239 results about "Adaptive learning rate" patented technology

The invention discloses a carbon emission prediction and optimization method, and the method comprises the steps: obtaining multi-source heterogeneous data including historical carbon emission data, meteorological data, economic indexes, energy consumption data, and Internet of Things sensor data, and constructing a three-dimensional feature matrix through employing an improved spatial-temporal feature extraction algorithm; based on a mixed architecture of a graph neural network GNN and a long and short term memory network LSTM, a prediction model is established in combination with an attention mechanism, and training is performed through an adaptive learning rate optimization algorithm; a prediction result is input into an improved NSGA-III algorithm, and three targets of total carbon emission, economic cost and social benefits are optimized at the same time; and establishing a feedback closed loop through reinforcement learning RL, and updating the model and the strategy on line according to real-time monitoring data. According to the method, multiple advanced technologies such as accurate data processing, dynamic prediction, multi-objective optimization and cross-domain collaboration are integrated, and a comprehensive and effective solution is provided for carbon emission management.

The invention relates to a supporting structure stress state monitoring method based on artificial intelligence, and belongs to the technical field of artificial intelligence and data processing. The method comprises the following steps: acquiring and marking strain data of a supporting structure; after abnormal values are removed, normalizing the multi-sensor data to generate a normalized strain sequence; a state monitoring model is constructed, a deep time sequence neural network architecture is adopted, and the state monitoring model comprises an input layer, a self-adaptive wavelet attention feature mapping layer, a time domain gating convolution module, a global maximum pooling layer, a dynamic feature importance reweighting layer and a full-connection classification layer; inputting a normalized data training model; optimizing a loss function through a quantile interval adaptive learning rate and a momentum updating strategy; after real-time monitoring data is processed, inputting the data into the training model according to time window slices, outputting four types of probabilities, and taking the maximum value as a prediction state; and if a plurality of continuous windows are early-warning and dangerous, triggering the terminal to give an alarm. The accuracy of monitoring the stress state of the supporting structure can be improved.

The invention provides a zero-set differential privacy self-adaptive federated learning system oriented to Non-IID data. The system comprises a central server and a plurality of clients, the central server maintains a global contribution value list in the federation training process and sends contribution values back to the corresponding clients; according to the contribution values, clients meeting conditions are screened out through a client filtering mechanism and are used for carrying out gradient weighting and generating global model updating; in local model training, each client uses a self-adaptive sliding window mechanism to divide a data set sampling and training range, selects an optimal learning rate through a self-adaptive learning rate selection algorithm, selects a gradient cutting threshold value of each round by using a self-adaptive gradient cutting threshold value mechanism, and performs local model training. And the privacy budget allocated to each round of gradient updating is adaptively adjusted through an adaptive differential privacy federated learning algorithm according to the received contribution value. According to the method, the problem of how to consider both data privacy protection and model precision optimization in a non-independent identically distributed (Non-IID) data environment is effectively solved.

The invention provides a machine language large model construction method and system, and the method comprises the steps: receiving user input data, defining an intelligent semantic association network, and extracting the semantic association between a text and visual or auditory data; a cross-modal attention mechanism and an adversarial generative network are utilized to enhance the multi-modal data relationship in the association, and a multi-modal perception mechanism is formed; based on the mechanism, a multi-level feedback system is constructed, feedback signals from different levels are processed, and an optimized prediction path is generated; key parameters are dynamically adjusted in combination with a task adaptive module and the optimized prediction path, and an adaptive learning rate adjustment strategy is generated; based on an incremental context extension module, an intelligent semantic association network, a multi-modal sensing mechanism, a multi-level feedback system and a task adaptive module, a machine language large model is constructed. According to the method, the understanding and processing capability of the model on complex multi-modal data is improved.

The invention relates to the technical field of intelligent data processing, in particular to a model training and content auditing management system based on intelligent data acquisition, which comprises an acquisition processing unit, a model training unit and a content auditing unit. The model training unit constructs a multi-modal fusion model, a joint training mechanism of the model learns modal association, adaptive learning rate strategy optimization training and early stop technology anti-overfitting, the trained model is used for content auditing, the content auditing unit receives multimedia content, extracts key frames, audio voice content and subtitle text, and performs feature representation and fusion to obtain a subtitle text. Whether violation occurs or not is judged through the trained model, all the units cooperate with one another, the collecting and processing unit provides a data basis for model training, the model training unit provides a precise model for content auditing, effective auditing of multimedia content is jointly achieved, and legitimacy and health of the content are guaranteed.

