10results about How to "Low latency requirement" patented technology

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.

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 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.