82results about How to "Realize dynamic optimization" patented technology

The invention discloses a point cloud intelligent cutting and dynamic optimization method based on edge computing, and belongs to the technical field of intelligent traffic systems and edge computing, and the method comprises the steps: collecting three-dimensional point cloud data through a multi-line laser radar disposed on an RSU; generating a three-dimensional bounding box through a three-dimensional target detection network in an RSU edge calculation unit, generating an ROI mask, expanding a buffer region, and performing adaptive region cutting on the point cloud to form an ROI point set and a background point set; a perception-driven cutting decision is realized, and ROI extraction is converted into a self-learning process based on semantic features from fixed threshold judgment; voxelizing the non-uniform resolution point cloud, generating a unique hash index for each voxel unit, comparing a hash set of a current frame with a hash set of a previous frame, extracting newly added and disappeared point sets, and constructing a differential data packet according to a change ratio; while high timeliness is kept, data redundancy is reduced, and the point cloud updating rate and the incremental transmission efficiency are improved; and dynamic optimization of communication-calculation cooperation is realized.

The invention relates to the field of numerical control machining and computer-aided manufacturing software, and particularly discloses a honeycomb core six-axis ultrasonic cutting track digital control and interpolation optimization system which comprises a track preprocessing module, a parameterized spline interpolation module, a six-axis linkage inverse solution module, a dynamic feed-forward compensation module, an ultrasonic energy adaptive control module and a multi-source information fusion and decision module. By the adoption of the technical scheme, smooth and efficient interpolation of a track, dynamic feed-forward compensation of motion errors and self-adaptive adjustment of cutting energy can be achieved, and therefore the machining precision, the surface quality and the system stability are improved. According to the invention, ultrasonic energy control is originally incorporated into an overall trajectory control closed loop, and real-time accurate regulation and control of cutting energy are realized through the ultrasonic energy adaptive control module. The module is based on force, amplitude, temperature and other multi-sensor information, adopts fuzzy reasoning to intelligently decide output power, and ensures that cutting energy is constant and optimal under different processing conditions.

The present application relates to the technical field of energy. The present application relates to a multi-power full-time energy tracking energy storage system. It comprises a data effective reservation unit, a PCS matching unit, a VSG control working unit, a parameter updating simulation unit and a simulation parameter reservation unit; the data effective reservation unit is used for collecting load power data, and simultaneously performing effectiveness analysis on the load power data; according to the analysis result, effectiveness reservation is performed; the PCS matching unit is used for establishing a PCS group, and simultaneously performing working parameter analysis on the multi-power PCS in combination with historical load power data in the PCS group; through the combination of the reserved real-time load power data and the PCS group, multi-power PCS efficiency value evaluation is performed, the predicted efficiency value of each multi-power PCS under the condition of real-time load power data can be obtained, the multi-power PCS with the highest predicted efficiency value is selected as the working parameter for input, working mode switching is performed, and the overall efficiency of the system can be improved.

This invention discloses a method for simulating the transportation efficiency of long-distance, steep-slope tunnel construction adit. The method divides the material transportation process of a tunnel boring machine (TBM) into two phases: heavy-load entry into the tunnel and empty-load exit. Based on the kinematic principles of force, motion models are sequentially established for heavy-load external tunnel transportation, heavy-load adit downhill transportation, heavy-load main tunnel transportation, empty main tunnel transportation, empty adit uphill transportation, and empty external tunnel transportation. A time-discrete numerical integration method is used to calculate the acceleration, velocity, displacement, and running time of each motion model at each moment with a small time step. By simulating the transportation efficiency under different equipment parameters and working conditions using the motion models, the selection of transportation equipment, adit slope design, and braking system configuration parameters are determined. The advantages of this invention are that it achieves accurate dynamic simulation calculation of the transportation efficiency inside and outside the tunnel during TBM construction, providing intuitive guidance for transportation equipment selection, braking system configuration, and optimization of adit slope design.

This invention provides a method for controlling the weight of a solid tire carcass and its application, relating to the field of tire carcass technology. The method includes the following steps: (a) weighing the base rubber after molding and comparing it to the standard weight of the base rubber to obtain a weight difference value; (b) correcting the amount of intermediate rubber based on the weight difference value from step (a), weighing the tire carcass after molding the intermediate rubber, and comparing it to the standard weight of the intermediate rubber tire carcass to obtain a weight difference value; (c) correcting the amount of tread rubber used in one stage based on the weight difference value from step (b), weighing the tire carcass after molding the first stage of tread rubber, and comparing it to the standard weight of the first stage of tread rubber tire carcass to obtain a weight difference value; (d) correcting the amount of tread rubber used in the second stage based on the weight difference value from step (c), and molding the tread rubber in two stages to obtain the finished tire carcass. This invention, through segmented weighing and multi-stage correction, gradually offsets the weight errors in each molding stage, achieving a tire carcass weight error of ±0.1 kg, far superior to existing weight control precision.