The invention discloses a complex terrain self-adaptive motion control method and system for a double-wheel-foot robot, and relates to the technical field of robot motion control. The method comprises the following steps: constructing a complex terrain model comprising a robot model, obstacles and environmental constraints; a motion decision controller is constructed, a reinforcement learning model is utilized to perform multi-target collaborative optimization training on the motion decision controller by adopting an asymmetric training strategy for different complex terrain models, and risk constraints are introduced during updating of the asymmetric training strategy to constrain behaviors of the strategy; strategy gradient back propagation is carried out according to a training result, and an asymmetric training strategy is optimized by using a self-adaptive learning rate adjustment method based on performance feedback. According to the method, an asymmetric training strategy and a segmented training mechanism are designed in the reinforcement learning process, and the efficient, robust and self-adaptive motion control problem of the double-wheel-foot robot in various complex terrain environments is solved.

The invention relates to the technical field of sales market prediction, and discloses a sales prediction method and system based on deep learning. The method comprises the steps of collecting historical sales data and external environment data to construct a training data set, and performing multi-dimensional feature extraction to generate a time sequence and a context feature set; constructing a prediction model fusing the deep residual network and the long-short-term memory network, wherein the prediction model comprises a convolution module and a circulation module; the model is trained in stages, initial parameters are optimized through unsupervised pre-training, and then supervised learning is combined with self-adaptive learning rate fine tuning; and predicting real-time input data by using the trained model, and if the confidence coefficient of a prediction result is low, triggering an incremental learning mechanism to update the model. The system comprises a data acquisition module, a feature extraction module, a model construction module and the like. Through multiple technical means, the accuracy, adaptability and reliability of sales prediction are improved, and powerful support is provided for enterprise sales decision.

The invention relates to the technical field of power system monitoring, in particular to an electric energy meter and concentrator data fusion monitoring system and method. Comprising the steps that data of an electric energy meter are periodically collected, integrity verification is conducted on the data, noise interference in the data is removed, then the data are fused to construct a power network diagram structure, and the power network diagram structure comprises node information and edge weight information; a graph convolutional network is used for modeling, a power network graph structure is used as the input of a graph neural network model, an adaptive learning rate adjustment algorithm is combined with an early stop method to prevent overfitting and deep training of the model, and space-time association between nodes is captured through multi-layer space-time graph convolution operation. According to the method, the power network graph structure is constructed by fusing the data of the electric energy meter through the concentrator, nodes and edges are analyzed by using the graph neural network, key nodes and potential fault points are identified, and the network health state is evaluated in real time, so that comprehensive monitoring and optimal management of the power system are realized, and the overall performance and reliability of the power system are improved.

The invention discloses an energy storage system health state assessment method and system, mainly relates to the technical field of health state assessment, and is used for solving the problems that in the prior art, key degradation signals on a frequency domain cannot be identified, an entropy change rule of non-stable change cannot be reflected, and a deep neural network is insufficient in perception of a degradation region. Comprising the following steps: acquiring original monitoring data of the energy storage system in real time, calculating to obtain normalized data after time domain and frequency domain processing, and calculating characteristics of the enhanced monitoring data; constructing a residual attention deep neural network, and autonomously adjusting the current adaptive learning rate; dynamic focus loss is obtained through calculation, and a total loss function is calculated; obtaining a trained residual attention deep neural network; and acquiring current monitoring data, calculating features of the enhanced current monitoring data, and further performing energy storage system health state assessment by using the trained residual attention deep neural network.

The invention belongs to the technical field of asynchronous federated learning, and discloses a data isomerism-oriented knowledge alignment asynchronous federated learning method. According to the method, a data quality perception aggregation strategy is introduced, and a knowledge distillation mechanism based on the old degree is combined, so that a global model is subjected to balanced training on heterogeneous data of different devices, and the generalization ability of the model is improved. Meanwhile, a self-adaptive learning rate adjustment mechanism based on aggregation frequency and weight is designed, and it is ensured that contribution of different devices to the global model is more fair. According to the method, the training deviation in asynchronous federated learning is effectively relieved, the accuracy and stability of a global model are improved, and the method has a considerable application value for a real federated environment.

