42 results about "Training evaluation" patented technology

This application discloses a method and apparatus for constructing a multi-scale adaptive gating MambaPlus network, belonging to the field of artificial intelligence and machine learning technology. The method includes: initializing the network configuration and constructing the basic structure; preprocessing the input data to generate a standard dataset; mapping the input data to the hidden space via an input mapping layer, and extracting backbone features from the Mamba backbone; constructing at least two parallel scale branches in the hidden space to obtain a multi-scale feature set; inputting the backbone features and multi-scale features into an adaptive gating module, dynamically allocating weights and adaptively fusing them through a hierarchical gating mechanism to generate fused features; further enhancing the features through cross-scale attention and feedforward enhancement, and then superimposing the residuals to generate the final discriminative features; finally, completing category prediction and model training evaluation. This application, while retaining the advantages of Mamba's long-range dependency modeling, addresses the problems of insufficient utilization of multi-scale information, poor adaptive feature fusion, and low robustness in complex scenarios.

This application discloses a training method, device, electronic device, and storage medium based on a large language model. The method includes pre-configuring training modes, which at least include open-ended training and restricted training. The open-ended training and restricted training can run independently or be switched on demand. Responding to the input of the user to be trained, the method selects the training mode and simultaneously conducts training based on a fallback process, outputting the final training evaluation result. The fallback process ensures the continuity of the large language model training process. This application provides a training method including open-ended training, process tree restricted training, and a fallback process. The open-ended training integrates a knowledge base and a large evaluation model; the process tree restricted training includes configurable nodes and eight core functions; and the fallback process has model monitoring, self-interruption, and dialogue switching capabilities, achieving integrated training with flexible interaction, standardized control, and stable operation.

This invention discloses a method and system for parallel training, evaluation, and control based on a digital twin using an equipment bus. It unifies the time-stamping and state-space mapping of multi-bus observation data, and solves for a unified state estimate based on quality indicators, fusion residuals, and cross-bus physical consistency constraints. This drives the state evolution of the digital twin, executing the Lyapunov descent admission condition under synchronization error and end-to-end delay constraints. Further, multi-scenario parallel simulations are conducted to obtain an upper bound on the risk probability and construct a training evaluation function. When the evaluation is below a threshold or the risk exceeds the limit, only the affected control components are subject to safe feasible region projection correction, and closed-loop control commands are output. This invention improves state estimation accuracy, synchronization stability, risk prediction capability, and training control safety.

The application discloses a kind of based on patrol trolley robot police training intelligent tracking evaluation method and system, it is related to intelligent monitoring technical field.The method includes: demarcate training area and robot activity area in training site, and patrol trolley robot is based on laser radar and dynamic window method (DWA) in activity area autonomous movement and real-time obstacle avoidance;Through the AI vision module and laser radar carried, the preset training team is accurately identified and locked, and the pan-tilt and the moving chassis are controlled to realize stable follow-up shooting, and training video is collected in real time.Further, based on the multi-angle video shot, using the action recognition algorithm based on skeleton key point detection, the team action specification of training team and the action specification of individual student are intelligently evaluated, and the evaluation report containing quantitative score, deviation positioning and rectification suggestion is automatically output.The application solves the problem that traditional police training evaluation relies on manual, strong subjectivity and low efficiency, realizes the automation tracking of training process, multi-angle recording and intelligent quantitative evaluation.

The application provides a pilot attention allocation training method, device and equipment and a storage medium, and belongs to the field of flight training. The method comprises the following steps: in the case that a pilot wears a head-mounted eye tracker for training, obtaining a fixation point video of the pilot and recording the triggering conditions of observable items; dividing the fixation point video into interest areas, performing data analysis on the interest areas, and determining the access times, Pearson correlation coefficients, pupil diameter change scatter plots, Markov transition matrices, access sequences and time lengths, in-cabin and out-of-cabin access times, cockpit saccade mode models and the pupil diameter change rates of the pilot; further determining the attention transfer portrait and the attention allocation score of the pilot; and adjusting the attention allocation training bias of the pilot according to the attention allocation score and the attention transfer portrait. In this way, the limitations of traditional subjective judgment can be overcome, and the training evaluation is more scientific and operable.

