201 results about "Information quantity" patented technology

The invention discloses a data security risk assessment method based on a big data model, and relates to the technical field of data security, and the method comprises the steps: collecting and preprocessing multi-source data, collecting security-related data from network equipment, a server and an application system, carrying out the preprocessing, carrying out the adaptive feature extraction, and carrying out the data security risk assessment. The feature importance is evaluated by calculating the mutual information amount of features and risk tags, a standardized feature vector set is constructed, multi-model collaborative analysis is performed, feature vectors are input into a cascade collaborative network composed of an anomaly detection model, a threat recognition model, a correlation analysis model and a prediction model, and a risk risk is obtained. Through cross-model feature transmission and a bidirectional information feedback mechanism, deep collaborative analysis and multi-model deep fusion decision making are carried out, a weight is calculated according to historical accuracy of each model, a comprehensive risk score is calculated by adopting dynamic gating deep fusion, and a dynamic threshold value is calculated based on a sliding time window. And the risk is divided into three levels of high risk, medium risk and low risk.

The invention relates to the technical field of machine learning models, in particular to a geological disaster risk dynamic assessment method based on multi-source data fusion, which comprises the following steps: constructing a basic geographic information database; dividing geological disaster risk areas by adopting a machine learning algorithm; calculating the contribution degree of each environment factor to the geological disaster through an information amount model; fusing the dynamic rainfall data, carrying out weighted fusion on the dynamic rainfall data and the static geological disaster factors, and calculating a comprehensive risk value; dividing risk levels according to the risk values, and generating a geological disaster risk zoning map; compared with the prior art which mainly depends on single static geological data for analysis and has the problems of incomplete evaluation dimensions and poor timeliness, the scheme realizes multi-dimensional fusion analysis of geological conditions and real-time meteorological factors by constructing the geographic information database integrating the multi-source environmental factors and the dynamic rainfall data; and the comprehensiveness and the momentality of risk assessment are obviously improved.

The invention relates to the technical field of computers, discloses an astronomical sensitive data security processing method, device and system based on privacy calculation, and a storage medium, and aims to solve the problems that in astronomical data cross-mechanism joint analysis, original data is easy to leak, the cooperation efficiency is low, and a traditional desensitization or encryption method is difficult to consider security and availability at the same time. The method specifically comprises the following steps: each participant deploys a private computing agent node locally, and original data is not out of a domain; the task coordination center issues an analysis task; the proxy node extracts and preprocesses local data, and loads a corresponding secure multi-party computing protocol template; according to the method, share segmentation is carried out on multi-modal data such as images, spectrums and star catalogues by adopting addition homomorphic secret sharing, and distribution is carried out through a national secret SM4 encryption channel; and multiple parties cooperatively execute task-oriented security calculation in an encrypted state, and the result is aggregated and returned with the minimum information amount. The method has the effect of realizing high-precision safe collaborative analysis which is available and invisible.

The invention belongs to the technical field of artificial intelligence and power system crossing, and discloses a marketing and distribution fusion scene-oriented multi-modal large model efficient fine tuning method and system, and the method comprises the steps: firstly constructing a small amount of high-quality multi-modal seed data with a thinking chain by field experts; performing domain knowledge injection and multi-modal alignment on the base model in stages by adopting a parameter efficient fine tuning technology; the method comprises the following steps of: selecting an unlabeled sample, further introducing an active learning iterative loop based on hybrid uncertainty perception, automatically screening the unlabeled sample with the most rich information amount by quantifying cognitive uncertainty and accidental uncertainty of a model, and labeling the unlabeled sample by an expert, so as to expand a data continuous optimization model and form a'fine tuning-evaluation-labeling 'closed loop. According to the method, rapid and accurate adaptation of the multi-modal large model in a marketing and distribution fusion complex scene is realized with extremely low expert labeling cost, the service reliability and safety of model output are ensured, and a long-acting mechanism of sustainable evolution of the model is established.

The invention discloses a stomach cancer postoperative survival prediction method and system based on machine learning, and belongs to the technical field of medical worker crossing and medical worker combination. According to the technical scheme, the method comprises the following steps: acquiring clinical data of a gastric cancer patient, wherein the clinical data comprises demographic characteristics, tumor pathology characteristics, operation related parameters and laboratory detection indexes; filling missing values in the clinical data by using an iterative random forest missing value filling method based on mutual information weighting; on the basis of the filled data, a feature subset with the most information content for postoperative three-year survival prediction is screened out through a dual feature selection strategy; training a machine learning model by using the feature subset so as to predict the survival risk of the gastric cancer patient in three years after operation; and outputting a prediction result. The method has the beneficial effects that a plurality of key challenges from data preprocessing, feature engineering and model construction to interpretability and clinical application are systematically solved, and an accurate, reliable, transparent and practical gastric cancer postoperative survival prediction solution is finally formed.

