialogue more fluent and the emotion more consistent, and thus improve the user experience.

The invention discloses a battery maintenance method based on a dynamic sparse neural network and a large language model, and belongs to the technical field of battery management. The method comprises the following steps: acquiring sensor time sequence data and an operation and maintenance text maintenance log of a battery pack, and respectively carrying out standardized preprocessing on the sensor time sequence data and the operation and maintenance text maintenance log; extracting time domain and frequency domain features of the preprocessed sensor time sequence data to obtain sensor features; carrying out semantic coding on the preprocessed text maintenance log by adopting a large language model subjected to battery domain knowledge fine adjustment to obtain text semantic features; performing feature fusion on the sensor features and the text semantic features based on a cross attention mechanism to generate fusion features; and inputting the fusion features into a dynamic sparse neural network which adaptively adjusts the sparseness of each layer of network according to the real-time health state of the battery, and outputting a battery fault probability and a fault root cause weight.

This invention discloses a method for active user detection in a distributed large-scale MIMO-UFMC system based on a covariance algorithm, applicable to the communications field. First, a signal model of the frequency-domain received signal in a large-scale MIMO universal filter multi-carrier system is given, deriving expressions for odd and even components. It is found that even components are independent of transmitted data and unaffected by inter-block interference and inter-carrier interference. A maximum likelihood detection problem is constructed using a covariance-based method, taking the received signal of each user's dominant AP cluster as input to obtain the cost function for each user. Iterative optimization is performed on the activity indicator factor until the cost function no longer decreases, completing the active user detection. This invention effectively improves detection performance by utilizing the differences in system fading coefficients between different access points and the sparsity of active users.

The invention relates to the technical field of wireless communication, and discloses a semi-persistent scheduling method, base station equipment and user equipment.The semi-persistent scheduling method comprises the steps that configuration information of semi-persistent scheduling parameters is sent to user equipment UE, so that the UE conducts semi-persistent scheduling data transmission based on the configuration information; and performing semi-static scheduling data transmission based on the configuration information of the semi-static scheduling parameters. According to the method provided by the embodiment of the invention, the low-delay requirement of URLLC can be met, so that the existing semi-static scheduling transmission mechanism can support higher low-delay requirements, multiple semi-static scheduling configurations are introduced at the same time, and different time-frequency resources are allocated for each semi-static scheduling configuration, so that multiple services can be supported at the same time.

The application discloses a display device color self-adaptive calibration method and system based on image real-time feedback, belongs to the field of display device color calibration and computer vision technology, and relates to the technical field of display device color calibration and computer vision.The method comprises the following steps: controlling a display device to output a preset color sampling sequence; collecting images, establishing a time sequence and space correspondence relationship, and extracting feedback measurement values; binding the display input values and the feedback measurement values into sample pairs, constructing a hierarchical deviation model comprising a global model and a local refinement model, and fusing the hierarchical deviation model; generating compensation parameters; acquiring running parameters and / or environmental parameters in real time, updating the compensation parameters when the changes of the running parameters and / or the environmental parameters meet a triggering condition; and loading the updated compensation parameters to a hardware processing unit to perform mapping processing on pixel data.The application realizes closed-loop calibration, supports dynamic updating of parameters, and improves end-to-end color consistency.

The application relates to a method and device for processing fragmented data, equipment and a storage medium. The method comprises the following steps: receiving target fragmented data to be processed; generating a target group number corresponding to the target fragmented data according to a multicast sequence number, a market channel sequence number and a message sequence number included in the target fragmented data; in the case that the target fragmented data is not the first fragmented data of a UDP message, determining a target storage area for storing the target fragmented data according to the target group number and a preset storage area set, wherein the target storage area is arranged in a cache space or an additional storage space; storing the target fragmented data into the target storage area; reading fragmented data stored in each storage area and performing recombination output. The method can effectively improve the processing efficiency of the fragmented reading and recombination link, and can meet the low-latency and high-reliability requirements in a high-frequency trading scenario.

The invention relates to a novel ultra-low delay GPIO control signal conversion OOK radio frequency signal and signal reduction circuit structure, which comprises a BBU unit and an RRU far-end unit, the BBU unit comprises an OOK signal generator and a radio frequency switch, the RRU far-end unit comprises a detector and a comparator, the output end of the OOK signal generator is connected with the radio frequency switch, and the output end of the radio frequency switch is connected with the detector. The output end of the radio frequency switch is connected with the detector, and the output end of the detector is connected with the comparator. By adopting the novel ultra-low-delay GPIO control signal conversion OOK radio frequency signal and signal reduction circuit structure provided by the invention, a BBU baseband GPIO signal is transmitted to a far-end RRU through an optical fiber in order to meet the requirement of an ROF distributed small base station product, and the low-delay requirement is met. According to the invention, GPIO control switching logic and NR radio frequency transmit-receive time slot synchronization can be maintained. The design scheme is flexible and simple in circuit control, low in cost and high in cost performance.

