75 results about "Domain prediction" patented technology

The invention discloses a vehicle fine granularity detection method based on a three-dimensional grid and YOLOv11 transfer learning. The method mainly comprises three components: a source domain prediction network structure, a target domain prediction network structure and a transfer learning module. Wherein the source domain prediction network and the target domain prediction network are dual-channel deep networks fusing two-dimensional images and three-dimensional grids, and efficient migration of source domain knowledge in a target domain is realized by aligning feature distribution of the source domain and the target domain. Through deep fusion of three-dimensional grid information and two-dimensional image features, the method can maintain robust detection performance under adverse conditions of vehicle attitude change, illumination interference, shielding and the like, can effectively reduce large-scale data annotation and training cost, improves the rapid adaptation capability of the model in a new scene or a new vehicle type, and improves the robustness of the model. And a high-precision, extensible and rapid-iteration fine-grained identification and detection solution is provided for intelligent traffic monitoring, unmanned driving perception, military equipment identification and digital twin systems.

The invention provides a multi-source domain invariant acoustic feature extraction method and system for an equipment operation state, and belongs to the technical field of equipment maintenance, and the method comprises the steps: constructing a multi-source domain invariant acoustic feature extraction network based on a DANN model, and the network comprises a feature extractor, a classifier, a domain discriminator and a multi-domain acoustic feature class boundary constraint module; the feature extractor extracts high-dimensional features of sound signals, the classifier carries out fault mode recognition, and the domain discriminator realizes domain prediction. The multi-domain constraint module generates an embedding space, and calculates the maximum mean value difference, the local maximum mean value difference and the Euclidean distance among different source domain features. The network constructs a loss function by taking minimization of inter-domain difference and classification loss and maximization of inter-domain distance and domain discrimination loss as targets, and carries out adversarial training through multi-source tagged acoustic data. According to the method, domain invariant features are extracted by using adversarial learning and hidden space constraint alignment of multi-source domain acoustic features, so that the influence of feature offset under a cross-working-condition condition is effectively reduced, and the fault mode recognition accuracy is improved.

A defect detection method and a model transfer method. The model transfer method comprises: acquiring a source-domain image and a corresponding annotation, and a baseline model obtained from a source domain; acquiring target-domain images, which are fully annotated, partially annotated, or unannotated; inputting the annotation of the source-domain image into a denoiser to obtain a denoised annotation; inputting the unannotated target-domain images into an initial baseline model to obtain second target-domain prediction results, performing data augmentation on the unannotated target-domain images to obtain second target-domain augmented images, and using the second target-domain prediction results as pseudo-annotations of the second target-domain augmented images; and using the source-domain image and the corresponding denoised annotation, the annotated target-domain images and the corresponding annotations, and / or the second target-domain augmented images and the corresponding pseudo-annotations to train the baseline model, so as to obtain a transfer model. The method can complete training by using a small number of annotated target-domain images, thereby solving the problem of model transfer performance being poor in the case of insufficient target-domain images.

The invention relates to a self-adaptive cross-domain battery pack health state probability prediction method, and belongs to the technical field of battery health management. The method comprises the following steps: S1, data preprocessing and trajectory normalization; s2, carrying out adaptive step length difference modeling based on local curvature; s3, multi-modal noise modeling; and S4, cross-domain migration probability prediction based on GMM-UT. Quantizing the nonlinear characteristics of the degradation track through local curvature, and dynamically adjusting the difference step length so as to accurately extract the degradation characteristics; a Gaussian mixture model is adopted to fit multi-modal noise, and accurate quantification of uncertainty is achieved; multi-modal noise statistical characteristics and unscented transformation are fused, cross-domain migration and uncertainty propagation are optimized, and battery pack degradation track multi-modal probability distribution is output. According to the method, the defects that an existing method is poor in fixed step length adaptability, inaccurate in noise modeling and insufficient in cross-domain prediction robustness are overcome, the health prediction precision and reliability of the battery pack are improved, and the method is suitable for health management of the battery pack under complex working conditions.

