706 results about "Stream data" patented technology

The invention relates to the technical field of computer video processing, and discloses a low-delay video stream real-time processing method and device, and the method comprises the steps: obtaining original video stream data, and processing the original video stream data through employing a lightweight motion prediction method; processing the macro block data set and the predicted coding configuration parameter by adopting multi-thread assembly line coding to obtain a coded data block; establishing a data transmission mechanism to perform data flow control on the unified memory access interface; a heterogeneous task scheduling strategy is adopted to distribute task division results; a lightweight neural network is adopted to carry out parameter adaptive adjustment, and an optimized video stream processing result is obtained; according to the method, a zero-copy data transmission technology is adopted, and optimal configuration and efficient utilization of computing resources are achieved.

The invention discloses a digital human interaction method and device based on multi-modal sentiment analysis and a medium, and relates to the field of artificial intelligence, and the method comprises the steps: collecting multi-modal data of a user in real time through a multi-source sensor device; the multi-modal data comprises face video stream data, voice audio stream data and text dialogue data; calling data analysis engines corresponding to different modalities, and extracting corresponding modal feature sequences; according to the current interaction scene, the modal feature sequence and the historical dialogue context features are fused, and a comprehensive emotion evaluation result is generated; outputting a corresponding multi-modal response data packet based on the modal feature sequence through an interactive response engine corresponding to a comprehensive emotion evaluation result; and executing the multi-modal response data packet. And after feature fusion is carried out in combination with the current interaction scene, the generated response can more accurately fit the current emotion demand and communication context of the user, so that the digital human can be more easily fused into various scenes needing emotion interaction.

The invention provides a photovoltaic power station energy storage management system and method, and relates to the technical field of power system energy storage. The photovoltaic power station energy storage management system comprises the following modules: a data acquisition module, a prediction analysis module, an optimization scheduling module, a health management module, a real-time control module, a power grid interaction module and an energy efficiency evaluation module. And the data acquisition module is used for cleaning photovoltaic array output voltage / current data by adopting an improved Kalman filtering algorithm based on an Internet of Things sensing network, integrating environmental parameters such as irradiance and temperature of a meteorological station through a multi-source data fusion technology, and generating a standardized operation data set. According to the method, photovoltaic array output data is cleaned and integrated through an improved Kalman filtering algorithm and a multi-source data fusion technology, a high-precision standardized data set is constructed, a dual prediction architecture of an attention mechanism long-short term memory network and a time convolution network is combined, and the space-time correlation between photovoltaic output and load demand prediction is remarkably improved.

The invention discloses a traffic network toughness diagnosis method under flood disasters based on a space-time diagram neural network, and relates to the crossing field of traffic engineering and artificial intelligence. The method comprises the following steps: collecting traffic topology, flood monitoring and traffic flow data, and carrying out space-time alignment; a flood coupling dynamic space-time diagram is constructed, a water depth-traffic capacity response mechanism is introduced, and real-time mapping from a disaster physical state to a network topology is realized by utilizing an attenuation function meeting physical monotonicity constraint and dynamically updating an edge weight of a diagram structure according to real-time water depth; inputting the dynamic graph into a pre-trained space-time graph neural network model, extracting space-time evolution characteristics and outputting a toughness diagnosis result; model training adopts a toughness label generated based on an anti-fact baseline to carry out supervised learning, and introduces a physical constraint loss function. According to the method, the problems of decoupling of disaster features and graph structures and unavailability of toughness labels in the prior art are solved, and the physical consistency and accuracy of diagnosis are improved.