The invention provides a network protocol fuzz testing method based on strategy gradient reinforcement learning, and relates to the technical field of fuzz testing, and the method comprises the steps: building a test corpus through obtaining a to-be-tested protocol data packet format; establishing an action space containing a mutation operation set and a state space of a benchmark test data packet feature vector; determining a reward value based on the test result; calculating a strategy gradient by using a reward value, and updating strategy network parameters in combination with an adaptive learning rate; and performing fuzzy testing by using the trained model. According to the invention, the test efficiency is improved, the variation strategy selection is optimized, and the effectiveness of the network protocol security test is enhanced.

The invention discloses a packaging printed matter printing quality detection method. The method comprises the following steps: S1, image acquisition; s2, image preprocessing; s3, feature extraction; s4, defect identification and classification; and S5, quality evaluation and report generation. According to the method, single detection limitation is broken through, spectrum, audio, touch and image multi-modal data are synchronously acquired, deep fusion is performed by means of a multi-modal fusion algorithm, feature extraction is optimized for different package types, detection comprehensiveness and accuracy are greatly improved, and model construction is more accurate. Structures and training strategies are customized for conventional paper cigarette cases and special-shaped paper cigarette cases, the adaptive learning rate and data enhancement are adopted, the ability of model generalization and complex feature learning is enhanced, in the defect recognition link, the process is optimized, and through multi-mode cooperation, various defects can be accurately positioned, new and special defect types can be recognized, and the defect recognition efficiency is improved. And the defect classification dimension is enriched.

The invention provides an electromechanical equipment fault prediction method and system based on multi-source information fusion, and the method comprises the steps: collecting operation data, including vibration data, temperature data and current data, during the operation of electromechanical equipment; performing feature extraction on the operation data based on a principal component analysis algorithm to obtain a fusion feature vector; inputting the fusion feature vector into a fault prediction model based on a deep belief network, and outputting a prediction result; wherein the deep belief network adopts a small-batch stochastic gradient descent algorithm combined with an adaptive learning rate adjustment strategy during training; and judging whether the electromechanical equipment has a fault hidden danger or not according to the prediction result. According to the method, the relevance between different types of data is mined, and the defect of low prediction precision is overcome.

The invention discloses a test risk identification method and system based on artificial intelligence, and particularly relates to the technical field of data processing. By introducing an adaptive learning rate and a feedback mechanism, the improved test risk identification method is provided for the problem of data imbalance existing in the test process of an existing machine learning model, and the test risk identification method is provided by comprehensively collecting test data including test case execution results and test environment states. Extracting a dependency relationship abnormal value and a network delay high-frequency value between code modules, constructing a data prediction model, and comprehensively calculating a deviation of a prediction result based on the data; by dynamically adjusting the deviation of the model to the successful use case, the identification capability of the failed use case and the high-risk area is enhanced, and the high-risk area is effectively prevented from being underestimated or the normal use case is misjudged to be high in risk, so that the prediction precision is improved, the configuration of test resources is optimized, the invalid test reexamination is reduced, the test efficiency is remarkably improved, and the cost and the period are reduced.

The invention discloses an electric heat storage multi-energy cooperative consistent energy flow intelligent control method and system, and relates to the technical field of electric power system intelligent control, and the method comprises the steps: carrying out the multi-energy system state modeling of an electric system, a heat energy system and an energy storage system; an adaptive deep Q network architecture is introduced, an evolutionary strategy is optimized, and local optimal traps are reduced; designing a dynamic weight adjustment mechanism, and dynamically adjusting each index in the reward function according to different energy flow cooperative control states; and dynamically correcting control parameters through a closed-loop correction mechanism fed back in real time. According to the invention, rapid convergence is realized under the condition of system load fluctuation, and peak-valley difference adjustment and voltage adjustment can be completed within 200 seconds. Through an energy flow consistency optimization model, the consumption rate of distributed energy such as photovoltaic energy is improved. Through an adaptive learning rate and a dynamic weight adjustment mechanism, the response speed of the system is improved, and power fluctuation is remarkably reduced. The dynamic adjustment of the control strategy enhances the stability and power supply quality of the power system.