The invention discloses a quality parameter detection model construction method based on multi-source domain adaptation, and relates to the technical field of spectral analysis, the method comprises the following steps: constructing a quality parameter detection basic model, and constructing a multi-source spectral data set for training the quality parameter detection basic model; during model training, extracting source domain features and domain invariant features of the spectral data, and respectively predicting to obtain a source domain prediction result and a domain invariant prediction result; dynamically adjusting the contribution weight of each source domain training set to model training based on the performance of the domain invariant prediction result on each source domain training set; and determining a loss function of model training based on the domain invariant feature, the source domain prediction result and the domain invariant prediction result in combination with the contribution weight, and training by using the loss function to obtain a quality parameter detection model. According to the method, the problem of multi-source spectrum cross-domain regression prediction can be effectively solved under the condition of no target domain spectral data, and the robustness and accuracy of quality parameter detection are improved.

This invention provides a medical ultrasound image recognition method based on generator domain alignment, relating to the field of image classification technology. The method includes: creating labeled generated medical ultrasound image samples for fine-tuning using a conditional diffusion model, where labels characterize the category corresponding to the generated medical ultrasound image sample; aligning each labeled generated medical ultrasound image sample to the generator domain using an unconditional diffusion model of the generator domain, obtaining a generator domain medical ultrasound image set; fine-tuning the source domain model using the generator domain medical ultrasound image set, obtaining a generator domain model; aligning a target medical ultrasound image to the generator domain using the unconditional diffusion model of the generator domain, obtaining a generator domain target medical ultrasound image; and inputting the generator domain target medical ultrasound image into the generator domain model to obtain a first predicted classification result for the target medical ultrasound image. This transforms a cross-domain task into an intra-domain prediction task, aligning both the source domain model and the target data to the same generator domain, thereby achieving accurate predicted classification results.

The invention discloses a multi-industry load joint prediction method based on time-frequency alignment, and solves the problems of large error and low reliability of a load prediction result in the prior art, and the method comprises the steps: obtaining historical load data of a to-be-predicted industry and corresponding external data; carrying out preprocessing and normalization processing on the data; extracting frequency domain information of the load data based on short-time Fourier transform; according to the time domain information and the frequency domain information of the load data, a multi-industry load time domain prediction model and a multi-industry load frequency domain prediction model are constructed based on an LSTM neural network; and on the basis of time-frequency alignment, forming a multi-industry load joint prediction model by the multi-industry load time domain prediction model and the multi-industry load frequency domain prediction model, and predicting the multi-industry load. According to the invention, the accuracy and reliability of the load prediction result are improved.

A communication device in a communications network that includes a network node can generate (1620) a report based on a mobility related time-domain prediction of a neighbor cell of a plurality of neighbor cells The report can include information associated with a subset of the 5 plurality of neighbor cells. Generating the report can include selecting the subset of the plurality of neighbor cells based on the mobility related time-domain prediction of the neighbor cell. Generating the report can further include sorting the subset of the plurality of neighbor cells based on the mobility related time-domain prediction of the neighbor cell. The communication device can further transmit (1630) the report to the network node.

The invention discloses a long-tail radiation source individual identification method and device based on field generalization and a storage medium. The core of the method is that a plurality of types of source domain data sets distributed in a long-tail mode are obtained; sampling to obtain batch training data; after extracting feature representation by using a feature extractor, inputting the feature representation into a classifier and a domain discriminator at the same time; based on a classification prediction result, combining a field prediction result of a gradient inversion mechanism and class balance loss to construct an integrated overall objective function; all model parameters are jointly optimized by minimizing the target function, and an optimized recognition model is obtained and is finally used for performing high-precision individual recognition on target domain radiation source signals. Through collaborative optimization of classification precision, category balance and domain generalization ability, the problem of insufficient recognition performance of the model on an unknown target domain in a complex scene with training data long tail distribution and domain offset is effectively solved.

The invention discloses a carrying equipment bearing fault diagnosis method and device based on time-frequency dual-domain prediction, and the method comprises the following steps: collecting a bearing acceleration signal, carrying out the normalization preprocessing of the bearing signal, dividing the data into a training set, a fine tuning set and a test set, carrying out the enhancement processing of the data, and carrying out the fault diagnosis of the bearing. And inputting the time domain representation and the frequency domain representation into an encoder for feature extraction, performing model pre-training according to a set predefined proxy task, constructing a target diagnosis model based on a pre-training model, performing fine tuning on the target diagnosis model by using a fine tuning set, and finally inputting a test set into the fine-tuned target diagnosis model for diagnosis. According to the method, diversified data can be generated, potential feature information in unlabeled data can be efficiently extracted, diagnosis is rapid and accurate, and bearing fault diagnosis of intelligent carrying equipment can be realized.