The invention relates to image processing, in particular to a complex environment-oriented power transmission line inspection image processing method, which comprises the following steps of: screening key frames with high geometric information content by quantitatively evaluating inter-frame motion amplitude and feature tracking quality; extracting features of each key frame by using a pre-trained visual model; performing intra-frame feature aggregation on the features of each key frame by using an intra-frame self-attention mechanism to obtain corresponding intra-frame feature representation; performing global feature aggregation on the intra-frame feature representations of all the key frames by using a global self-attention layer to obtain inter-frame feature representations; constructing a multi-task learning network, and carrying out end-to-end collaborative optimization on depth estimation, image defogging and high-level semantic segmentation; inputting the inter-frame feature representation into a multi-task learning network, and obtaining a restored clear image through deep fusion of scene depth information and image degradation priori; according to the method, the defect that the dual requirements of quality improvement and feature retention of the power transmission line inspection image in a complex environment cannot be met can be overcome.

The invention discloses a privacy enhanced CPPS anomaly detection method based on longitudinal federated learning, and relates to the field of CPPS anomaly detection. According to the method, through longitudinal federated learning, an SCINet model and a Transform model are deployed at clients of a physical side and an information side respectively and are used for local deep feature extraction; in the feature uploading stage, feature compression processing is performed on the physical side data deep features and the information side data deep features, so that effective compression of the uploaded features is realized, and meanwhile, the sensitive information amount possibly leaked in the middle features is greatly reduced. In addition, a bidirectional collaborative optimization mechanism between the client and the server is constructed, and a local feature extraction strategy can be optimized in real time. According to the method, the data privacy protection effect is remarkably improved while the anomaly detection accuracy is guaranteed, the defects of a traditional anomaly detection method in the aspects of privacy security, feature compression effectiveness, bilateral collaborative optimization and the like are overcome, and the practicability and security of CPPS anomaly detection are effectively improved.

The invention discloses a power corridor disaster risk assessment method and system based on an entropy weight method, and particularly relates to the technical field of power corridor disaster risk assessment, and the method comprises the steps: obtaining original multi-source point location data, related to disaster risk assessment, of a power corridor region in the southwest region, carrying out the coordinate unification and time alignment, and carrying out the coordinate unification and time alignment; an original multi-source point location data set is obtained and subjected to quality control processing, the consistency and stability of the original multi-source point location data set are checked, missing values and abnormal values are removed, and the multi-source point location data set is output; on the basis of electric power corridor axis segmentation, double-threshold amplitude limiting conditions of an information amount retention rate and an extreme value order are set, a segmentation optimal interpolation kernel and a time window parameter set are constructed, and an interpolation scale and time aggregation granularity are dynamically backtracked and adjusted, so that risk integral offset caused by terrain fracture and window averaging is eliminated, and the power corridor axis segmentation accuracy is improved. And the credibility of the information amount of entropy weight evaluation is ensured.

The invention discloses a performance self-improving large model recommendation system and method, and the system comprises an initialization module which is used for pre-training a large language model through supervision and fine tuning, and generating an initial recommendation model; the self-optimization module comprises three iteratively executed sub-modules; the sample selection sub-module is used for screening historical data samples of which the information amount is higher than a threshold value on the basis of comparison between the model prediction probability and the preset threshold value; the response fusion sub-module is used for generating K candidate responses for a selected sample and generating a preference data set based on model evaluation; and the DPO optimization sub-module is used for updating model parameters by utilizing the improved DPO loss function and generating an optimized recommendation model. According to the method, the dependence on static preference data in the prior art is broken, the recommendation quality and robustness are improved, and adaptive optimization is realized.