The application discloses a kind of dystonia evaluation system and method based on wearable inertial sensor.The system includes: data acquisition module, including wearable inertial sensor and portable muscle strength tester, for the synchronous acquisition of subject's multidimensional motion data and muscle strength data;Data preprocessing module is used to optimize the data collected;Characteristic processing module is used to extract multidimensional features from the processed data, and then complete feature dimension reduction through principal component analysis to obtain a core feature subset;Evaluation model module is based on machine learning algorithm Training evaluation model, the core feature subset is used as model input, to realize the accurate classification of dystonia type;Result output module is used to generate quantitative evaluation report, to provide objective data support for clinical diagnosis and treatment plan formulation.The application solves the problem that traditional evaluation relies on subjective experience and lacks quantitative standard, and improves the accuracy and standardization of dystonia evaluation.

The application relates to the technical field of massage training, and discloses a pediatric traditional Chinese medicine acupoint massage training system, which comprises a nursing terminal interaction module, a training data acquisition module, an intelligent evaluation module and a management terminal interaction module. The real-time adjustment amount is generated by corresponding training standards of the training case selected by nursing staff and the massage data collected when the nursing staff is trained, the real-time adjustment amount is used for real-time reminding and correction of wrong methods, the accuracy of acupoint massage is improved, the efficiency of learning massage of the nursing staff is improved, the comprehensive evaluation coefficient is obtained by evaluating the massage data when the nursing staff is trained, the subjectivity of the traditional evaluation mode is avoided, the objectivity of the training evaluation is improved, the nursing staff is rated and the rating is predicted according to the comprehensive evaluation coefficient, the training report of the massage training process of the nursing staff is generated, and the training report is checked by a management doctor and the nursing staff.

The application discloses a kind of unmanned plane simulation training method and system based on simulation model full task scene, belong to simulation training technical field.This technology includes a kind of unmanned plane simulation training system and method, system includes man-machine interaction unit, high-precision simulation model unit, full task scene construction unit, real-time deduction unit, training evaluation unit and system management unit;Method includes the steps of constructing high-precision unmanned plane simulation model, constructing full task training scene, receiving operation instruction, real-time calculation state change, updating scene and display, training evaluation, generating training report etc..The application accurately simulates unmanned plane characteristics by high-precision simulation model, supports full task scene construction and real-time deduction, provides real-time interaction and multidimensional feedback, realizes intelligent evaluation and targeted guidance, with the characteristics of high authenticity, high integrity, high interactivity and high adaptability, effectively improves the training effect and training efficiency of unmanned plane operator.

This invention discloses a practical single-pulse radar training simulator architecture, employing a three-layer structure design consisting of a user layer, a middleware layer, and an execution layer. The user layer, based on a B / S architecture, handles human-computer interaction, training processes, and scenario configuration. It pre-generates and distributes target trajectory-related simulation parameters and pre-set information for abnormal events, while simultaneously collecting operational information and device status and conducting training evaluation. The middleware layer is an embedded control platform used to acquire servo differential angle information in real time, calculate angular error, attenuation, phase, and Doppler echo control parameters, as well as abnormal control quantities, according to the PRF beat. The execution layer is a DRFM simulation sub-unit that receives radar transmitted pulses and applies time delay, Doppler, and amplitude-phase modulation based on control parameters, generating the main target, debris, co-frequency station, and noise echoes for output to the trained radar. This invention achieves a balance between complexity and real-time performance through functional decoupling and separation of responsibilities, supporting high-fidelity training under multiple targets and multiple abnormal conditions.

This invention discloses a method and apparatus for joint virtual-real simulation training of an unmanned intelligent system. The method includes: firstly, acquiring a set of physical entity equipment configuration data, training task configuration data, and training parameter configuration data; then, performing modeling and scene generation processing on the set of physical entity equipment configuration data, the training task configuration data, and the training parameter configuration data to obtain a set of digital equipment model data and virtual training scene data; next, performing joint virtual-real training processing on the set of digital equipment model data and the virtual training scene data to obtain a set of training process data; and finally, evaluating and performing closed-loop optimization processing on the set of training process data to obtain a target training evaluation report and optimized training parameter configuration data.

The application relates to the technical field of intelligent agent cooperation and discloses a practical training behavior analysis method based on multi-intelligent agent cooperation, which comprises the following steps: first, performing time sequence alignment and aggregation on multi-source practical training data to generate a time sequence sequence containing scene states and operation instructions; using a behavior recognition intelligent agent to establish a scene evolution model, performing multi-step deduction on a virtual environment state based on the sequence; in view of a preset reference instruction set, evaluating the reachability of the instruction set relative to risk constraints and task targets in the deduced state, and calculating a decay index reflecting the degree of recoverable space contraction; then, using a target matching intelligent agent to calculate task completion degree which is deeply coupled with the decay index, and combining resource consumption data of the practical training process to generate a comprehensive score; finally, mapping the comprehensive score to a qualified or unqualified evaluation result based on an offline calibrated qualified benchmark value. The application can accurately identify high-risk operation trends and realize the standardization of practical training evaluation.