This invention belongs to the field of copper alloy material design and preparation technology, and discloses a new process for efficiently designing high-performance copper-iron alloys. Based on the 5% Fe content vertical cross-sectional characteristics of the Cu-Fe-Si ternary phase diagram, the Si content is designed in stages (0-4%) to achieve targeted performance optimization. Combined with a three-stage heat treatment process of "homogenization-basic aging-gradient aging", a closed-loop system of "phase diagram design-preparation-detection-optimization" is constructed. Through precise composition control by EPMA and real-time feedback of mechanical properties, the problems of blind composition design and insufficient precipitation of strengthening phases in traditional processes are solved. The resulting alloy has excellent tensile strength, yield strength, and total elongation at break, with synergistic improvement in mechanical and magnetic properties, and a design efficiency improvement of more than 40%, making it suitable for electronic, mechanical and other fields.

The invention discloses an intelligent empty can detection device for a ring-pull can production line and a detection method of the intelligent empty can detection device. An input station, a tank opening detection station, a tank body inner wall detection station, a tank body inner bottom detection station, a tank body outer wall multi-detection station, a tank body bottom outer side multi-detection station and an output station are sequentially arranged on the circumference of the tank body, a positioning assembly capable of being matched with tank bodies with different diameters is arranged, and the input station is provided with a height and diameter detection sensor; a digital twinborn system is innovatively introduced, a'physical detection device-virtual twinborn model 'real-time mapping system is constructed, and detection full-process visualization, data full-link tracing and virtual simulation optimization are achieved. According to the invention, all-directional defect detection of the zip-top can is realized, the dynamic optimization capability of detection parameters, the equipment fault pre-judgment efficiency and the production collaboration level are improved through the digital twin technology, the method is adaptive to cans of different specifications, the detection precision and generalization are remarkably improved, and the method is suitable for a high-standard zip-top can production line.

This application relates to the field of computer technology, and in particular to a method and system for dynamic optimization of station guidance areas based on reinforcement learning. The method includes: acquiring station operation data and mapping it to preset station guidance areas to construct real-time state distribution results; determining regional change trend results for each preset station guidance area based on the real-time state distribution results; identifying bottleneck areas within the preset station guidance areas based on the regional change trend results, and generating candidate adjustment results for guidance areas by filtering adjustable boundaries around the bottleneck areas along the passenger flow direction; constructing a reinforcement learning environment state by combining the real-time state distribution results and the regional change trend results, and using a dual deep Q-network to perform action optimization in the candidate adjustment results to determine the target guidance area division result; and optimizing the preset station guidance areas based on the target guidance area division result. This application can dynamically optimize station guidance areas based on real-time station operation status and passenger flow distribution changes.

The application provides a production method of high-corrosion-resistance and high-strength galvanized steel strand, relates to the technical field of hot-dip plating, and comprises the following steps: collecting hot-dip plating related characteristic parameters and setting basic hot-dip plating time; pre-processing the hot-dip plating related characteristic parameters and outputting a hot-dip plating calibration coefficient; establishing a coupling formula and obtaining a hot-dip plating calibration time; collecting strength interference characteristic parameters; based on the collected strength interference characteristic parameters and the obtained actual hot-dip plating required time, calculating a tensile strength prediction score of the produced galvanized steel strand, and determining the eligibility of the galvanized steel strand prepared under the current working condition parameters. The application calculates a most suitable galvanizing time, ensures that the zinc layer is fully penetrated, and avoids excessive reaction to cause the steel strand to be brittle. Meanwhile, the application can predict whether the tensile strength of the finished product is qualified before production, realizes dynamic optimization, early warning and self-learning in the whole process, and effectively reduces the waste product rate.

