1698 results about "Multiple factors" patented technology

The invention provides a pre-control and monitoring method for a whole process of actual grouting engineering based on digital geological model, comprising: extracting discontinuous fracture surfaces in a geologic body structure, to build a underground geologic body structure model; build a multi-source geologic body attribute model, optimizing and solving grouting simulations of a disaster high-risk region by using multiphase flow calculation method, and controlling grouting equipment by using optimized grouting parameters to carry out an actual grouting engineering; obtaining a multi-factor diffusion range of variable coefficients by adjusting physical values of the parameters in the grouting simulation and attribute data structure model by using a control variable method, learning and capturing complex mapping relationship between different parameters by using a neural network, optimizing the grouting parameters in the actual engineering in real-time, so as to realize the pre-controlling, monitoring and optimization of the whole process of the actual grouting engineering.

Disclosed is a method for evaluating the fatigue damage and life of a bridge structure under a multi-factor coupling effect. The method comprises the following steps: S1, reproducing complex service environment action values; S2, analyzing performance values of a bridge structure; S3, calculating coupling fatigue damage of a vulnerable part; and S4, predicting a remaining life of the bridge structure. By adopting the method, the fatigue damage condition and remaining life of an in-service bridge under a multi-factor coupling effect can be accurately evaluated, and the effect of each factor / action and the coupling effect the factors / actions in fatigue damage of a vulnerable part are quantitatively analyzed, thereby providing data support and decision-making reference for the operation and maintenance management, reinforcement and repair, and optimal design of bridges.

The invention discloses a power system load prediction method based on multivariate data fusion, and relates to the technical field of power systems. The problems that in the prior art, multi-source data fusion processing is complex, timeliness and processing efficiency are insufficient, and under the complex working condition, the response capacity to short-term emergencies is weak are solved. By collecting and preprocessing the multi-source data in real time, various factors influencing load changes can be comprehensively considered, the response capability to complex working conditions and emergencies is enhanced, the time sequence window is dynamically adjusted to improve the data validity and prediction performance, the accuracy and reliability of power system load prediction are improved, and the power system load prediction efficiency is improved. According to the method, multiple models are constructed and trained to obtain an accurate load prediction result, intelligent prediction and management of the load of the power system are realized through accurate load prediction, abnormal identification and processing are performed in combination with an emergency database, stable and reliable load prediction is provided for the power system, and thus power resource configuration is optimized.

The invention discloses a power load prediction method and device based on multi-factor influence, and particularly relates to the technical field of power load prediction. Historical weather data and power load data are collected, preprocessing and feature extraction are carried out, temperature change deviation features and power load lag features are extracted and input into a prediction model for training, the accuracy of a prediction result is evaluated, and when the prediction accuracy is low, the prediction result is subjected to prediction. The abnormal degree is further analyzed and predicted, and the power dispatching strategy is adjusted in time, so that the violent fluctuation of the load demand is dealt with, the stable operation of the power grid is ensured, the accuracy of power load prediction is effectively improved, and the risk of overload or collapse of the power grid is reduced especially in the case of extreme weather.

The invention provides an aircraft digital twinning situation deduction and risk assessment method in a complex meteorological environment, and belongs to the technical field of data processing, the aircraft digital twinning situation deduction and risk assessment method in the complex meteorological environment comprises the following steps: collecting and processing weather data covering parameters such as wind direction and wind speed; obtaining aircraft performance parameters and building a database; and inputting and analyzing the flight plan. By establishing a three-dimensional deduction model including dynamics, weather influence and environment models, aircraft situations under different weathers are deduced, and flight safety, performance and task risks are evaluated. The method has the advantages that weather factors are comprehensively and accurately analyzed, and influences are quantified through data acquisition and processing and a meteorological model; the system is suitable for various aircrafts, can integrate navigation and control systems, and realizes remote monitoring management; risk assessment is scientific and reliable, multiple factors are integrated, powerful support is provided for flight decision making, and aircraft safety and task success rate in complex weather are effectively improved.