The invention discloses an industrial mechanical arm joint rigidity online identification method based on an optimal excitation track, and belongs to the field of mechanical arm parameter identification. The method comprises the following steps: firstly, establishing a mechanical arm rigid-flexible coupling dynamic model, and constructing a linear regression equation taking joint rigidity and viscous damping as parameters; a Fisher information matrix is derived based on the regression model to quantify the amount of information of the excitation trajectory. And further, parameterizing the excitation trajectory by using finite term Fourier series, and solving the optimal excitation trajectory by using a scalar index of a Fisher information matrix and a sustainable excitation condition as optimization targets under the condition of meeting joint motion constraints. And finally, the mechanical arm is driven to execute the track, motion data are collected and input into parameter identification algorithms such as unscented Kalman filtering and recursive least square, and online and high-precision estimation of the time-varying joint stiffness is achieved. According to the method, the excitation track is designed directly from the information theory, and the precision, robustness and real-time performance of joint rigidity identification are effectively improved.

The invention discloses a dual-channel collaborative unknown signal reconstruction error control method, and belongs to the technical field of wireless communication. The method solves the problem that the existing method cannot give consideration to both the signal reconstruction precision and the information content contained in the signal. The invention provides a scheme of combining dual-channel collaborative quantization and error control of a receiving end, a dual-channel quantization mechanism realizes information complementation through different quantization characteristics, single-channel error accumulation is inhibited, and the signal detail retention capability is improved. And in combination with an interpolation processing method, estimating and compensating quantization information, optimizing reconstruction output, and completing high-fidelity signal recovery. According to the method, the reconstruction precision and fidelity of unknown signals can be remarkably improved, the robustness and adaptability of the system under the condition of lack of prior information are ensured, meanwhile, extreme dependence on the performance of a single-channel quantizer can be reduced, and then efficient balance between computing resources and reconstruction quality is achieved. The method can be applied to the field of wireless communication.

The application provides a multi-view clustering method and system based on a graph attention automatic encoder, relates to the technical field of multi-view clustering, and specifically includes the following steps: selecting a view with the largest information quantity from different views of the same group of nodes; learning a graph structure and node content by using a trained graph attention encoder based on the view with the largest information quantity and node content information, so as to obtain a node feature representation; performing specific constraint on the node feature representation by using an l1,2-norm penalty, so as to obtain a constrained node feature representation; inputting the constrained node feature representation into a self-optimizing clustering module to perform clustering, so as to obtain a final clustering result; and the application applies the graph attention network to multi-view graph clustering, simultaneously reconstructs the graph structure and the node content, and makes the latent representation well preserve the graph structure and the content information of the nodes, so that the application is more suitable for cluster tasks.

The invention provides a glioblastoma patient lifetime prediction method and system, and relates to the technical field of intelligent medical treatment, and the method comprises the steps: obtaining and coding a digital full-slice image of a patient; screening high-information-content image blocks based on information entropy; executing a dendritic calculation rule and a structural plasticity rule through internal learnable dendritic neurons by using a dendritic morphological neural calculation model, and extracting global and local features from the image blocks; fusing the features through feature fusion neurons to obtain depth features; and finally, predicting the total lifetime through the output neurons. By simulating hierarchical calculation and plasticity of neuron dendrites, a double-level feature learning model capable of integrating global context and local details of a full-slice image at the same time is constructed, and the technical problems that an existing prediction method is insufficient in feature integration and limited in model calculation structure are effectively solved. And more comprehensive and accurate prediction of the lifetime of the glioblastoma patient is realized.

The invention discloses an information guide identification and form filling system and method, belongs to the technical field of intelligent information interaction and data processing, and aims to solve the problems of fuzzy demand positioning, fragmentation of information acquisition, low form filling efficiency and insufficient data credibility in traditional information interaction. After initial basic information of a user is collected, a question and answer association graph is constructed through a graph neural network, the initial information is mapped to graph generation node association features, and a guide label is generated through clustering analysis; matching a core guiding main line based on guiding labels and node weights, dividing interaction stages, generating a verbal skill construction guiding plan, starting self-adaptive question and answer to construct an original demand pool, and verifying and integrating the original demand pool into an initial demand information set; and matching a multi-modal acquisition target to generate a standardized data set, correcting structured information to position a target demand, matching a scale template, and realizing intelligent form filling through information quantitative conversion and deletion completion. According to the method, the cross-scene information interaction efficiency and the data availability are remarkably improved.