The invention discloses an AI-based illumination control system self-adaptive dimming method, which comprises the following steps of: acquiring a two-dimensional coordinate, a motion track curve and voice communication data of a personnel position, and identifying social interaction scene categories of different time slices; determining a social interaction strength score according to the voice communication data, and predicting the most adaptive spectrum configuration data in the current social interaction scene in combination with the social influence factor matrix and the social interaction scene type; according to a historical dimming log and an energy consumption metering curve, identifying a short-term high-frequency dimming segment, and adjusting a real-time response strategy and a disturbance classification processing mechanism of the spectrum configuration data; and adjusting the illumination power distribution values of the high-frequency use area and the low-frequency use area according to the occupation raster data generated by the millimeter-wave radar and the infrared array in combination with the lamp area mapping relation and the power budget constraint. Through multi-source data fusion and a self-adaptive regulation and control mechanism, unification of real-time performance, pertinence and energy-saving performance of spectrum regulation is realized, and the dynamic response capability and the user perception experience of the intelligent illumination system are improved.

The invention discloses a congestion management system and method based on vehicle-road cooperation and dynamic group path optimization, and relates to the technical field of intelligent traffic systems. The method comprises the following steps: S1, collecting original traffic data through a multi-modal sensor array to carry out space-time alignment, and constructing a microscopic traffic flow data set; s2, bidirectional information interaction is carried out through a vehicle-road cooperative communication network; s3, calculating a multi-modal fusion confidence coefficient parameter based on the data quality index, and generating traffic state information and road section real-time saturation; s4, fusing historical and static road network data, constructing a dynamic digital twinborn model and a road network state feature map, and generating a road network load balancing index; and S5, making a decision by adopting a hierarchical-distributed architecture, and generating a group path induction strategy through multi-objective optimization. The efficient collaborative decision is realized through the graph convolutional network and the attention mechanism, and the road network traffic efficiency is comprehensively improved through stable and reliable multi-objective optimization on the premise of guaranteeing the user fairness.

The invention relates to the technical field of subway passenger flow prediction, and discloses a space-time perception subway passenger flow prediction method based on large model and multi-source information fusion, which comprises the following steps: constructing a data agent based on a large language model and a model context protocol; generating text description of each subway station and a corresponding spatial context embedding vector based on the peripheral interest point data of each subway station by utilizing a data agent; constructing an enhanced graph neural network according to the spatial context embedded vector, and performing spatio-temporal joint prediction based on the historical passenger flow data and the spatial context embedded vector by using the enhanced graph neural network to obtain a preliminary passenger flow prediction result; and constructing a cue word template, and performing domain knowledge calibration based on the cue word template by using the pre-trained large language model to obtain a final passenger flow prediction result. According to the method, external multi-source information is deeply fused, so that the urban rail transit passenger flow prediction task performance, robustness and interpretability are remarkably improved.

The invention relates to the technical field of energy systems, and discloses an energy system intelligent decision-making method, which comprises the following steps: acquiring four-flow data of an industrial energy system, and constructing a four-flow parameter model based on a production flow topological graph; constructing a multi-agent reinforcement learning environment, taking the four-flow parameter model as a state space of the multi-agent reinforcement learning environment, and determining an action space and a reward function of each agent; performing domain adjustment on the large language model based on the industrial energy system knowledge base; through combination of a multi-agent reinforcement learning model and a large language model, a decision strategy is dynamically adjusted according to real-time data feedback, and decision feasibility verification is performed based on a physical mechanism model. According to the method, the four-flow integrated parameter model is constructed, and a multi-agent reinforcement learning and large language model cooperation mechanism is introduced, so that the problems of data splitting and the like in a traditional energy management system are effectively solved, and the intelligent level of energy system decision making is improved.

The invention relates to the technical field of fire early warning, in particular to an early fire early warning system based on AI image recognition, and the system comprises an image collection module which is used for obtaining the video stream data of a monitoring area in real time; an AI image analysis module which is in communication connection with the image acquisition module and is used for receiving the video stream data and carrying out real-time analysis on video frames based on a pre-trained fire identification model so as to extract visual features related to the fire; and the early warning judgment module is in communication connection with the AI image analysis module and is used for receiving an analysis result of the visual features. According to the early fire early warning system and method based on AI image recognition, through video image analysis, the system can recognize weak flame or smoke characteristics at the initial stage of a fire and when naked eyes do not obviously see the characteristics, and the delay problem that a traditional smoke-sensing and temperature-sensing detector needs to wait for physical parameters to be diffused to the detector to be triggered is solved.