The invention provides a joint point detection dynamic model algorithm for racing boat movement, and relates to the technical field of computer vision and movement posture detection, and the algorithm selects a YoloBoat model to carry out preprocessing on racing boat movement image data, and comprises normalized and diversified data enhancement operations. A backbone network of the model adopts CSPDarknet53 to extract features, a feature fusion layer adopts PANet to optimize the features, and an output layer generates a detection result. The key point track is smoothed through spline interpolation, and optimization and improvement such as confidence coefficient filtering are carried out. A data set is constructed by using about 1000 own pictures, key points are manually marked, and the data set is divided into a training set, a test set and a verification set in proportion. In training, an adaptive learning rate and a plurality of data enhancement methods are adopted. And through comparison with OpenPose and other models, the performance of the model is evaluated through multiple indexes. The method is high in detection precision, high in reasoning speed and high in robustness in close-range and long-range scenes, and can cope with various complex scenes. Accurate and real-time joint point detection data are provided for boat racing training and competition.

The invention provides a flexible power distribution network intelligent management and control method based on cloud side-end cooperation, and belongs to the technical field of intelligent power grids. An intelligent decision-making model is constructed and issued to an edge server through the cloud, the power grid state is monitored, and a management and control strategy is formulated. The edge servers exchange strategy information and work cooperatively. The sending end edge server computes policy utility based on multi-edge collaboration. And if the strategy is rejected, the sending end edge server updates the intelligent regulation and control judgment and decision model parameters based on the abnormal operation index probability. And if the strategy is accepted and the strategy utility exceeds the utility threshold, the edge server updates the model according to the operation condition of each regional power grid. If the strategy is rejected for multiple times or does not meet the multi-threshold stability requirement, the cloud side uses a ternary loss function containing multiple deviation factors to update the model, and the self-adaptive learning rate and the ternary loss function are introduced by judging the multi-domain stability requirement. According to the invention, the real-time performance and accuracy of decision making are improved.

The invention discloses a method and device for achieving accurate lifting of a steel plate through automatic magnetism adjustment, and belongs to the technical field of steel plate lifting control. The method comprises the steps that core parameters of an electromagnet of an electromagnet lifting appliance and observation data in the historical lifting process are collected; the observation data in the historical hoisting process are preprocessed; constructing a reinforcement learning model, and training the constructed reinforcement learning model by using the preprocessed observation data in the historical hoisting process; wherein the reinforcement learning model adopts a self-adaptive learning rate and updating frequency; and the specification and number of the current to-be-hoisted steel plates are obtained, the optimal current of the electromagnet is obtained through the trained reinforcement learning model, and steel plate hoisting is achieved. According to the technical scheme, the electromagnetic attraction can be intelligently and accurately controlled, and the hoisting efficiency and precision are improved.

The invention discloses an industrial chain situation prediction analysis system and method based on AIGC, and belongs to the field of data processing systems or methods specially suitable for the purposes of commerce, finance, management, supervision or prediction. An AIGC model based on the combination of a generative adversarial network (GAN) and a variational auto-encoder (VAE) is constructed; through adversarial training, KL divergence optimization and adaptive learning rate adjustment, a data rule is accurately captured, a moving average method, an autoregressive integral moving average model (ARIMA) and a random forest algorithm in situation prediction are cooperated, prediction stability and accuracy are enhanced, traditional defects are improved, in the aspect of analysis and management, polynomial fitting interpretation trend is deeper, and the prediction efficiency is improved. Risks and opportunity points are accurately identified through a comprehensive index system and a scientific method, decision making is assisted, on the whole, the industrial chain situation prediction analysis process is comprehensively innovated, scientific industrial development decision making is powerfully promoted, and enterprise competitiveness and industrial resource allocation efficiency are improved.

A computing system and method can be used to implement a version of federated learning (FL) that incorporates adaptivity (e.g., leverages an adaptive learning rate). In particular, the present disclosure provides a general optimization framework in which (1) clients perform multiple epochs of training using a client optimizer to minimize loss on their local data and (2) a server system updates its global model by applying a gradient-based server optimizer to the average of the clients' model updates. This framework can seamlessly incorporate adaptivity by using adaptive optimizers as client and / or server optimizers. Building upon this general framework, the present disclosure also provides example specific adaptive optimization techniques for FL which use per-coordinate methods as server optimizers. By focusing on adaptive server optimization, the use of adaptive learning rates is enabled without increase in client storage or communication costs and compatibility with cross-device FL can be ensured.