The application belongs to the field of artificial intelligence model compression and edge deployment in intelligent operation and maintenance and fault diagnosis of power equipment, and discloses a transformer cooperative distillation incremental learning method and system based on time sequence consistency, which comprises the following steps: a teacher model is used to screen a no-label sample set to obtain a pseudo-label sample set, the pseudo-label sample set is combined with an original label sample set to obtain a distillation training sample set; a multi-mechanism cooperative distillation training scheme containing soft and hard label joint distillation, time sequence consistency distillation and multi-task distillation is constructed, and a student model is subjected to distillation training through the multi-mechanism cooperative distillation training scheme; and a cooperative distillation incremental learning scheme in which a cloud end continuously updates a teacher model to adapt to new multi-source monitoring data and a student model after edge distillation generates a pseudo-label to expand the distillation training sample set is constructed. The application can be used for realizing high-precision and low-cost training and continuous updating of a student model in a resource-limited terminal or an online scene.

This application provides a method and system for dynamic knowledge updating of a large-scale model for operation and maintenance (O&M) technical services. It relates to the field of O&M technology, and models the knowledge generation triggering factors, iteration impact range, and knowledge association strength in multi-source O&M knowledge update sources to obtain a knowledge full-link evolution set including knowledge generation trigger records, iteration impact assessment results, and cross-knowledge association networks. Then, dynamic coupling rules are generated based on the real-time scenario characteristics of O&M tasks and this knowledge full-link evolution set. These rules are used to filter and integrate the knowledge full-link evolution set to obtain a scenario-coupled knowledge set. This set is embedded into the knowledge architecture of the large-scale O&M technical service model, and storage node parameters and retrieval logic are adjusted to obtain the basic updated model. Finally, the basic updated model is dynamically adapted in multiple dimensions, and the knowledge call priority sequence and storage node parameters are adjusted in conjunction with changes in scenario characteristics to generate an adapted updated model, thereby improving the adaptability of the large-scale O&M technical service model.

The invention discloses a method for regulating and controlling efficient conversion of methanol based on (p) ppGpp signal molecules. Specifically, by regulating synthesis and hydrolysis balance of (p) ppGpp signal molecules in microbial cells and combining promoter engineering and nutrition condition optimization, methanol metabolic pathway key gene expression and amino acid metabolic flux are enhanced, and the methanol utilization efficiency and target product synthesis capacity of microorganisms are remarkably improved. The regulation and control mechanism of the (p) ppGpp on methanol conversion under the stress of a low nitrogen source is defined for the first time, efficient conversion of methanol to organic acids such as acetic acid and butyric acid is realized, a brand new strategy is provided for preparation of high-added-value chemicals through methanol bioconversion, and the method has the advantages of being economical, environmentally friendly and high in conversion efficiency and has a wide application prospect.

This invention relates to the field of tunnel construction technology, and more particularly to a system and method for dynamically optimizing tunnel construction parameters based on geological information from point cloud data. At multiple monitoring time points, point cloud data is collected from the tunnel segment to be constructed, obtaining multiple point cloud datasets. The tunnel segment is divided into multiple construction areas, and multiple regional point cloud datasets are obtained from these datasets. Point cloud change analysis is performed on each regional dataset to obtain multiple point cloud change rates. Based on these regional point cloud datasets, reflection intensity analysis is conducted to obtain multiple reflection intensity coefficients. Combined with the multiple point cloud change rates, geological analysis is performed to obtain the distribution of concealed geological information. Based on this distribution, tunnel construction parameters are dynamically optimized to obtain optimized tunnel construction parameters for tunnel construction. This achieves the technical effect of improving the efficiency and safety of tunnel construction while reducing construction risks and costs.

This application provides a path planning method, apparatus, device, storage medium, and vehicle, relating to the field of autonomous driving technology. The method includes: acquiring road information of a target road where the vehicle is located and first driving information of the vehicle; wherein the first driving information indicates the current driving state of the vehicle; discretizing the target road based on the road information to obtain a discrete target road; wherein the discrete target road includes multiple drivable segments, and each drivable segment includes multiple drivable positions; determining second driving information of the vehicle at each of the drivable positions based on the first driving information and / or the road information; the second driving information indicates the driving state of the vehicle in a preset dimension; and determining the optimal driving path of the vehicle based on the second driving information. The method of this application can improve the accuracy and applicability of path planning.