The invention relates to the technical field of data processing, in particular to a relational database intelligent query method based on machine learning, which comprises the following steps of: processing multi-modal flow data through time sequence alignment, generating a unified semantic representation vector, constructing a dynamic psychological state map, and modeling a psychological state evolution track by utilizing a neural common differential equation mechanism. After user query is received, historical dialogue nodes are retrieved from the graph, enhanced query intention representation is generated, the enhanced query intention representation is converted into an execution plan through a neural symbol inference engine, and a graph neural network is adopted to predict execution cost. And finally, a personalized analysis report is generated by combining a causal discovery algorithm, and system adaptive optimization is realized through feedback signals. According to the method, the problems of inconsistent time sequence semantics and strong context dependency of the multi-modal psychological data are effectively solved, and the query accuracy and the personalized level in a psychological dialogue scene are improved.

The invention relates to the technical field of ocean current data analysis, in particular to a multi-scale modeling and verification method and system based on ocean current spring layer structure analysis. And obtaining the thermocline and halocline parameters of the target sea area and carrying out abnormal value elimination and spatial interpolation to form a thermocline data set. A convolutional neural network is used to extract spring layer space structure features, and based on Gaussian curvature quantification boundary complexity, a thermocline and halocline equivalent geometric model is constructed. On the basis, taking the equivalent geometric model as a boundary condition, and performing multi-scale decomposition on the Navier-Stokes equation set to generate a dynamic approximate equation; and then inputting the approximate equation and the equivalent geometric model into a physical information neural network for constraint training to obtain a multi-scale flow field prediction model, and comparing the multi-scale flow field prediction model with actually measured flow field data to complete verification. According to the method, automation, refinement and multi-scale coupling of spring layer structure modeling are achieved, and scientificity and engineering applicability of ocean current modeling are improved.

The invention discloses a cable temperature detection method, which comprises the following steps: laying distributed optical fibers along a to-be-monitored cable, synchronously obtaining whole-line temperature, vibration sound wave signals and load current data, and carrying out dynamic compensation and correction on a temperature measurement value; on the basis of multi-source information fusion, abnormal points are accurately positioned, a cable multi-physical field digital twinborn model is constructed, a health index reflecting the insulation condition is calculated in real time, and quantitative evaluation of local and overall states is achieved; further simulating different load scenes to predict the time when the insulation performance is degraded to a failure threshold value, so as to output probability distribution of the remaining service life of the cable; according to the method, the whole-process intelligentization from accurate temperature measurement and abnormal positioning to state evaluation and life prediction is realized, and the perspectiveness and reliability of cable operation and maintenance are remarkably improved.

The invention discloses a data forwarding processing method and system for a campus network, and the method comprises the steps: controlling a plurality of independent cameras to generate a multicast video stream carrying a permission identifier according to an access control terminal triggering event; the method comprises the following steps: synchronously creating (S, G) table entries with permission marks and never expired (*, G) table entries on a path multicast router according to a first video stream data packet received by a multicast source side designated router source DR; according to an IGMP join message which is sent by a receiver and carries a permission identifier, through a matching result of an RP comparison (*, G) table entry output interface permission and a multicast data permission identifier, whether a data stream is forwarded along a shared tree (RPT) is decided; and according to a matching result of the permission identifier of the new receiver and the pruned (S, G) flow permission, activating a pruning state on the multicast path and recovering data flow forwarding. According to the embodiment of the invention, occupation of invalid data streams on campus network bandwidth can be reduced, and the overall utilization rate of network resources is improved.