The present invention provides a data transmission method and system, which relates to the field of communication technology. The method includes: locking a temporary circuit when it is determined that a locking signal sent by a data receiving end is received; setting an enable control signal of the temporary circuit to a first signal, and sending transmission data to the data receiving end along the temporary circuit; stopping sending transmission data to the data receiving end so that the data receiving end can verify the transmission data. The system executes the method. The present invention establishes a routing path from a routing source node to a routing target node based on an on-chip network, and builds the routing path as a temporary circuit, and performs data transmission based on the temporary circuit. Its low latency and determined resources provide large bandwidth and low latency requirements for parallel real-time transmission of multiple data, thereby improving data transmission efficiency.

This invention relates to the interdisciplinary fields of biomedical engineering, artificial intelligence, and neuromodulation, and provides a cognitive fatigue assessment and electrical stimulation intervention system based on a combined EEG-behavioral AI model. The system includes: a data acquisition and preprocessing module for simultaneously acquiring and preprocessing the user's EEG signals and behavioral data; a real-time fatigue assessment module for extracting and fusing features from the preprocessed EEG signals and behavioral data using the combined AI model, outputting the user's real-time fatigue assessment results; the combined AI model is constructed based on the Transduction Information Maximization (TIM) algorithm to achieve cognitive fatigue assessment in scenarios with few samples; and a closed-loop electrical stimulation intervention module for dynamically adjusting transcranial electrical stimulation parameters and performing targeted intervention based on the real-time fatigue assessment results. This invention achieves real-time, accurate detection and personalized targeted intervention of cognitive fatigue states through multimodal data fusion and closed-loop intervention.

This invention provides a data processing method, apparatus, and passive optical network (PON) system. The method includes: allocating a first portion of bandwidth to a first optical network unit (ONU) within a first time window, wherein the first ONU has completed registration and is in an operational state; receiving a first data frame from the first ONU within a time corresponding to the first portion of bandwidth; and detecting a second data frame from a second ONU within the first time window, wherein the second ONU has not yet completed registration. This solves the problem of transmission delay in the PON system caused by the quiet window during the ONU registration process, which prevents the system from meeting the low-latency requirements for carrying mobile services, thus achieving the effect of reducing the transmission delay of the PON system and meeting the low-latency requirements for carrying mobile services.

The invention discloses a data asset table entering method based on a large language model, which belongs to the technical field of data processing, comprises data analysis, semantic annotation, data processing, asset evaluation and dynamic optimization, and realizes an automatic data capitalization process from data extraction, semantic annotation and structured processing to asset evaluation and optimization. The technical problems of efficient analysis, semantic annotation, data processing, asset evaluation, data asset in-table disclosure and dynamic optimization of text data and real-time streaming data are solved, the requirement for manual annotation is reduced, automatic annotation and processing of data are achieved, large-scale text data and real-time streaming data can be efficiently processed, and the data processing efficiency is improved. The method meets the low-delay requirement, provides a comprehensive score and a visual tool, helps a user to visually understand the value of the data assets, and gives higher commercial value to the data assets.

The invention discloses a micro light emitting diode dynamic control system, control method and control device based on a multi-core heterogeneous chip. A task management and light field analysis unit TMU writes an analyzed light field descriptor set into an on-chip shared memory and sends a doorbell signal to a neural mimicry computing cluster NPU; a neural mimicry computing cluster NPU receives the light field descriptor from a task management and light field analysis unit TMU, reads historical data from a distributed shared memory, writes a processing result into the distributed shared memory, and triggers a high-precision driving computing cluster HPU and a distributed pixel driving network DPS; the high-precision driving calculation cluster HPU receives the instruction and the data, reads accurate parameters from the distributed shared memory and issues a final driving micro instruction to the distributed pixel driving network DPS; and the distributed pixel driving network DPS is used for receiving microinstructions and local parameters from the high-precision driving calculation cluster HPU. And inherent contradictions among power consumption, bandwidth, service life and display quality are solved.

The application discloses a kind of automatic wharf horizontal transport equipment dirty detection method and device, construct first level convolutional neural network for detecting IGV from image and second level convolutional neural network for detecting dirty area from IGV, first level convolutional neural network and second level convolutional neural network are constructed into double-branch cascade detection model, the original image obtained from the automatic wharf operation scene is input as the double-branch cascade detection model, and the first level convolutional neural network identifies IGV from complex scene, and outputs the image labeled IGV boundary box, according to the boundary box, IGV subgraph containing only IGV image is cut out from image, so as to remove background interference and transmit to second level convolutional neural network, and second level convolutional neural network focuses on detecting dirty area from IGV subgraph, so as to improve the detection efficiency and detection precision of IGV dirty area.

The present invention belongs to the field of wireless communication technology, and discloses a random access method, system, device and terminal of an adaptive distributed queue. If a device conflict occurs, an adaptive distributed queue conflict resolution method is executed, otherwise the device successfully accesses the network; wherein the adaptive distributed queue conflict resolution method is: when the number of devices in the group is less than the optimal value, the number of devices in the group in the conflict resolution queue is adaptively adjusted according to the device type requirements, so that the relevant performance is optimized, including delay-sensitive devices obtaining the lowest delay or low-power devices consuming the lowest energy consumption; analyzing the impact of uplink authorization restrictions on related performance, so that resource utilization efficiency is optimized. The present invention can adaptively adjust the number of devices in the group in the conflict resolution queue according to the device type requirements, so that the performance of average conflict resolution delay and average conflict resolution energy consumption can be improved. For both delay-sensitive devices and low-power devices, the present invention can achieve optimal performance.