This invention relates to the technical field of simulation training, and specifically to a real-time simulation training system for equipment operation posture based on the fusion of a three-axis gimbal and an IMU. The system is applied to a training platform composed of a three-axis gimbal and a six-axis IMU sensor, and includes: a multi-source data acquisition module for receiving first sensing data, wherein the first sensing data includes angular velocity data from the three-axis gimbal and acceleration data from the six-axis IMU sensor; an operation mirroring module for determining mirrored operation behaviors of the equipment based on the first sensing data, wherein the mirrored operation behaviors characterize the operator's posture; and a virtual synchronization output module for generating driving instructions for a virtual training scenario based on the mirrored operation behaviors, thereby generating synchronized operation postures in the simulation training space and outputting training data; wherein the simulation training space includes preset standard operation behaviors, and the training data includes training evaluation data based on the preset standard operation behaviors.

This application proposes a simulation training method, system, electronic device, and storage medium, which includes: acquiring a preset geographic mapping model, fixed virtual geographic coordinates of a virtual reference station, and satellite ephemeris data; real-time acquisition of physical coordinate data fed back by the mobile platform; converting the physical coordinate data according to the geographic mapping model to obtain the virtual geographic coordinates of the mobile station; calculating the first satellite observation data using the fixed virtual geographic coordinates based on a unified time reference and precise satellite ephemeris data, and calculating the second satellite observation data using the virtual geographic coordinates of the mobile station; generating differential correction data based on the first satellite observation data; acquiring the positioning result output by the receiver of the tested mobile station; calculating the deviation index between the positioning result and the virtual geographic coordinates of the mobile station, and generating a training evaluation report.

This invention proposes a method, apparatus, equipment, and storage medium for simulation training of ship section construction. It achieves the construction of a high-performance virtual reality simulation environment, supporting collaborative interaction across multiple terminals such as VR, PC, and CAVE, significantly improving the system's adaptability and interactive immersion. Through a multi-user network synchronization architecture and workstation-level dynamic simulation scene construction, it realizes the realistic restoration of section construction process logic and operational synchronization, improving collaboration and training realism. Employing a multi-level detail model and dynamic mesh optimization technology based on view distance and bounding box overlap rate effectively reduces the rendering load of complex scenes and improves the system's real-time response performance. By establishing a multi-projection plane world coordinate system and real-time viewpoint matrix calculation, it achieves consistent rendering across multiple terminals. Combined with workstation cycle data synchronization and an operation scoring mechanism, it forms a quantifiable training evaluation system, thereby comprehensively improving the efficiency, safety, and scientific rigor of ship section construction training.

This invention discloses a radar countermeasure simulation training method and system based on a multimodal large model, belonging to the field of radar simulation training technology. The method includes: receiving training requirements input by trainees in natural language, parsing and extracting core elements, and initializing radar and target parameters. Based on these requirements, it generates complete electromagnetic environment scenarios, multi-system radar signals conforming to tactical situations, synthesizes waveforms, simulates complex environmental effects, and outputs standardized files. A real-time simulation engine unit simulates radar power-on and search scanning, generating target echoes. An intelligent countermeasure strategy unit perceives the situation, intercepts signals, identifies threats, and generates adaptive countermeasure strategies. An intelligent evaluation and feedback unit collects data throughout the process, quantifies and scores data through a large model, checks compliance, and generates training evaluation results in real time. This invention solves the problem in existing technologies where radar countermeasure training is difficult to adapt to complex and ever-changing electromagnetic countermeasure requirements.

This invention discloses a real-time guidance method for pelvic floor muscle exercises based on multi-parameter sensing using an abdominal belt, belonging to the field of physiological monitoring technology. The method includes: constructing a set of training rhythm windows based on respiratory phase sequences and training quality sequences; calculating the gating confidence of each training rhythm window in the set; generating a real-time guidance sequence; performing anomaly detection processing on the real-time guidance sequence to generate an anomaly prompt sequence; and simultaneously extracting training evaluation indicators for each training rhythm window to generate a training score record. This invention, by generating abdominal pressure change sequences and real-time guidance sequences, achieves collaborative analysis of the relationship between respiratory rhythm and abdominal pressure changes based on multi-source signals such as abdominal wall displacement and abdominal belt tension collected by an intelligent abdominal belt sensor, thereby improving the real-time guidance capability and training effect evaluation capability of the pelvic floor muscle training process.