The invention discloses a machine learning virtual power plant load prediction method based on multi-factor fusion, and the method comprises the following steps: S1, collecting multi-source heterogeneous data, including historical load data, meteorological data, calendar information and social activity data, and carrying out the preprocessing of the historical load data and the meteorological data; s2, a multi-head attention mechanism and a cross-modal learning technology are adopted to construct a multi-factor fusion model, and fusion vectors of historical load data, meteorological data, calendar information and social activity data are calculated; s3, constructing a multi-scale load prediction model, and constructing a training set for the multi-scale prediction model; s4, training the multi-scale prediction model by using the training set to obtain a trained multi-scale prediction model; and S5, performing power load prediction according to the trained multi-scale prediction model. The method not only can effectively process complex and changeable influence factors, but also has high-precision prediction and high adaptability.

The invention provides an intelligent inspection operation robot, which comprises a self-adaptive interaction assembly for collecting unstructured features in an inspection environment of the inspection operation robot in real time, converting continuous signals of the unstructured features into discrete interaction decision nodes, and generating an interaction parameter set capable of being iteratively optimized through a cross validation mechanism; the autonomous intention analysis component constructs a probabilistic decision-making program based on implicit features of an operator behavior track of the inspection operation robot, extracts a potential mode of an operation intention through nonlinear correlation analysis in combination with an interaction parameter set, and recognizes an interaction demand which is not clearly expressed; and the collaborative response component constructs a response decision mechanism with self-organizing characteristics, and jointly inputs an interaction parameter set and an interaction demand into the response decision mechanism by establishing a dynamic association rule between different interaction channels of the inspection operation robot and an operator to generate a composite interaction scheme. According to the method, the composite interaction scheme comprehensively considering multiple factors can be generated.

The invention discloses an intelligent science and technology project matching and pre-scoring system based on a large model, and relates to the technical field of project matching evaluation, and the system comprises a data collection module which collects multi-source science and technology project information; the data preprocessing module is used for cleaning, converting and standardizing the data; the large model construction module comprises a feature coding layer, a matching decision layer and a pre-scoring prediction layer; the intelligent matching module outputs a matching degree and similar items; a pre-scoring module; and a user interaction module, a model updating module, a security module and a privacy protection module. The large model fuses multiple technical training, each module has a specific function and algorithm, accurate and efficient operation of the system is guaranteed, and data security privacy is protected. According to the method, a large number of science and technology projects can be efficiently processed, the evaluation period is shortened, the matching accuracy and the pre-scoring reliability are improved, multiple factors are considered, the subjective influence is reduced, the system can be dynamically updated, the knowledge graph is updated, the management monitoring function is perfected, intelligent development of science and technology project management is promoted, and resource configuration is optimized.

The invention relates to the technical field of path planning, and discloses an unmanned aerial vehicle path planning method based on large-range cruise, which comprises the steps of geographic information acquisition, geographic information processing, regional discretization, starting point and target point setting, feasible path search and path optimization. Geographic information collection is comprehensive and accurate, data of terrains, obstacles, no-fly zones and the like are covered, and a foundation is laid for path planning; regional discretization fits the characteristics of an unmanned aerial vehicle, cells are divided according to the minimum turning radius and the effective range of a sensor, and flight safety and perception are guaranteed. Feasible path searching is efficient and reasonable, and the optimal path is searched by well-designed cost and heuristic functions and considering multiple factors; path optimization is effective, based on a B spline curve smoothing method, curvature and length are reduced, energy consumption is reduced, and abrasion is reduced; safety verification is added, collision and cross-zone risks can be accurately judged, and safety guarantee is enhanced.

The invention relates to the technical field of data processing, and discloses a motor vehicle safety performance classification method and system based on multi-factor coupling influence. The method comprises the steps of collecting multidimensional safety data of a motor vehicle, calculating a mutual information value and transfer entropy to generate a coupling relation network diagram, constructing a three-layer safety index system to form a comprehensive score, obtaining low-dimensional features through phase-space mapping, determining a dynamic boundary value by applying density clustering, and classifying real-time states to output risk early warning information. According to the technical method for realizing accurate classification and early warning, the defects of neglect of factor coupling relation, lack of scene adaptability, insufficient dynamic prediction capability and the like in the prior art are overcome, and the accuracy and practicability of safety performance evaluation and early warning of the motor vehicle are improved.