The invention discloses an eddy current data compression transmission optimization method, relates to the technical field of data compression transmission, and solves the technical problems that a compression algorithm lacks scenario adaptation and data transmission lacks priority and link adaptation. Scene demand-algorithm characteristic accurate matching is achieved, compressed data overall reconstruction errors and defect area reconstruction errors are reduced, delay and compression ratio requirements of different scenes are met, data priorities are judged in a multi-dimensional mode, it is ensured that defect data acquisition resources are inclined, key information is prevented from being lost, and the method is suitable for large-scale popularization and application. A transmission protocol is optimized for a short / medium / long distance scene, the protocol adaptability and the transmission efficiency are improved, a transmission scene-link state-data priority three-dimensional linkage transmission strategy is constructed, the bandwidth occupation proportion is dynamically adjusted, core data are preferentially transmitted when links are congested, and resources are fully utilized when the links are idle.

According to the invention, an unmanned aerial vehicle platform is creatively combined with the improved YOLOv11, and the unmanned aerial vehicle construction site safety helmet intelligent inspection method based on the improved YOLOv11 is provided. The method comprises the following steps: firstly, carrying out digital planning on an inspection route, and generating an executable flight path file; the unmanned aerial vehicle carries high-definition camera equipment, and carries out omnibearing video acquisition on a construction area according to a preset route; and the ground control center receives the real-time video stream data and inputs the real-time video stream data into a specially-designed safety helmet identification model to complete intelligent analysis of the state of personnel safety equipment and record of violation events. The detection model is subjected to triple optimization on the basis of a YOLOv11 architecture: a C3k2SPD module is adopted to replace a C3k2 module in an original backbone network, an ASPP module is introduced to replace an original SPPF module, and meanwhile, an NWD loss function is fused at a detection head part, so that the small target recognition capability and the complex scene fitness are remarkably improved.

The invention provides a video data processing method and apparatus, and an electronic device. The method comprises the steps of obtaining initial video data; the initial video data comprises initial video stream data and initial audio stream data corresponding to the initial video stream data; determining a role feature of at least one audio generation role in the initial video data; determining scene features of the initial video data; based on the role features, the scene features and the content translation text corresponding to the initial audio stream data, determining input data; inputting the input data into a preset voice generation model, and generating target audio stream data through the voice generation model; and generating target video data based on the initial video stream data and the target audio stream data. According to the mode, in the process of generating the translated audio corresponding to the video, tone cloning and emotion restoration of the translated audio are realized through the role features of the speaker of the original video and the scene features embodied by the video, the translation effect of the video is improved, and the video translation cost is reduced.

The invention relates to a lightweight acoustic feature extraction method for end-side audio and video quality inspection. The method comprises the following steps: end-side equipment separates audio and video stream data through a double-time-sequence anchor point alignment method, and acquires independent audio data; performing grading preprocessing on the independent audio data to obtain noise-reduced independent audio data; on the basis of the segmented audio, a time domain and frequency domain collaborative extraction algorithm is adopted, multi-dimensional time domain features and frequency features sensitive to audio quality difference are obtained, and a lightweight acoustic feature vector is obtained through an incremental principal component analysis method; based on the lightweight acoustic feature vector, utilizing a dual-threshold matching judgment mode to obtain similarity between features; and on the basis of the feature similarity, through feature dimension deviation verification, obtaining a quality inspection result containing a standard grade and feature anomaly positioning. Lightweight acoustic feature extraction of end side audio and video quality inspection is realized.

The invention relates to the technical field of numerical control machine tool precision compensation and intelligent control, and discloses a numerical control machining center spindle dynamic compensation control method and system.The numerical control machining center spindle dynamic compensation control method comprises the steps that surface topography point cloud data of an online visual sensor, spindle motor current data and feed shaft position encoder feedback data are obtained, and a multi-source state monitoring data set is generated; a regularization method is adopted to solve reverse mapping from the surface low-frequency error component to radial runout and axial movement of the main shaft, and a main shaft rigidity state index and a tool abrasion index are output; describing an interaction relationship between the tool wear and the main shaft rigidity based on a coupling degradation kinetic equation, and calculating contribution weights of the tool wear and the main shaft rigidity to the machining precision; training a deep Q network strategy by using the deduced reward function, and outputting an optimal action; and executing spindle dynamic compensation parameter adjustment or generating a replacement instruction according to the optimal action type. According to the method, the factor of lacking a continuous learning improvement mechanism from historical data is overcome.