The invention discloses a plastic product surface defect detection device based on machine vision, which relates to the technical field of plastic product detection, and comprises a support plate, a positioning table, a three-axis adaptive assembly arranged between the support plate and the positioning table, a compensation assembly arranged in the assembly, and a visual detection soft communication system. The three-axis adaptive assembly realizes three-axis movement of the visual camera, and the compensation assembly can finish micro-motion compensation of the camera and make up for a detection blind area of a large-size product; the visual detection soft communication system is in two-way signal connection with all execution and sensing components, automatic linkage control, image acquisition and processing and intelligent defect recognition and judgment are achieved, detection signals can be received, action parameters can be dynamically adjusted, and closed-loop detection is formed. The device solves the problems that traditional detection equipment is limited in detection range and low in recognition precision, improves the adaptability to products of different specifications, achieves defect global accurate detection, reduces defective product outflow, and is suitable for the production quality control link of plastic products.

A brake handle with a dynamically variable rotation center in the braking process comprises a brake shell, a piston assembly, a push rod assembly, a brake handle, a multi-fulcrum linkage component and a rotation switching mechanism. The shell is provided with a piston cavity. The handle drives the piston to move through the push rod. The linkage component is provided with two rotating fulcrums, and the switching mechanism can switch two working states. When the front section of the handle is pinched, the handle rotates around the first fulcrum, idle stroke is eliminated, and operation is labor-saving. When the rear section is pinched, the second fulcrum is switched to rotate, and the lever ratio is increased to improve braking force. By means of automatic switching of double fulcrums, labor-saving control and large braking force output are both considered, and the problem that a traditional hydraulic brake handle is poor in performance is solved; the modular structure is easy to adapt to mass production, and switching nodes are accurate and controllable; a locking and failure protection structure is arranged, so that the operation is stable and safe; meanwhile, eccentric wear of parts is eliminated, the service life of sealing pieces is prolonged, and braking efficiency and use safety are improved.

The application discloses a motion control method for preventing glue stacking of a glue dispenser, and relates to the technical field of intelligent control.The method comprises the following steps: constructing a non-Newtonian fluid dynamic model through a power law equation, and generating a four-dimensional compensation coefficient matrix; predicting glue amount deviation through a convolutional neural network, calculating compensation glue amount based on linkage of a piezoelectric valve and a screw valve, obtaining a control instruction set; constructing a Jacobian matrix to calculate an end equivalent inertia force, and injecting a reverse compensation current through a servo driver to synchronously optimize a multi-axis motion trajectory, and re-planning a path through a curvature radius constraint at an inflection point; detecting glue line height through laser height measurement, triggering an online learning algorithm to update convolutional neural network weights, switching process parameter files, and controlling synchronization of a conveying belt beat and a glue dispensing gap time; monitoring glue line height, executing an abnormal glue stacking classification processing strategy, and recording fault data packets.The application realizes glue stacking prevention of the glue dispenser through lightweight convolutional neural network prediction and multi-axis motion trajectory optimization.

The utility model discloses an anti-deformation waste material collecting device for metal material manufacturing, which relates to the technical field of metal waste material collection, and comprises a waste material box, a lifting structure arranged on the waste material box and used for limiting the descending distance of metal waste materials; the lifting structure comprises a first sliding groove, the first sliding groove is formed in the bottom of an inner cavity of the waste box, a driving motor is installed on one side of the waste box, a rotating shaft is fixed to an output shaft of the driving motor, the top of the waste box is sleeved with a hopper, and a second sliding groove is formed in the hopper. And a fixing shaft is fixed in the second sliding groove, and the outer surface of the fixing shaft is sleeved with a shear fork type lifting frame. The height of the hopper is dynamically adjusted through the shear fork type lifting frame, the height difference of metal waste falling into the waste box can be flexibly shortened or increased, impact force generated by free falling is reduced, the impact load of the waste on the waste box and an internal structure is remarkably reduced, and equipment is effectively prevented from being deformed and damaged.

The invention provides an intelligent putting algorithm optimization method and system applied to a short video streaming platform, and belongs to the technical field of internet advertisement putting, and the method comprises the steps: firstly determining an initial content pool containing short video materials and content tags classified according to themes, and a target user group reaching range covering user behaviors and regional features; then collecting real-time interaction data of the target user group to the initial content pool, generating a content adaptation coefficient reflecting matching association of the material and the user based on the real-time interaction data, and adjusting the initial content pool delivery priority and the target user group reaching range coverage dimension according to the content adaptation coefficient; and finally, circularly executing the operation until a stable delivery strategy is formed and is applied to continuous stream delivery. Therefore, the short video stream delivery accuracy and efficiency can be improved.