The invention relates to the technical field of information, and discloses a multi-factor space-time coupling water environment prediction method, which comprises the following steps: acquiring water environment internal data and water environment external data; preprocessing the water environment internal data and the water environment external data to obtain preprocessed multi-source data; the preprocessed multi-source data serve as a training set to be input into a random forest regression model, importance selection is conducted on the multi-source data, an influence factor set with the minimum error of the random forest regression model is selected, and the influence factor set serves as an optimal feature selection set; and taking the optimal feature selection set as a training set, inputting the training set into a least square support vector machine model, training the least square support vector machine model, obtaining a multi-factor coupled water quality prediction model, and obtaining a water environment prediction result based on the multi-factor coupled water quality prediction model. According to the method, the dimensionality of the least square support vector machine training model is reduced, and the prediction precision is improved.

The invention relates to an intelligent maintenance decision method and system based on a knowledge graph, and relates to the technical field of intelligent operation and maintenance and decision support and information, and the method comprises the steps: obtaining the operation state data and historical maintenance data of industrial equipment; constructing an industrial knowledge graph and generating an incidence relation model; an optimal maintenance path is obtained based on the analysis of the relationship between the fault type and the maintenance path; optimizing a maintenance task allocation scheme through the optimal maintenance path, the incidence relation model and the maintenance resource constraint condition; carrying out maintenance time window dynamic adjustment analysis to obtain an optimal maintenance strategy; and finally, performing multi-objective optimization iterative analysis to obtain a final maintenance decision. According to the scheme, various factors can be comprehensively considered, and the scientificity and the accuracy of maintenance decision making of the industrial equipment are improved.

The invention provides a north pole atmosphere coupling forecasting method capable of automatically fusing sea ice concentration and thickness, which belongs to the technical field of meteorology, and comprises the following steps: firstly, acquiring GFS background field data and correcting by applying an ice-sea heat exchange correction equation; and then sea ice concentration and thickness information is extracted, and physical consistency verification is carried out based on a thermodynamic equilibrium equation. Performing multi-element fusion through an Arctic ice sea assimilation model, calculating the sea ice surface temperature by applying a polar region sea ice heat flux equation, and establishing a sea ice concentration thickness boundary layer interaction equation set to describe the modulation effect of sea ice on a boundary layer; the method comprises the core steps of performing deep fusion on various sea ice features by applying a feature fusion model, adaptively adjusting the weight based on an ice-gas interface balance index, generating an optimized mode initial field and boundary conditions, and finally setting a parameter scheme to execute an integral program to obtain a forecasting result. The technical problem of low forecasting accuracy caused by insufficient physical consistency of sea ice information is solved.

The invention relates to the technical field of geological monitoring, and discloses a self-adaptive phase unwrapping foundation SAR deformation monitoring method and system. The method comprises the steps of collecting ground-based SAR echo data, extracting phase gradient features and noise distribution, constructing a model based on a compressed sensing theory to solve a phase unwrapping result, establishing a deformation field dynamic evolution model fusing multiple factors, and generating a deformation monitoring value through an adaptive weighted fusion algorithm. And an abnormal deformation detection model is also constructed to realize adaptive re-calibration of model parameters, and a multi-sensor collaborative calibration network is constructed to improve the monitoring precision. The problem of phase unwrapping is effectively solved, multiple influence factors are comprehensively considered, surface deformation can be accurately monitored in real time, abnormity can be timely found, the model can be calibrated, multi-sensor data can be fused, and the method has important application value in the fields of geological disaster early warning, engineering structure safety assessment and the like.

The invention provides a regulation factor threshold effect evaluation method and device for vegetation response sudden drought characteristics, equipment and a storage medium. Relates to the technical field of drought risk prediction and influence evaluation. The method comprises the following steps: acquiring hydro meteorological data, vegetation productivity data, topographic feature data, soil characteristic data and biological variable data of a target area; constructing a drought suddenness index, extracting a drought suddenness event according to the drought suddenness index, and determining a drought suddenness feature of the drought suddenness event; extracting vegetation productivity abnormal data before and after the sudden drought event, and determining the response characteristic difference of the vegetation to sudden drought; the method comprises the following steps: determining influence factors of vegetation on sudden drought response characteristics, fusing XGBoost, a random forest regression model and an SHAP interpretability analysis method, analyzing regulatory factors of the vegetation response sudden drought characteristics, and determining a threshold effect of the regulatory factors. According to the method, the nonlinear cooperative regulation effect of multiple factors on vegetation response sudden drought characteristics can be disclosed more comprehensively.