The invention relates to the technical field of education evaluation, in particular to an English reading understanding ability self-adaptive evaluation method and system. The method aims at solving the technical problems that the traditional fixed difficulty test is easy to cause inaccurate capability estimation, the existing adaptive test is rough in subject description difficulty, the reading skill diagnosis is fuzzy, and the question bank security is insufficiently considered. According to the technical scheme, the method comprises the steps that a pre-training language model is used for conducting word, syntax, chapter and subject multi-dimensional difficulty feature extraction on a reading material; correcting question parameters based on an item reaction theoretical model in combination with answer data; dynamic evaluation is realized by adopting a self-adaptive test engine based on Bayesian capability estimation and maximum information amount topic selection; outputting the fine-grained skill mastering degree through the cognitive diagnosis model with the Q matrix constraint; a question exposure control strategy is integrated to balance measurement precision and question bank safety. And the system finally outputs a capability level, a skill diagnosis radar map and personalized reading recommendation.

The present invention optimizes the amount of input information with respect to a language model for determining the confidentiality of information, and enables identification of confidential information with excellent efficiency and accuracy. This information processing system 100 is configured to comprise: an auxiliary storage device 203 for retaining information about sentences; and a processor 201 that calculates, on the basis of the information about the sentences, an anticipated computational cost and certainty factor during confidentiality identification with regard to respective sets of a specific sentence and a peripheral sentence, that generates a peripheral information list indicating peripheral sentences selected on the basis of results thereof, and that inputs specific sentences and peripheral sentences to a language model on the basis of the peripheral information list to thereby identify whether the specific sentences are confidential information.

The present application relates to the technical field of cross-modal retrieval, and particularly relates to a strong correlation unsupervised cross-modal retrieval method guided by information quantity, which is realized through the following steps: firstly, image local features and global features, and text features are extracted; the image local features and global features are enhanced; the enhanced local features are subjected to regularization processing; then, the image global features and local features are subjected to orthogonal fusion using an image feature fusion network; next, the image features and text features are fused according to a different modal feature information quantity conversion ratio principle using a multi-modal fusion network; finally, different modal features are mapped into hash codes, and Hamming distance is used for similarity sorting, so as to obtain a retrieval result. The present application focuses on the enhancement and fusion of data features, can obtain more semantic information, and improves the retrieval efficiency.

The invention discloses a novel multi-component conversion multiple imaging method for a deep-sea rugged seabed structure, and relates to the technical field of geophysical exploration for petroleum, and the method comprises the steps: inputting a longitudinal and transverse wave velocity field, a seismic source wavelet, observation system parameters, and a rugged seabed elevation; constructing a non-uniform curved grid of the acoustic-viscoelasticity model; calculating an acoustic-viscoelastic primary wave field continuation operator of vector wave separation under the curved coordinate system; calculating an acoustic-viscoelastic primary wave adjoint wave field of vector wave separation under the curved coordinate system based on an adjoint state theory; calculating an acoustic-viscoelastic n-order multiple wave field continuation operator of vector wave separation under the curved coordinate system; and calculating an acoustic-viscoelastic n-order multiple accompanying wave field of vector wave separation under the curved coordinate system, and generating a multiple imaging result by applying an elastic multiple imaging condition of vector wave separation. According to the invention, full-path compensation and longitudinal and transverse wave vector imaging of multiple waves can be realized, the imaging range is expanded, and the information amount of imaging is increased.

The present disclosure provides an information compression system that is capable of achieving higher compression efficiency. A data acquisition section acquires data. A generation section (segmentation section and integration section) determines each object depicted by the data and a sense of each object, and according to results of the determination, generates compression target data by converting values of elements in the data to identification information indicating each object and the sense of each object. A data storage section generates compressed data by compressing the compression target data. This makes it possible to convert highly random element values to slightly random identification information and compress the resulting converted information while reducing the amount of information. Consequently, the compression ratio can be increased.

The invention discloses a device, method and system for detecting water quality through full-automatic collection and positioning, and the method comprises the steps: firstly putting a device cabin into a target water body, and recording geographic position information by a Beidou satellite navigation system; the automatic water sample collecting device collects a water quality sample and pumps the water quality sample to the liquid storage tank, the coloring agent feeding device injects a coloring agent to form a mixed solution, and the mixed solution is conveyed to the color-sensitive sensing device through a third metering pump; the color-sensitive sensing device acquires image information and transmits the image information to the data acquisition and processing device; extracting an RGB proportion value, distinguishing polluted and unpolluted areas through a critical relationship method, calculating matrix entropy and average information amount in combination with an information entropy theory, and evaluating the pollution degree; an evaluation result and a geographic position are fused to generate a pollution path map, a Beidou satellite navigation system plans a route, and a displacement device drives a device cabin to move towards a low-pollution area; and repeating the sampling evaluation process, and dynamically updating the pollution path diagram until the pollution source is locked. The problems of insufficient pollution source positioning accuracy and efficiency and the like in the prior art are solved.