The application discloses a kind of robot post-disaster rescue collaboration methods based on deep learning and large language model, by introducing " reflection mechanism " in natural language processing, realize task semantic analysis, dynamic optimization and adaptive scheduling.Large language model generates task semantic graph and combines reinforcement learning to adjust priority in real time, form closed-loop task reflection and optimization process, greatly improve the flexibility and adaptability of robot in complex environment.Design " distributed perception bridge " module, through attention mechanism dynamic screening and sharing key perception data, realize the efficient cooperation of heterogeneous robot, even in the condition of unstable communication still can guarantee rescue efficiency.Based on the reinforcement learning model driven by environmental characteristics, real-time perception of physical properties such as fragment density and slope, optimize behavior strategy, improve the generalization ability and robustness of model.Through " force-semantic coupling " motion planning, dynamically adjust the operation path and force of manipulator, improve the success rate of complex task.

The invention relates to the technical field of production traceability, and particularly discloses a cable production traceability method and system based on big data, and the method comprises the steps: collecting core traceability data of extrusion and vulcanization processes in cable production, dividing a causal disturbance interval through constructing a disturbance conduction chain, and carrying out the dynamic processing of the data into a variable-granularity traceability unit set; based on the traceability unit set, granularity matching analysis is carried out under the scenes of quality traceability, process optimization and responsibility definition, and an adaptive fault of granularity imbalance is positioned; extracting causal features of missing node pairs in the fault, and defining an applicable boundary of the traceability model; and establishing a coupling mapping relation by combining boundary features and granularity-fault features, and dynamically calibrating a granularity adjustment coefficient. The system comprises a unit construction module, a fault extraction module, a boundary defining module and a granularity calibration module. According to the invention, through variable granularity processing and dynamic calibration, the cable production traceability precision and scene adaptability are improved.

The invention discloses a geosteering method based on interactive seismic forward modeling establishment, and relates to the technical field of geosteering in oil-gas exploration and development. The method comprises the steps of firstly establishing a work area and constructing a seismic wave velocity field through an interactive interface, then generating a depth domain seismic section, then interactively drawing a stratigraphic position line and a fault line based on the section, discretizing the stratigraphic position line and the fault line into a space point set with geological attributes, then constructing a triangular mesh model based on the point set, identifying a stratigraphic framework, and finally, constructing a seismic wave velocity field based on the triangular mesh model. Then dynamically estimating shale content and porosity based on measured data of a drilled section, obtaining longitudinal wave velocity through rock physical forward modeling, extrapolating and interpolating the longitudinal wave velocity by combining a stratigraphic framework, establishing a velocity model for seismic forward modeling, and performing seismic wave field forward modeling to obtain a seismic wave field; and finally, carrying out multi-dimensional feature comparison on the forward modeling response and the actual seismic section, if a matching criterion is met, making a guiding strategy, otherwise, reversely adjusting model parameters and carrying out forward modeling iteration again until matching.

The present application relates to the technical field of intelligent control of industrial robots, and particularly relates to a vision transfer mechanical arm dual-arm cooperative assembly control method and system. The method comprises the following steps: collecting the action posture of the mechanical arm dual arms and the relative pose information between the dual arms in real time; identifying the process switching node of the current assembly process stage; constructing the master-slave role dynamic control logic of the dual-arm cooperation, setting the initial motion control parameters of the dual-arm cooperation based on the collected dual-arm posture and relative pose information, and determining the initial master-slave role division of the dual arms. The present application realizes seamless connection of processes, adaptation to various tasks and high-precision stable assembly through posture node identification technology, dynamic adjustment cooperation technology and vision transfer technology, so as to improve the assembly process efficiency, the mechanical arm adaptation ability and the overall assembly quality.

The application discloses a preparation method of a red mud composite phase change heat storage material and relates to the technical field of material preparation process optimization. The method constructs a historical database containing temperature, power and product latent heat, calculates basic heating power based on data quality, collects current heating data in real time, calculates an inertia factor representing heat diffusion capacity and a temperature rise factor representing specific heat capacity characteristics, constructs a thermal response feature vector to quantify the composition similarity between the current batch and the historical record, dynamically determines the real-time target temperature for precise control by screening high-similarity high-quality historical records, and updates the database with new data. The application indirectly reflects the composition difference of red mud raw materials through thermal response characteristics, solves the problem of process window drift caused by composition fluctuation, realizes self-adaptive temperature control without complex composition detection, and significantly improves the phase change latent heat and batch consistency of the product.