The invention relates to the technical field of industrial equipment control, and discloses a dynamic maintenance management method and system based on the full life cycle of test equipment, and the method comprises the steps: obtaining a time domain vibration signal and torque current data during the operation of the equipment through a high-frequency vibration sensor and a current loop feedback interface; calculating an accumulated damage gradient parameter representing the physical recession degree of the mechanical transmission chain; the parameters are input into a preset damage stress coupling model to solve the maximum radial impact force allowed to be borne by the damaged part, and the impact force is mapped into a dynamic capability boundary for carrying out logic constraint on the motion planning instruction; and the motion controller reconstructs an ideal speed curve by using the boundary in an interpolation period, and reshapes a torque load spectrum output to a servo motor by reducing the jump of acceleration and deceleration sections. The problem that an existing static control strategy is mismatched with the dynamic aging characteristic of equipment is solved.

The invention belongs to the technical field of traffic flow green wave control, and particularly relates to a green wave control method based on traffic flow prediction driving. The method comprises the following steps: acquiring multi-dimensional real-time data in a distributed manner; performing abnormal value detection and data interpolation based on space-time neighborhood analysis to generate high-quality time sequence traffic flow data; modeling a road network into a dynamic graph fusing static connectivity and real-time traffic flow coupling degree, and constructing a double-time-scale time sequence diagram sequence; inputting an improved space-time diagram neural network, extracting time features by optimizing Bi-GRU, capturing spatial association by a multi-head attention mechanism, and outputting an accurate road section-level traffic flow prediction result through residual fusion; and finally, constructing a multi-objective optimization model based on prediction, solving by using an improved multi-objective particle swarm algorithm, obtaining a signal timing scheme, and issuing and executing the signal timing scheme. The method can dynamically adapt to traffic changes, improves the traffic efficiency, and avoids green wave failure and congestion.

The invention discloses a locomotive depot turnout state recognition and linkage warning method based on visual tracking, relates to the technical field of rail traffic signal processing, is applied to a locomotive depot station track automation system, and comprises a station track automation manipulator and an algorithm server which are connected through a local area network. The method comprises the following steps: S1, receiving and analyzing a route selection and arrangement command, generating a starting command according to a constructed turnout-camera mapping table to trigger a target turnout to act, and executing an image recognition task by a target camera; s2, the target turnout performs corresponding actions according to the route selecting and arranging command, and meanwhile, a target camera is triggered to dynamically track the actions of the turnout and collect a video stream; s3, performing multi-dimensional image intelligent identification according to the collected video stream data, wherein the identification specifically comprises turnout state identification, personnel and obstacle detection and crossing safety behavior identification; and S4, performing intelligent warning and bidirectional feedback according to an image intelligent identification result. According to the invention, the turnout state identification accuracy and efficiency are improved.

The invention relates to the technical field of public transport organization, in particular to an urban traffic station connection evaluation method based on multi-source data, which comprises the following steps of: performing passenger flow analysis on an upstream public transport section of a station, dividing passenger flow time periods, and acquiring multi-source connection situation data by adopting a sampling or real-time monitoring mode based on different time periods; calculating a transport capacity supply saturation index based on the station upstream public traffic section passenger flow data, calculating a dynamic transport capacity service efficiency index based on the mobile passenger-carrying end parking characteristic data, weighting to obtain a comprehensive connection bearing coefficient, estimating the real-time passenger flow arrival flow based on the arrival shift data of the traffic station and the number of resold tickets, and determining the real-time passenger flow arrival flow. And evaluating the urban traffic connection capability of each station by generating a real-time connection efficiency value. According to the method, the quantitative evaluation of the operation state of the connection system of the urban traffic station is realized, the defects of a traditional method in the aspect of multi-source data fusion are overcome, and the analysis precision of the connection capability of the urban traffic is improved.