The invention discloses an engineering quality safety risk intelligent assessment method and system, relates to the technical field of engineering quality safety assessment, adopts multiple types of sensors to collect data, breaks through the limitation that the traditional structure monitoring only depends on a single index to judge the risk, and realizes cross-parameter and cross-component structure state comprehensive perception. The problem that the structure risk is commonly evolved by multiple factors is solved, and the overall recognition precision is improved. A sliding window and a periodic variable coefficient calculation mechanism are introduced, and the system can dynamically extract fluctuation characteristics of structure response and identify different evolution forms of short-time fluctuation, medium-term creep and long-term mutation. And the defects of insufficient abnormal precursor identification capability and serious hysteresis response in the traditional method are overcome. According to the method, an abnormal mutation index AEI and an evolution acceleration factor EAF are constructed, the formation process and the development path of the risk are identified no longer only on the basis of whether a certain moment value exceeds a threshold value or not, and the risk process perception capability is improved.

The invention provides an aircraft digital twinning situation deduction and risk assessment method in a complex meteorological environment, and belongs to the technical field of data processing, the aircraft digital twinning situation deduction and risk assessment method in the complex meteorological environment comprises the following steps: collecting and processing weather data covering parameters such as wind direction and wind speed; obtaining aircraft performance parameters and building a database; and inputting and analyzing the flight plan. By establishing a three-dimensional deduction model including dynamics, weather influence and environment models, aircraft situations under different weathers are deduced, and flight safety, performance and task risks are evaluated. The method has the advantages that weather factors are comprehensively and accurately analyzed, and influences are quantified through data acquisition and processing and a meteorological model; the system is suitable for various aircrafts, can integrate navigation and control systems, and realizes remote monitoring management; risk assessment is scientific and reliable, multiple factors are integrated, powerful support is provided for flight decision making, and aircraft safety and task success rate in complex weather are effectively improved.

The invention relates to the technical field of building curtain wall keel structure design, and discloses a curtain wall keel structure intelligent optimization design method, which comprises the following steps: firstly, obtaining geometric parameters, load distribution data and material performance indexes, and verifying the geometric parameters, the load distribution data and the material performance indexes; analyzing the verified data to generate multi-dimensional structure characteristic parameters, and constructing a structure optimization reference model; keel structure data are verified in real time according to the model, and abnormal matching of node displacement and stress distribution is quantitatively evaluated; according to a verification result, generating a structural anomaly index to judge a design defect; and when defects exist, optimization design schemes including keel section adjustment, node reinforcement and the like are automatically generated and sequenced. According to the method, multiple factors are comprehensively considered, multi-objective optimization is achieved, the curtain wall keel structure design quality and efficiency can be improved, the structure safety is guaranteed, and the cost is reduced.

The invention relates to a multi-factor fusion Internet of Things outlier data identification method, and belongs to the technical field of Internet of Things, and the method comprises the steps: determining an identified target factor, fusing a plurality of influence factors associated with the target factor, obtaining corresponding historical data, and carrying out the preprocessing; performing correlation analysis on the specific indexes, and selecting index data having significant influence on target factors to form a final effective index combination; constructing a prediction model combining the LSTM with a feature attention mechanism, performing prediction by using the data after correlation analysis screening, outputting a prediction value in a future period of time, and obtaining an actual prediction value through reverse normalization; a prediction model is adopted, an outlier data identification method fused with multi-factor comprehensive analysis is adopted, an isolated forest algorithm is used for modeling, prediction residual errors and influence factors serve as input features, and the abnormal score of each data point is calculated; and based on the distribution of the abnormal scores, dynamically setting a threshold value by adopting a sliding window technology, and judging the data points exceeding the threshold value as outlier data.