The invention discloses a milling chatter stability domain prediction method and system based on an approximate analysis method, and the method comprises the steps: employing a Fourier approximation method to represent a milling steady-state response based on a milling direction coefficient and a time delay characteristic in a milling dynamic model; applying small disturbance to the steady-state response, substituting the steady-state response into the milling dynamic model, and obtaining a milling stability analysis expression by applying a Galerkin process; substituting the bifurcation condition into the stability analysis expression based on the characteristics of the stability domain boundary corresponding to the milling stability bifurcation condition, and deducing an approximate implicit analytical expression of the milling stability boundary; solving the implicit analytical expression by adopting a numerical continuation method to obtain a stability boundary expressed by rotating speed-cutting depth; according to the method, the stability lobe graph is obtained, the problems of initial continuation point selection and transition between lobes with different stability in the solving process are solved, complete and accurate construction of the stability lobe graph is achieved, theoretical guidance of flutter-free working conditions is provided for milling, and the method has important engineering application value.

The invention discloses an equipment remaining service life prediction method, equipment and a medium, and the method comprises the steps: firstly obtaining operation monitoring data of source domain and target domain equipment, then constructing according to the source domain data to obtain a first input time sequence, and inputting the first input time sequence into a source domain prediction model framework; a feature extraction model and a probability prediction model are arranged in the model framework. Performing feature processing on the sequence through a feature extraction model to obtain comprehensive degradation features, training a probability prediction model by using the degradation features, and obtaining a trained model framework and corresponding posterior parameters based on the trained prediction model; and after a target domain prediction model is constructed according to the target domain monitoring data and the parameters, predicting the to-be-detected equipment through the model to obtain a prediction result of the equipment and corresponding uncertainty information. According to the method, robust prediction in a cross-domain scene can be realized, and meanwhile, corresponding uncertainty information is provided, so that a more comprehensive reference basis is provided for an equipment maintenance decision.

The application provides a building group energy consumption prediction method and system based on space-time graph convolution and adversarial domain adaptation, comprising: obtaining source domain related data and target domain related data, preprocessing to obtain initial input data, etc.; constructing a space-time graph convolution network model and pre-training to obtain a pre-trained space-time graph convolution network model, splitting the model into a space-time feature extractor and a predictor; performing weight initialization and adversarial training on the target domain space-time feature extractor and the target domain predictor to obtain a trained target domain space-time feature extractor and a trained target domain predictor; inputting a target domain test set into the trained target domain space-time feature extractor and the trained target domain predictor to obtain a target domain regional building group energy consumption prediction result; and the application captures the spatial dependence relationship between buildings based on graph convolution and graph attention mechanism, adopts a transfer learning strategy to transfer source domain prediction related knowledge to a target domain, and accurately predicts energy consumption by using a small amount of data.

The invention relates to the technical field of voice recognition, and particularly discloses a self-adaptive voice recognition method and system based on domain feature fusion, and the method comprises the steps: obtaining hardware parameters of voice interaction equipment in response to an interaction request, and determining a current computing power mode of the voice interaction equipment based on the hardware parameters; based on the determined computing power mode, activating a corresponding speech recognition model structure; end-side preprocessing is performed on an input original voice signal to generate a preprocessed voice signal. According to the invention, through hardware computing power perception and dynamic model scheduling, the optimal balance between the identification precision and the calculation efficiency under different resource constraints is realized; through a domain prediction and hierarchical feature fusion mechanism, the adaptability and recognition robustness of multi-domain voice are enhanced; and non-standard voice calibration, context memory management and end-cloud collaborative updating capabilities are further fused, so that the understanding accuracy of personalized pronunciation, continuous dialogues and dynamic scenes is remarkably improved.

The application designs a new energy aircraft lithium battery health condition deep learning cross-domain prediction method, and belongs to the field of artificial intelligence and aviation technology; firstly, the discharge current data of the battery cycle process is collected, all SOH of the battery is extracted and arranged in time sequence to obtain the SOH time sequence of the battery; then, the source domain and the target domain of the battery are determined; a convolutional neural network is used as the framework of the prediction model, and the training is carried out, so that the differences between different battery categories can be well reflected, the differences can be made up in the optimization process, and then the prediction model meeting multiple battery health state indexes is obtained; the cross-domain health state index of the battery is predicted; the technical scheme has the advantages of low complexity, strong robustness and the like.