The application discloses a kind of intelligent sports comprehensive training platform systems based on behavior characteristic identification, it is related to sports training, characteristic identification technical field, it includes multi-source data acquisition, data preprocessing, feature extraction, feature fusion and evaluation and risk intervention module.System obtains movement and pressure data by visual sensor matrix and flexible pressure sensing unit, extracts and fuses geometric posture, barycenter trajectory and muscle power feature using inverse dynamics algorithm and Transformer architecture, and constructs biomechanics constraint model.The application can deeply perspective the internal power mechanism of the trainee, accurately identify the barycenter deviation and joint compensation phenomenon, realize the active intervention of sports injury risk and the quantitative evaluation of energy transmission efficiency, significantly improve the objectivity of training evaluation and the scientificity of training, provide digital closed-loop management and personalized guidance scheme for physical education.

The application provides a ship electromechanical system operation training evaluation method and device based on a three-dimensional environment, and relates to the field of ship electromechanical system simulation. The method comprises the following steps: obtaining a simulation data set corresponding to a ship electromechanical system in a three-dimensional virtual environment; obtaining operation interaction data in different types of ship training consoles; preprocessing the simulation data set, and constructing a plurality of evaluation indexes based on the preprocessed simulation data set; inputting the operation interaction data into the multi-evaluation index model to output an index score corresponding to each evaluation index; inputting the index score into a comprehensive training evaluation model to output a training quality score corresponding to a target trainee; and outputting a training quality evaluation result of the target trainee based on the training quality score. The application solves the problem that the traditional ship operation training evaluation system lacks a coupling computer mechanism between operation behavior and physical simulation response, so that the evaluation result cannot reflect the comprehensive ability of the trainee to control the system.

The application provides an aeration quantity prediction method and system based on MSTP-Net, and relates to the technical field of sewage treatment. The method comprises the following steps: obtaining historical operation data and performing feature selection based on correlation analysis to screen out a key feature subset; a multi-scale time sequence feature extraction method is used to capture short-term fluctuations and long-term dependence features in parallel and to fuse them; time sequence features are used as training inputs, and a time sequence cross-validation strategy is used to train and optimize the hyperparameters of MSTP-Net; and the optimized model is used to output aeration quantity prediction values at multiple future time points by using an Informer prediction method. The application solves the problems that existing models cannot simultaneously consider short-term fluctuations and long-term dependence and that training evaluation has a forward-looking bias by constructing a MSTP-Net hybrid architecture, fusing multi-scale feature extraction and efficient long sequence prediction, and combining time sequence cross-validation, realizes high-precision, multi-time-scale aeration quantity prediction, significantly reduces prediction errors, and enhances the robustness of the model.

A phased array antenna thermal deformation error compensation method based on small sample data machine learning modeling adopts ground high-precision modeling, on-orbit temperature and typical position sensing, and periodic ground calibration to obtain small sample data, and through data modeling and correction, data preprocessing, model construction and training evaluation, a high-precision thermal deformation error compensation model is established based on small sample data, the self-attention mechanism of the model is used to improve the generalization ability and training efficiency of the deep learning model, an intelligent model for phased array antenna thermal deformation error compensation is established, real-time on-orbit working condition parameters including working mode, duty cycle and temperature information are collected as input parameters of the intelligent model, the beam controller issues control instructions according to task requirements, calls the intelligent model according to compensation instruction requirements, loads it to the active channel of the phased array antenna, and adjusts the amplitude and phase in real time to offset the error caused by thermal deformation.

This invention provides an interactive sports training assessment method and system, belonging to the field of sports training assessment technology. The method involves: distributing an assessment item set in a virtual training assessment scenario; sorting the items according to the trainee's exercise habits and psychological state for each assessment item; sequentially displaying the items and capturing training behavior posture sets; globally comparing these with a standard posture set to construct joint point abnormal posture vectors; and combining joint point behavior trajectories with standard trajectories to construct point abnormal vectors; combining and analyzing these results to obtain single-item assessment results; and finally, processing all results and sending them to the assessment terminal. This invention achieves standardized quantification of assessments, improves the objectivity and accuracy of assessments, and provides a scientific reference for sports training.