The invention relates to the technical field of machine learning, and discloses an image processing method and device, equipment and a medium, and the method comprises the steps: extracting the image features of a target image, recognizing the coding vector of a task node in an image recognition task flow, determining an image reasoning path of a preset image reasoning model according to the coding vector, acquiring the corresponding image quality index of each task node in the image reasoning model, determining the node weight of each task node according to the performance index, and performing target identification processing on the target image according to the image reasoning path to obtain the node output of each node in the image reasoning path, and calculating node loss values of each node output and a preset real label, performing weighted summation to obtain a total loss value, determining a to-be-updated parameter according to the image reasoning path, optimizing the to-be-updated parameter in combination with the total loss and the image features to obtain an updated image reasoning model, and applying the updated image reasoning model to processing of a new image to output an identification result. According to the invention, the efficiency and accuracy of image processing are improved.

This invention discloses an artificial intelligence-based distribution management method, system, device, and medium, relating to the field of intelligent distribution management technology. The method includes: S1, data collection; S2, hierarchical identification of referral relationships; S3, intelligent calculation of multi-dimensional rewards; S4, intelligent calculation of regional profit sharing; S5, integrated reporting of distribution operation data; and S6, intelligent optimization of the distribution mechanism. This invention relies on a compliant two-tier distribution core architecture, combined with scientific hierarchical determination, diversified rewards, and support mechanisms. It avoids the compliance risks of multi-level distribution and, through the "first to get rich, then to get rich later" support logic and fair and reasonable profit distribution rules, fully stimulates the promotional enthusiasm of participants at all levels, promoting the stable and sustainable expansion of the distribution network. By using artificial intelligence technology to accurately process data throughout the entire distribution process, it simplifies the entry and operation processes for participants, eliminates the pressure of inventory buildup, lowers the entry threshold, and reduces participation costs.

This invention discloses a combustion monitoring and evaluation method for an automatic combustion adjustment system of an F-class gas turbine, comprising the following steps: Step 1, acquiring multiple Humming signals and multiple ACC signals from the gas turbine; Step 2, performing a filtering process on each Humming and ACC signal; Step 3, selecting and voting on the results of the first filtering based on set criteria to obtain qualified Humming and ACC signals; Step 4, performing a second filtering process on the qualified Humming and ACC signals to generate Humming control signals and ACC control signals; Step 5, combining the Humming control signals and ACC control signals with NO... X The corrected values ​​are analyzed to obtain combustion evaluation results. This invention can achieve advanced identification of combustion instability and can achieve NO... X Closed-loop control for emission reduction.

The invention discloses a prison abnormal behavior recognition method and system based on edge intelligent analysis, and belongs to the technical field of intelligent security protection, and the method comprises the steps: deploying edge nodes at each monitoring point of a prison, collecting a synchronous monitoring image flow, recognizing a low-light environment image frame through illumination evaluation, and storing the low-light environment image frame in a database; enhancing the image by using an improved Retinex algorithm optimized by a prison scene noise sample library; inputting the clear enhanced image into a lightweight CNN model, performing multi-dimensional matching with a prison abnormal behavior category knowledge base of a three-layer structure through an embedded behavior classification engine, and outputting a classification result of calibrated confidence; when the confidence exceeds a threshold value, triggering graded early warning, and associating the monitoring platform with peripheral sensor synchronous data to form a multi-mode alarm evidence chain; the edge node regularly encrypts and summarizes the early warning data packet and the negative sample data, and feeds back to the training end to realize model incremental learning; according to the method, the low-light environment adaptability and the feature discrimination capability are improved, and the recognition accuracy and the real-time performance are guaranteed.

This invention discloses a real-time voice interaction and incremental backfilling method and system for interrogation transcripts, relating to the field of voice interaction technology. The method includes: acquiring a real-time voice stream and segmenting it into several voice segments; extracting voiceprint feature vectors, performing similarity comparisons, and determining speaker role labels; binding the voice segment corresponding to the stable transcription segment at the end of the sentence with the current target question identifier and the speaker role label to generate a four-tuple; establishing a voice backfilling index; and generating a voice playback control for each incremental backfilling answer field, which, when triggered by the user, plays back the corresponding voice segment according to the index. This invention solves the technical problems in interrogation transcript scenarios where the answer text and the original voice segment cannot be accurately bound, leading to the inability to verify the transcript; the lack of automatic attribution and persistent recording of speaker roles; and the inability to associate newly generated transcripts with new role information after new role registration.