The invention provides a power distribution network resonance key element identification method and device based on harmonic monitoring data, and the method comprises the steps: collecting the harmonic voltage and branch harmonic current data of a key node through the deployment of an online harmonic monitoring device, converting a time domain signal to a frequency domain through band-pass filtering for a preset harmonic number set, and extracting a complex amplitude value, carrying out normalization processing to obtain standardized data; secondly, constructing a node admittance matrix in a frequency domain in combination with the newest power grid topology and equipment static parameters, identifying a feature root closest to zero and a corresponding resonant frequency thereof through eigenvalue decomposition, and obtaining a modal vector; tiny disturbance is applied to inductive reactance / capacitive reactance element parameters, and the sensitivity of the inductive reactance / capacitive reactance element parameters to modal admittance is calculated and normalized into dimensionless indexes; and screening and sorting according to the sensitivity threshold, and outputting a group of resonance key elements. The method does not need off-line simulation, has the advantages of high real-time performance, high engineering adaptability, low implementation cost and the like, and provides efficient technical support for active treatment of harmonic waves of the power distribution network.

This application provides an encoding, encapsulation, and display method, apparatus, and electronic device. The encoding method includes: acquiring an image to be encoded; then, determining the encoding scale N (N is an integer greater than 1) for the image, and determining the encoding parameters corresponding to each of the N encoding scales, resulting in N sets of encoding parameters. Subsequently, based on the N sets of encoding parameters, the image is encoded N times using a pre-set single-scale encoder to obtain N sets of bitstream data. In this way, encoding the image yields bitstream data with a smaller data volume (i.e., low-encoding-scale bitstream data), thus increasing the probability of the image bitstream data reaching the decoding end during network fluctuations, thereby ensuring smooth video playback. For high-encoding-scale bitstream data, if lost, it will not affect normal decoding and playback at the playback end, and will not cause screen tearing; if it reaches the decoding end, it can also improve the playback quality.

The invention provides a video encryption transmission method and device and a video decryption output method and device. The video encryption transmission method comprises the following steps: constructing a security parameter set according to an encrypted video encryption key and an initialization vector of an image group; performing XOR operation on the first L-1 bytes of the original data byte stream by adopting the stream key to obtain encrypted byte stream data; setting a network abstraction layer header for each network abstraction layer unit after the image group is processed, wherein the network abstraction layer header at least comprises original coding standard information and encryption state information; performing hash signature processing on the encrypted byte stream data of the image group to generate signature data, and packaging the signature data into an authentication network abstraction layer unit; packaging the security parameter set, the network abstraction layer unit provided with the network abstraction layer head, the authentication network abstraction layer unit and the original parameter set of the image group according to the transmission sequence of the original code stream of the image group to obtain an encrypted video code stream; and sending the encrypted video code stream to a receiver.

The invention discloses a communication data analysis system and method for network security, and relates to the technical field of computer internet. Time sequence features, protocol semantic features and interactive topology features of communication session historical data are extracted based on network flow data; the method comprises the following steps of: acquiring a time sequence feature, a protocol semantic feature and an interactive topology feature, fusing the time sequence feature, the protocol semantic feature and the interactive topology feature into a unified high-dimensional feature vector, acquiring a communication behavior record and a communication behavior dynamic feature vector of network equipment in a communication network, calculating a digital feature of a coupling relationship evaluation value, and when a certain communication session occurs, judging whether the communication session occurs or not. The method comprises the following steps of: calculating real-time coupling relationship evaluation values among network equipment, quantifying the difference degree of the real-time coupling relationship evaluation values through digital characteristics, accumulating the coupling relationship evaluation values in the process of performing a certain communication session, calculating the total anomaly degree of the communication session, and calculating the abnormal degree of the communication session. The method aims at solving the problems that advanced persistent threats are difficult to effectively recognize, feature expression is insufficient and the perceptual ability is weak in the prior art.