This paper presents a graph neural network node classification method that integrates meta-learning and mini-batch training, belonging to the field of information technology. First, the method utilizes the METIS algorithm to divide the original large-scale graph data into multiple non-overlapping connected subgraphs. Then, by constructing a hybrid selection mechanism based on label coverage and label entropy, subgraphs with high information content and strong representativeness are selected from the subgraph pool as the meta-learning task. Subsequently, iterative training is performed on the selected subgraphs using the meta-learning framework to capture the general prior features of the graph structure, thereby obtaining a set of initial parameters for the model with rapid adaptability. Finally, these optimized initial parameters are transferred to the mini-batch training stage on the full dataset, guiding the model to achieve rapid convergence through high-quality initialization. On large-scale benchmark datasets, this method significantly reduces the number of iterations required by the model while maintaining the same classification accuracy as current mainstream graph neural network models, thus greatly shortening the overall training time.

The application discloses a non-circular sparse array DOA estimation performance evaluation method based on information theory, which comprises the following steps: firstly, a multi-dimensional probability density function (PDF) of a received signal is constructed, a joint PDF of the received signal and DOA and a DOA posteriori PDF are derived through information theory, and a system DOA information quantity is obtained by simplifying the DOA posteriori PDF with a Bessel function; then, a DOA posteriori PDF of given noise is derived, and a DOA information approximate upper limit is obtained by simplifying the DOA posteriori PDF with a Taylor expansion; finally, a DOA estimation performance index entropy error is obtained by using the posteriori differential entropy; the application builds a non-circular sparse array system DOA information theory framework based on information theory, in actual signal processing, the entropy error of a parameter can be calculated only by estimating the posteriori PDF of the parameter, and a performance limit independent of an algorithm is provided; in addition, it is found through simulation that the DOA information quantity approaches the DOA information upper limit under a high signal-to-noise ratio, and the entropy error approaches the Cramer-Rao limit, thereby verifying the rationality of the index.

The invention discloses an image detail enhancement and noise suppression method in a low-light environment, and belongs to the technical field of image processing. The method comprises the steps of firstly extracting a low-illumination image brightness value to generate an original brightness map, and adding random disturbance to obtain a disturbance sensitive feature map; calculating a local brightness information amount based on a pixel local gradient, and generating an information effectiveness weight map in combination with a brightness random degree and a structure disorder degree; extracting a brightness detail image from the original brightness image, and fusing the feature images to obtain an anti-disturbance and credible detail image; and finally, through image reconstruction neural network processing, outputting a detail-enhanced denoising brightness image. According to the invention, accurate balance between detail reservation and noise suppression of the low-illumination image is realized, and the image quality is improved.

The application discloses a kind of low-light environment under image detail enhancement and noise suppression method, belong to image processing technical field.The application first extracts low-light image luminance value to generate original luminance graph, obtains disturbance sensitive feature graph by adding random disturbance;Then, based on the local gradient of pixel, the local luminance information quantity is calculated, and the information effectiveness weight graph is generated by combining luminance randomness and structure disorder degree;Subsequently, the luminance detail graph is extracted from the original luminance graph, and the anti-disturbance, reliable detail graph is obtained by fusing the above feature graph;Finally, through image reconstruction neural network processing, the detail enhancement denoising luminance graph is output.The application realizes the accurate balance of low-light image detail retention and noise suppression, and improves image quality.

The application discloses a kind of crew workload self-adaptive prediction methods based on multi-source heterogeneous data fusion, the method includes: the physiological characteristic data of crew is collected to obtain the current physiological load of the individual;Based on the current physiological load of individual and the physiological load parameter of historical time, the individual physiological load prediction value of next time period is obtained;Obtain the navigation environment data of the ship where the crew is in next time period, and obtain the comprehensive stress level of the navigation environment where the crew is in next time period;Based on the comprehensive stress level of the navigation environment where the crew is in next time period and the individual physiological load prediction value of the next time period, input causal reasoning graph model, predict the workload level of crew.The application comprehensively, multidimensionally reflects the working state of crew by collecting and fusing the physiological characteristic data of crew, historical physiological load parameter and navigation environment data, improves the information quantity and comprehensive analysis ability of prediction model.