The application provides a cricothyroid membrane puncture rescue simulation training method and system, a cricothyroid membrane puncture rescue simulation training method comprising the following steps: step S1, starting an infrared locator, a pressure-sensitive puncture needle and a head-mounted display, and establishing a connection; step S2, calibrating the head-mounted display coordinates; step S3, the infrared locator determines the positions of the pressure puncture needle, the head-mounted display and the elastic simulation human body, and transmits the position information to the head-mounted display; step S4, the head-mounted display updates the position according to the received position information; step S5, the head-mounted display renders the training scene and produces environmental sound; step S6, the training personnel perform a cricothyroid membrane puncture action; step S7, the contact module in the head-mounted display judges whether the puncture reaches the specified position; step S8, when the training personnel puncture to the specified position, the pressure puncture needle is fixed, and training evaluation is prompted and given; and step S9, the training is completed. The beneficial effect is that the cost of training construction is controlled, and more realistic feedback experience can be given to students.

The application discloses a multi-scale adaptive gating MambaPlus network construction method and device, belongs to the technical field of artificial intelligence and machine learning, and comprises the following steps: initializing network configuration and constructing a basic structure; pre-processing input data to generate a standard data set; mapping to a hidden space through an input mapping layer, extracting backbone features from a Mamba main body; constructing at least two parallel scale branches in the hidden space to obtain a multi-scale feature set; inputting the backbone features and the multi-scale features into an adaptive gating module, dynamically distributing weights and adaptively fusing through a hierarchical gating mechanism to generate fused features; then, cross-scale attention enhancement and feedforward enhancement are performed, and residual superposition is performed to generate final discriminative features; and finally, class prediction and model training evaluation are completed. While retaining the advantages of Mamba long-range dependence modeling, the application solves the problems of insufficient multi-scale information utilization, poor feature fusion adaptability and low robustness in complex scenes.

The application discloses a kind of short-term memory of hearing quantification evaluation method based on snare drum rhythm imitation training, the test system comprising host computer, PPG sensor and electronic drum is first constructed, and the score of the short-term memory of hearing of the subject is obtained as supervision label by DMTS;Snare drum rhythm imitation training is then executed, and the average piece of time-consuming, training times and other behavior data are recorded synchronously, and resting pulse wave signal is collected before and after training;IBI sequence is constructed by band-pass filtering, valley detection and other pretreatments to pulse wave signal, and the difference features before and after training are extracted by extracting pulse rate variability features;Finally, the supervised learning regression model is constructed by fusing physiological features and training behavior features, the non-linear characteristics are fitted by smoothing function, the overfitting is inhibited by regularization constraint, and the objective quantification result of the short-term memory of hearing of new subject is output.The application realizes the integration of measurement and training evaluation, improves the objectivity, stability and repeatability of evaluation, and provides a systematic and deployable technical solution.

This invention discloses a desktop table tennis shot quality quantification training device and method, belonging to the technical field of sports training equipment. The device includes a fixed base, a shot response component, a measurement module, a data processing module, and a data feedback module. The measurement module includes at least one of a spin measurement submodule and a velocity measurement submodule, preferably both; the spin measurement submodule collects shot spin parameters, and the velocity measurement submodule collects velocity parameters of the moving component of the moving subunit. The data processing module simultaneously collects, aligns, and fuses the two types of parameters within a sampling window to generate shot quality quantification data, which is output by the data feedback module. This invention also provides a scheme containing only a single submodule and discloses the corresponding quantification method, training data feedback system, and computer-readable storage medium, which can simultaneously acquire spin and velocity parameters, output objective quantitative feedback, and improve the accuracy of training evaluation.

This invention discloses a method for analyzing table tennis training movements based on multi-source information fusion. It employs at least two video sources, a nine-axis posture sensor at the trainee's wrist, and an audio source to acquire training process information. After unified time synchronization, frame alignment, and preprocessing, a YOLO-based detection and keypoint extraction module outputs bounding boxes for the trainee, racket, and table tennis ball, as well as keypoints on the human body. A nine-axis posture analysis module outputs posture sequences and motion features, while an audio event detection module outputs peak values ​​for ball impact and table contact. A multi-source feature matrix is ​​then constructed and input into an improved TriDet temporal motion localization model to obtain the motion category, start time, end time, and confidence level. Finally, the method outputs training movement analysis results, comprehensive evaluation indicators, and training suggestions. This approach improves the accuracy of motion boundary localization, the accuracy of ball impact recognition, and the comprehensiveness of training evaluation, and is suitable for edge deployment.