The application provides a feasible domain prediction method, device, system and storage medium, which is applied to vehicle automatic driving or auxiliary driving, and the method comprises the following steps: acquiring a surround view image at a current time, and obtaining bird's eye view features according to the surround view image; the surround view image comprises images of multiple perspectives collected by multiple cameras on the vehicle; extracting the bird's eye view features to obtain bird's eye view high-dimensional image features at the current time; generating a future feasible domain prediction map according to a time sequence queue formed by the bird's eye view high-dimensional image features at the current time and bird's eye view high-dimensional image features at multiple historical times, and outputting the feasible domain prediction map. The application realizes analysis and prediction of a future scene, can provide a basis for behavior decision of vehicle automatic driving or auxiliary driving, avoids repeated calculation and information accumulation error caused by dividing the feasible domain segmentation and obstacle prediction into two modules by combining scene perception and behavior prediction, and does not need to independently perform behavior prediction.

The application relates to a target detection pre-tracking method and system based on fuzzy control optimized particle filtering, and belongs to the field of target tracking. Particle states are predicted, and particle weights are calculated, the particle with the largest weight is regarded as an optimal particle, and the space near the optimal particle is regarded as an optimal region; the distance between the particle and the optimal particle and the particle ratio in the optimal region are calculated and taken as inputs of a fuzzy controller, a moving coefficient is output after a process of fuzzification, fuzzy reasoning and defuzzification; the particle state is updated according to the moving coefficient, and the particle with a lower weight is moved to a high-likelihood region; the iteration is performed for multiple times until the number of particles in the optimal region exceeds a threshold value; and the state of the target is estimated according to the particle state. The application can effectively alleviate the particle impoverishment problem, improve tracking accuracy, control particle movement through the fuzzy controller, be adaptively applied to different environments and scenes, and has good robustness and interpretability, and the rules can be expressed in natural language.

The present application relates to power distribution network fault diagnosis technical field, especially in based on cross-domain prediction and context comparison's power distribution network short-circuit grounding fault diagnosis method, system and equipment, method includes: collecting power distribution network normal working condition under voltage, current one-dimensional time sequence signal, completes data preprocessing and constructs time domain-frequency domain double view data set;Based on double view data set constructs the cross-domain prediction and context comparison diagnosis model that adapts to power distribution network, extracts the cross-domain prediction similarity feature, cross-domain context global feature, frame level local feature three kinds of core features;Based on cross-domain feature consistency constructs fault discrimination benchmark, judges whether power distribution network fault has and determines fault starting time;According to cross-domain feature matching degree constructs fault type exclusive benchmark feature library, and short-circuit grounding fault type is identified through comprehensive feature matching. Through the present application, effectively solve the problem of low precision, easy misjudgment of traditional fault diagnosis method under high proportion of distributed power access, and the problem that supervised comparison learning relies on a large number of labeled data and is difficult to scale.

This invention discloses a large-scale digital twin simulation and prediction method and system for rockburst, relating to the fields of mine safety and digital technology. The prediction method includes the following steps: constructing a three-dimensional geometric model, obtaining the geomechanical parameters of each rock layer, and assigning the geometric model to form an initial digital twin; collecting multi-source monitoring data from the mine, fusing multi-source heterogeneous data and mapping it to the digital twin model, driving the model state update to keep it synchronized with the physical mine; establishing constitutive equations and performing dynamic time-step solutions; inputting different future mining plans or prevention schemes, simulating possible morphologies of unknown areas, analyzing the impact of mining activities on mine stability, and predicting the dangerous areas, evolution paths, and degree of danger of rockburst under these future scenarios; repeating the above steps with the data obtained after on-site adjustments and monitoring, dynamically adjusting and continuously optimizing the digital twin model.

Systems and methods for triggering reports based on fulfillment of joint events associated to measurements and predictions are disclosed. In one embodiment, a method performed by a User Equipment (UE) for triggering a report comprises performing one or more measurements on at least one cell, performing one or more mobility related time-domain predictions on the at least one cell, and triggering transmission of a report comprising the one or more measurements performed on the at least one cell and / or the one or more mobility related time-domain predictions performed on the at least one cell, responsive to both a first condition and a second condition being fulfilled wherein the first condition is associated to the one or more measurements and the second condition is associated to the one or more mobility related time- domain predictions. The method further comprises transmitting the report responsive to the triggering of the transmission of the report.