The invention discloses a vehicle-mounted and road scene traffic information quantity collaborative threshold determination method, which comprises the following steps of: respectively dividing a traffic information source of a road scene and display elements of a vehicle-mounted interactive interface into a plurality of categories, calculating information quantity of each category based on an information entropy model, and determining weight of each category, respectively obtaining the total information amount of the road scene and the total information amount of the vehicle-mounted interaction interface; constructing a function of the traffic information amount and the gaze entropy based on a Wundt curve; acquiring eye movement gazing data of a driver in a combined driving scene with different total information amounts of a road scene and a vehicle-mounted interaction interface, and calculating a gazing entropy value; and by taking the traffic information amount as input and the gaze entropy value as output, performing parameter fitting on the function to obtain a minimum threshold value and a maximum threshold value of the traffic information amount. According to the method, the gaze probability distribution differences corresponding to different traffic information amounts are reflected through the gaze entropy values, so that the information amount change can be realized through quantitative indexes, and the influence of information complexity on driving behaviors is truly reflected.

The invention discloses a data collection method and device and electronic equipment, the data collection method can be applied to the technical field of automatic driving, and the method comprises the steps: obtaining original data related to a to-be-executed data collection task; extracting feature information from the original data based on a preset feature dimension; based on the feature information, the contribution degree of the original data in a preset evaluation dimension is quantified, and a trigger priority score of the data acquisition task is obtained; determining a target trigger level matched with the trigger priority score in preset trigger levels; different trigger levels in the preset trigger levels correspond to different resource allocation rules; and based on a resource allocation rule corresponding to the target trigger level, allocating system resources to the data acquisition task. According to the method, a complex driving environment is described through multi-dimensional feature fusion, collaborative optimization classification logic is provided, accurate resource allocation according to classification values is realized, and collaborative improvement of dimension, precision and resource efficiency of automatic driving data acquisition is achieved.

The application discloses an abnormal file processing method and device, a storage medium and an electronic device. It relates to the field of big data. The method comprises the following steps: receiving an abnormal file, and obtaining a plurality of pieces of abnormal information in the abnormal file; calculating the total information quantity of the plurality of pieces of abnormal information, and determining whether the total information quantity is greater than a target information quantity; in the case where the total information quantity is less than or equal to the target information quantity, sequentially identifying a preset identifier in each piece of abnormal information, processing the abnormal information according to the preset identifier in each piece of abnormal information, obtaining an updated abnormal file, and storing the updated abnormal file in a database; in the case where the total information quantity is greater than the target information quantity, sending the abnormal file and the total information quantity to a server, and displaying alarm information on the server. Through the application, the problem of low efficiency in processing abnormal information in an abnormal file by manual work in the related art is solved.

The application provides a polarization and visible light image fusion texture extraction method and device. A visible light reflection component is first extracted to screen a reliable linear polarization degree image texture and generate a preliminary fusion texture image. Then, a deep learning network model is trained by taking the preliminary fusion texture as a supervised true value, combining a reasonable data enhancement method and a scientific training strategy, so as to eliminate residual block effects and noises in artificial textures. Finally, a fusion texture with high information quantity, high precision, low noise and low block effect is output. In this way, the complementary characteristics of the polarization and visible light images are fully utilized, the texture extraction result is clear and free of artifacts, and the method is suitable for image fusion, edge detection and other real complex scene image detail enhancement and quality optimization.

The application discloses a soil and rock dam deformation field reconstruction method based on a diffusion model, belongs to the field of water conservancy engineering safety monitoring and engineering intelligent analysis, and comprises the following steps: constructing a soil and rock dam random finite element deformation field data set, performing grid mapping, geometric mask processing and global normalization on the deformation field sample, and forming a two-dimensional deformation field training sample; training the diffusion model by using the training sample, wherein an active sampling strategy is adopted in the training stage, and high-information samples are preferentially selected according to sample information quantity to participate in training; in the sampling stage, sparse monitoring data is mapped to a section grid coordinate, a monitoring consistency loss function is constructed, and the gradient thereof is embedded into an inverse diffusion sampling process as a guide item to dynamically correct the sampling state, and a reconstructed deformation field is generated. The application can realize high-precision and real-time deformation field reconstruction of a key section of a soil and rock dam under a sparse monitoring condition, and significantly improves the expression ability of the model to a complex deformation mode and the diversity of generated results.