The invention discloses a generative adversarial network architecture search method and system and an image generation method, and belongs to the technical field of network architecture search. The searching method comprises the following steps: performing single-path sampling on a pre-constructed generator super network according to a parameter quantity constraint range to obtain an effective subnetwork; training the generator super-net by adopting a complexity adaptive learning rate optimization strategy to obtain a pre-trained generator super-net; adversarial training is carried out on the generator hypernet and the discriminator to obtain a pre-trained discriminator; generating a generator super-network candidate architecture through a genetic algorithm in the early stage of the evolution stage and through a covariance matrix self-adaptive evolution strategy in the later stage of the evolution stage; and performing multi-target non-dominated sorting on the generator super-network candidate architecture, updating an effective sub-network and keeping a Pareto optimal solution to obtain a searched optimal generator architecture and further obtain a searched optimal generative adversarial network architecture. The method not only ensures the search quality, but also improves the calculation efficiency.

The invention relates to the technical field of radar weak target detection, in particular to a radar weak target detection method and device based on deep learning. According to the technical scheme, the method comprises the following steps: data preprocessing and enhancement: receiving an original echo signal of a radar, generating three-dimensional time frequency-wavelet coefficient spatial data by combining time frequency analysis with wavelet transform, and expanding a data set in multiple modes; and feature extraction network construction: constructing a 3D-CNN and LSTM fused network, wherein the 3D-CNN has a convolution operation with a variable convolution kernel size. According to the method, the data are richer and more comprehensive through time-frequency analysis in combination with wavelet transform and data enhancement, and multi-dimensional features can be accurately captured through 3D-CNN and LSTM fusion and an adaptive convolution kernel; and the training effect is ensured through the adaptive learning rate and the early stop method, and the accuracy and adaptability of weak target detection are jointly improved.

The invention discloses a power transmission line risk monitoring method and device based on a learning rate strategy, electronic equipment and a storage medium, and belongs to the technical field of power transmission line risk monitoring. The method comprises the following steps: acquiring system operation information of a power system, wherein the system operation information comprises real-time meteorological data and line operation data; extracting real-time meteorological characteristics from the real-time meteorological data, extracting line operation characteristics from the line operation data, and constructing vector representation to be measured based on the real-time meteorological characteristics and the line operation characteristics; inputting the to-be-measured vector representation into a risk assessment model to obtain a risk assessment result output by the risk assessment model; the risk assessment model is obtained by training a logistic regression model based on a plurality of training samples and updating a self-adaptive learning rate, and the training samples comprise sample vector representations and risk assessment labels corresponding to the sample vector representations. The method can improve the assessment precision of the power transmission line risk, and guarantees the safe and stable operation of a power system.

The invention discloses an automatic history fitting method based on multiple data assimilation of an improved set smoother, and relates to the technical field of petroleum engineering. The method comprises the following steps: firstly, constructing a plurality of oil reservoir models, setting a prior model, a real model and an initial expansion factor, obtaining the yield of each oil reservoir model by utilizing oil reservoir model numerical simulation, and calculating a residual error; based on corrected data covariance matrix singular value decomposition, production observation data disturbance enhancement, adaptive learning rate matrix scaling and geological boundary constraint, improving a set smoother, updating the permeability of each oil reservoir model, performing numerical simulation again, updating an expansion factor, judging whether the expansion factor meets a preset condition or not, continuing iteration if the expansion factor meets the preset condition, and if the expansion factor does not meet the preset condition, continuing iteration until the expansion factor meets the preset condition; and if not, updating the expansion factor of the iteration, ending the iteration, and outputting the permeability field of each updated oil reservoir model, thereby solving the problems of parameter overshoot, covariance statistical deviation and low calculation efficiency in oil reservoir history fitting, and facilitating history fitting of complex oil reservoir production parameters.

The invention discloses an internet advertisement accurate putting method and system based on big data, relates to the technical field of internet advertisements, and is used for solving the problem that the accuracy and dynamic adaptability of interest prediction are reduced due to the fact that advanced algorithms such as deep learning or reinforcement learning are not fully adopted by a model. User interaction features are captured in real time through a dynamic behavior analysis module, multi-dimensional data modeling is performed in combination with deep learning, and user interest prediction and real-time adjustment of an advertisement recommendation strategy are achieved. The dynamic behavior feature vector is fused with the time sequence and the spatial distribution feature, so that the prediction accuracy is improved; a multi-layer perceptron and a self-adaptive learning rate optimization algorithm are adopted to train a model, and generalization and response capabilities are enhanced; an advertisement adaptation degree score is calculated in combination with semantic similarity, and recommendation correlation is ensured; rapid feedback and strategy optimization are achieved through incremental learning, the defects of the prior art in the aspects of real-time performance and automation are overcome, and the advertisement putting efficiency and accuracy are improved.