The application relates to the field of fault diagnosis, in particular to a double-flow confrontation fault diagnosis method and system based on a symmetric point pattern image. The method comprises the following steps: obtaining a fault diagnosis result based on vibration data of a device to be diagnosed and a diagnosis model; a training process of the diagnosis model comprises the following steps: obtaining historical vibration data and / or historical fault labels of a plurality of devices; preliminarily optimizing a feature extraction unit by using the historical vibration data of the devices; obtaining aligned symmetric point pattern image features and vibration signal features based on the preliminarily optimized feature extraction unit and the historical vibration data of the devices, and performing feature fusion to obtain fused features; inputting the fused features into a domain discriminator to obtain a domain prediction probability, and optimizing the preliminarily optimized feature extraction unit based on the domain prediction probability, and optimizing a fault classifier based on the fused features and the historical fault labels to obtain a trained diagnosis model. The method is used for improving the accuracy and reliability of cross-machine fault diagnosis of mechanical devices.

The application discloses an abdominal aorta automatic segmentation method based on a segmentation sampling neural network, and specifically comprises the following steps: obtaining a historical CT sequence scanning image, performing image preprocessing and cutting with a preset size to obtain an original cutting file; inputting the original cutting file into a pre-trained front segmentation network model after down-sampling to perform forward propagation, and obtaining a full receptive field prediction file; performing global segmentation sampling on the original cutting file based on a segmentation step to obtain a plurality of first sub-domain files; performing local segmentation sampling on the original cutting file based on the segmentation step to obtain a plurality of second sub-domain files; constructing and training a sub-domain prediction network model, wherein the sub-domain prediction network model takes the full receptive field prediction file, all the first sub-domain files and all the second sub-domain files as inputs; and using the front segmentation network model and the trained sub-domain prediction network model to perform target segmentation on a real-time CT sequence scanning image to obtain a voxel segmentation model of the abdominal aorta.

The invention discloses a pear maturity nondestructive testing method and system based on mixed spectrum multiple tasks, and belongs to the crossing field of nondestructive testing and spectrum analysis technologies. The prediction method comprises the following steps: acquiring spectral reflectivity data of a target pear fruit, and performing three-view preprocessing to generate an original spectrum, a first-order derivative spectrum and a standard normal variable standardized spectrum; inputting the processed spectral data into a pre-trained mainstream pear variety global model, and synchronously outputting predicted values of hardness and soluble solids; and finally, obtaining a corresponding specific judgment threshold according to the variety of the target fruit, and comprehensively judging the maturity by comparing a predicted value with the threshold. According to the method, the advantages of traditional machine learning and deep learning are fused, the mainstream pear variety global model with the cross-variety generalization ability is constructed, the variety specificity threshold value is combined, rapid, lossless and accurate prediction of the maturity of the multi-variety pears is achieved, and the method is suitable for multi-scene application such as orchard harvesting, storage grading and market circulation.

The invention discloses a critical parameter prediction method, a readable storage medium and a device, and belongs to the technical field of nuclear systems, and the prediction method comprises the steps: determining a to-be-predicted new nuclear system and a calculated value of an existing critical experiment; calculating sensitivity vectors of the new nuclear system and the critical experiments to the nuclear data; on the basis of the calculated value and the sensitivity vector, in combination with the relationship between the experimental value of the existing experiment, the experimental uncertainty and the critical experimental coefficient, constructing a calculation formula of the critical parameter predicted value and the predicted uncertainty of the new nuclear system; solving a coefficient of a critical experiment enabling the uncertainty of the prediction parameters to be minimum; and substituting the coefficient into the two calculation formulas to obtain a predicted value and the uncertainty thereof. The method is stored in the readable storage medium. The device comprises a terminal program for executing the steps of the method. According to the method, the information of a plurality of critical experiments can be effectively integrated, the critical parameters can be more reliably and accurately predicted under the condition of lack of very similar critical experiments, and the uncertainty of the prediction parameters can be more effectively compressed.

This application provides a method for predicting the remaining service life of a turbine engine based on cross-domain transfer. The method includes: training a raw remaining service life prediction model using source domain training data; the raw remaining service life prediction model consists of a multi-scale feature extraction module and a prediction module, the multi-scale feature extraction module including a Transformer network, a CNN network, and a feature coupling unit; performing cross-domain learning on the pre-trained prediction model using target domain data to obtain a target remaining service life prediction model corresponding to the target domain; acquiring the operating data corresponding to the turbine engine to be predicted, and inputting the operating data into the target remaining service life prediction model to predict the remaining service life, thereby obtaining the remaining service life corresponding to the turbine engine to be predicted. Through this method and apparatus, the performance degradation problem caused by differences in data distribution in cross-domain prediction is effectively alleviated, achieving accurate prediction of the remaining service life result in the target domain.