The invention relates to a multi-language voice content recognition method and system, and belongs to the technical field of voice signal processing, and the recognition method comprises the steps: collecting original audio stream data, executing the noise reduction filtering processing, and segmenting the original audio stream data into a plurality of audio segments; extracting acoustic feature vectors of the audio clips; inputting the audio clip into the speech recognition model group, and obtaining a text clip and a confidence score; fusing all the text fragments to generate a preliminary recognition text, segmenting the preliminary recognition text into text blocks, and translating the text blocks into a target language text block by block through a streaming translation model in combination with a semantic feedback tag received in real time; executing bidirectional translation verification on the text blocks of which the confidence scores are lower than a set threshold value; inputting the translated text stream and the acoustic feature vector into a semantic analysis model, and fusing an output result to obtain a semantic feedback tag; and constructing a cross-modal association graph through the graph neural network, outputting a key information abstract and triggering an alarm signal. The speech recognition accuracy can be improved while the recognition efficiency is guaranteed, and risk response is carried out.

The invention discloses a network load balancing method, a computer program product, equipment and a medium, relates to the technical field of data centers, is applied to a virtual switch of a data center, and obtains to-be-sent data; packaging the to-be-sent data to obtain initial packaged data; in the quintuple, determining an item to be transformed; in the initial encapsulation data, transforming a value corresponding to a to-be-transformed item in the outer layer message to obtain target encapsulation data; performing hash operation based on the quintuple of the target encapsulation data to obtain a hash value; and transmitting the target encapsulation data according to a transmission path corresponding to the hash value. The method and the device have the beneficial effects that different transmission paths can be used for transmitting the target encapsulation data, and finally, different to-be-sent data of the same stream are transmitted through different transmission paths, so that the occupation of a single transmission path by the data of the same stream is avoided, load balancing between the transmission paths is realized, and the transmission efficiency is improved. The network load balancing capability of the data center is improved; and the universality is good.

The invention discloses a large-model-driven urban railway newly-opened station flow prediction method, and relates to the technical field of urban railway traffic operation management. The technical problems of low prediction precision, poor interpretability of a traditional model and insufficient reasoning flexibility in a complex urban scene caused by lack of historical passenger flow data of a newly opened station of an urban rail are solved. The method comprises the following steps: firstly, collecting pre-processed passenger flow related multi-dimensional data, and then constructing a multi-dimensional reference system to retrieve a target station similar reference or extract a macroscopic index reference; judging a scene category through a rule engine, calling a corresponding template, and filling information to generate a structured cue word; inputting the cue word into the large language model, and outputting a content prediction value and a result of a natural language reasoning chain according to the reasoning thinking chain template; and finally, checking the result, and completing optimization of the method based on feedback. The method has a good application prospect in the technical field of urban rail transit operation management.

The invention relates to the technical field of hydraulic engineering safety monitoring, and discloses a hydraulic engineering slope deformation monitoring method and system based on a video stream. The method comprises the following steps: deploying video acquisition equipment in a slope stable area, and acquiring continuously shot video stream data containing a slope monitoring area; extracting a plurality of natural feature points from an initial frame of the video stream data; calculating a pixel displacement vector based on the position change of the natural feature point in the subsequent video frame; based on the spatial distribution characteristics of the pixel displacement vectors, natural feature points with consistent motion trends are identified as the same deformation area, and cluster displacement vectors are generated; and calculating the physical space displacement of the deformation area according to the cluster displacement vector and a coordinate conversion model. The system comprises corresponding function modules. According to the invention, automatic and real-time monitoring of slope deformation is realized, the monitoring efficiency and accuracy are improved, and the cost is reduced.