The invention relates to the technical field of power equipment state operation early warning, and discloses a power equipment state operation time sequence early warning system and method, and the system comprises a data collection module which collects the multi-source operation data of power equipment through infrared, ultrasonic TEV, high-frequency, ultrahigh-frequency and vibration sensors; the data preprocessing module is used for carrying out normalization processing on the collected multi-source operation data by adopting an extreme value method or an averaging method and mapping the data to a [-1, 1] interval; and the time sequence analysis modeling module is used for constructing a Prophet time sequence model for the normalized data, and decomposing a trend item, a season item and a holiday item. A dynamic factor is added in weight updating through a dynamic parameter adjustment method, network training is effectively prevented from falling into a local minimum point, fluctuation in the learning period is reduced, the learning rate is flexibly adjusted according to error changes in combination with an adaptive learning rate algorithm, and slow convergence caused by the too low learning rate is avoided.

The invention belongs to the technical field of computers and information science, and relates to a power network asset key feature extraction method and device, equipment and a medium. According to the method, an improved k-means clustering algorithm is adopted to perform clustering analysis on power network asset data to obtain key features of power network assets; wherein the k-means clustering algorithm is improved in that in the clustering process, the mahalanobis distance between each data point and each clustering center is calculated, and the data points are distributed to the clustering center with the shortest distance; updating a clustering center based on dynamic weight updating and an adaptive learning rate; performing feature weighted summation and nonlinear transformation processing on the key features obtained by clustering by using an attention mechanism to obtain an optimized feature vector matrix; and powerful support is provided for subsequent asset management, fault diagnosis and predictive maintenance.

The present invention discloses a tightening quality assessment method based on deep learning, which relates to the technical field of tightening quality assessment. The present invention collects torque, preload, lubrication status and environmental data in real time during the tightening process, uses an LSTM dynamic model to capture the nonlinear fluctuation of the friction factor during repeated tightening, and promptly corrects the torque-preload conversion relationship. Adaptive learning rate and fully connected layer mapping are adopted to enable the model to automatically update parameters under different environmental conditions, dynamically adjust the tightening speed, preset torque and lubrication strategy. The environmental integration module quantifies changes in temperature, humidity and salt concentration, corrects process parameters in real time, and effectively suppresses instantaneous abnormal fluctuations.

The invention discloses a point cloud single tree segmentation method and system. The method comprises the steps of obtaining point cloud data of a target tree and an environment in a power transmission corridor area; constructing a segmentation network model, and performing model training through the optimized loss function and adjustment of the adaptive learning rate to obtain an optimized segmentation network model; inputting the preprocessed point cloud data into the optimized segmentation network model, calculating the category of each point through an activation function, and outputting the category prediction of each point; and point cloud segmentation is carried out based on the category, and a segmentation result is evaluated. According to the invention, point cloud data segmentation is carried out through the segmentation network architecture, the learning rate and the loss function in the architecture are optimized, and the recognition and segmentation precision of the tree monomers is improved in combination with the probability distribution of deep learning; the robustness in a complex environment is enhanced, the problems of noise, sparse point clouds and data imbalance can be effectively solved, and the method is suitable for tree monitoring, accurate positioning and safety evaluation of the transmission line corridor.

The invention relates to the technical field of artificial intelligence and natural language processing, in particular to an online incremental learning AI chat robot response generation system which comprises a cache reweighting module, a semantic offset calculation module, an adjusting module, an updating module, a response generation module and a feedback module. According to the method, cache reweighting based on semantic similarity and time decay, a self-adaptive learning rate based on semantic offset and rollback protection are introduced, low-rank increment updating of LoRA and a weight adjustment closed loop of multi-target weighted decoding and feedback driving are adopted, so that the system can effectively improve the real-time performance of the system while preferentially utilizing aging and related historical contexts, and the real-time performance of the system is improved. And the updating strength is dynamically controlled, and stable parameters are recovered during abnormal offset, so that the problem that the response reliability is reduced due to model updating delay or parameter offset during high-frequency interaction or semantic drift caused by conflict between speed and stability in online incremental updating is effectively solved.