The invention discloses an industrial equipment time series data prediction method and system based on multi-frequency component modeling, and the method comprises the following steps: carrying out the multi-layer wavelet decomposition and reconstruction of the obtained time series data of industrial equipment; obtaining multiple layers of low-frequency trend signals for representing long-term stable change of equipment operation data and high-frequency detail signals which are complementary with the low-frequency trend signals and are used for capturing short-term fluctuation; performing interactive modeling on the low-frequency trend signal and the high-frequency detail signal through a feature fusion mechanism to generate an updated low-frequency trend signal; constructing a low-frequency prediction model according to the updated low-frequency trend signal, and generating a low-frequency prediction result for describing the overall operation trend and long-term change of the system; constructing a high-frequency prediction model according to the high-frequency detail signal, and generating a high-frequency prediction result used for supplementing low-frequency prediction result detail information; and fusing the low-frequency prediction result and the high-frequency prediction result of each level, integrating prediction information of different frequency bands, and generating time domain prediction output.

The application discloses a kind of time prediction methods based on process clustering, it is related to the research field of time series prediction.The time of processing target process of specific target team is predicted, which provides reference basis for the selection of scheduling scheme and the evaluation of production cycle.According to the clustering results of the process, the different types of process are fitted with the corresponding work time data, and the learning curve model associated with the cumulative processing quantity of the team, the personnel shift system and the team size is obtained respectively.After determining the target process and the target team, the target process category is calculated according to the clustering results, and the three-factor learning curve of the target team under the task of this category is used to predict the target work time.

The invention relates to the technical field of mold life management, in particular to a mold life evaluation method and system, and the method comprises the steps: collecting mold multi-source data, and generating a structured multi-source data set after lightweight preprocessing; calculating an early warning proportion through curve fitting and a parameter fine tuning algorithm in combination with mold life characteristics and maintenance states, calculating life-increasing parameters through a constraint matching formula, rejecting low-efficiency data, and generating high-precision evaluation parameters; a double-time-domain prediction model is constructed, and prediction results of short-term early warning risks and medium-term life consumption trends are output; performing abnormity judgment by adopting a dynamic threshold algorithm, correcting core parameters through a data calibration mechanism according to a judgment result, generating an evaluation task adjustment instruction, and generating system optimization data according to an instruction execution result, model calibration information and mold operation data; and through the collaborative strategy model, the management strategies of different scenes are adapted. According to the scheme, the service life of the mold can be accurately controlled, and the production efficiency is improved.

This invention discloses a method, system, device, and medium for power energy consumption analysis based on multi-source data fusion, belonging to the field of power energy consumption analysis technology. The specific steps are as follows: Historical and real-time multi-source working data are collected and preprocessed; behavioral event feature vectors are extracted from a multi-layer multi-source aligned dataset; a multi-level energy consumption and carbon emission baseline dataset is established using a graph neural network; multi-time-domain energy consumption and carbon emission prediction is performed; based on the prediction results, energy-saving and carbon optimization strategies are output and verified in a simulation model; the verified energy-saving and carbon optimization strategies are deployed and executed, and the strategy generation process is optimized based on execution feedback. This invention achieves collaborative optimization management of energy consumption and carbon emissions in complex industrial parks, improves state characterization accuracy through multi-source fusion and behavior-driven approaches, achieves risk identification through multi-time-domain prediction, and ensures the safety and adaptability of the strategies through closed-loop verification and dynamic adjustment.

The invention discloses an intelligent terminal operation scene optimization method and system, and relates to the technical field of intelligent terminal operation scene optimization and computing resource management, and the method comprises the steps: obtaining environment state data and user interaction data; predicting resource gap information based on the state data; predicting a scene task to be triggered based on the interaction data, wherein the second time period is longer than the first time period; generating a resource allocation instruction and a preloading instruction according to the resource gap information and the scene task prediction result; dynamically distributing the computing resources and executing preloading operation; and collecting running state data when the scene task is executed, and performing feedback optimization on the prediction operation based on the running state data. According to the method, passive response of resource configuration and software preparation is converted into prospective execution through dual-time-domain prediction cooperation, and prediction parameters are continuously optimized in combination with execution feedback, so that scene triggering time delay and system jitter are remarkably reduced in a dynamic environment, and the stability of a control center, resource utilization efficiency and user experience consistency are improved.