The invention discloses a typhoon-rainstorm flood composite disaster monitoring and evaluating method and system, and belongs to the field of intelligent disaster monitoring, and the method comprises the following steps: building a disaster recognition model through a machine learning algorithm based on multi-factor data, and obtaining a composite disaster judgment result; based on the composite disaster judgment result, constructing a disaster evaluation model for the area forming the disaster; dynamically updating parameters of the disaster assessment model based on real-time monitoring data to obtain a real-time composite disaster assessment result; based on the real-time composite disaster assessment result, dynamically adjusting loss probability distribution to obtain a comprehensive disaster damage assessment result; and carrying out disaster situation space-time evolution based on the comprehensive disaster damage assessment result to obtain a disaster situation development trend.

The invention discloses a loess slope stability intelligent monitoring method and system, and the method comprises the steps: carrying out the regional differentiation modeling of multi-source data of a target region, and outputting slope change data; according to the period-by-period displacement time sequence data, a slope deformation trend is obtained through prediction based on an LSTM algorithm; and inputting the slope change data, the slope deformation trend and the meteorological rainfall data into a slope stability evaluation model to predict a safety coefficient, and if the safety coefficient is lower than a preset threshold value, issuing early warning information. According to the loess slope stability intelligent monitoring method and system provided by the invention, data such as gradient change and deformation trend are accurately extracted, multi-source data and meteorological rainfall data are fused by means of cross-modal attention, a random forest model is used for learning a multi-factor coupling rule, and finally multi-factor data prediction is integrated to obtain a safety coefficient for early warning. The problem of low early warning reliability caused by difficulty in accurately evaluating the slope stability by fusing multiple factors such as gradient, deformation and meteorological rainfall can be solved.

The invention relates to the technical field of electrical digital data processing, and discloses a statistical entropy and multi-factor fused text similarity data processing method, which comprises the following steps that: a processor extracts substring sets which do not contain maximum common values of a first data sequence and a second data sequence, and calculates the quadratic sum of the lengths of substrings to generate local statistical entropy; traversing the maximum common substring set to obtain storage address indexes of the maximum common substring set in the first data sequence memory space and the second data sequence memory space, and constructing a topological mapping vector of a mapping structure displacement relationship; calculating the total number of inverted pairs of the topology mapping vector by using a merge sorting algorithm, and generating a normalized topology dissipation index; and by taking the local statistical entropy as an information carrier and taking the topological dissipation index as a structural damping factor, executing nonlinear damping modulation operation to obtain a final similarity score, and solving the technical problem that the block-level displacement cannot be identified by linear scanning logic by quantizing topological entropy increase of data distributed in a storage space.

The invention relates to a method for setting power loss of a mountain photovoltaic array. The method comprises the following steps of: establishing a coupling causal spectrum of multi-source factors including terrain, angle, sunlight, temperature, dust and wiring mode to component output; identifying a dominant loss path based on an identification method of a Bayesian network; outputting a structure weight matrix of the power loss, and identifying an inclination angle gt; a temperature gt; a set of influence factors of the wiring topology pair setting; calculating a power loss response index of each component based on the loss path weight; forming a response index thermodynamic diagram by using a space weight interpolation algorithm, and performing array space sensitive partitioning; the high response area is used as a preferential setting target, and the low response area enters an observation maintenance mechanism; by constructing the response index and the correction response potential function based on multi-factor coupling, the power response potential of the component to the setting operation can be accurately identified, the problems of average adjustment and low-efficiency resource allocation in a traditional setting method are effectively avoided, and it is ensured that setting resources are intensively input into a high-response area.

The invention discloses a fruit and vegetable logistics supply chain management system based on aging analysis, relates to the technical field of fruit and vegetable logistics management, and solves the technical problems that a traditional logistics path selection method is single, multiple factors are not comprehensively considered, an optimal path is difficult to select, and an effective real-time monitoring and adjusting mechanism is not available in the transportation process. According to the invention, when the logistics path is generated, factors such as the traffic flow, the vehicle speed and the slow moving distance of different paths between the logistics starting point and the logistics ending point are comprehensively considered, the path with the minimum logistics value is selected as the standard logistics path by calculating the path value, and the remaining path in the standard logistics path can be monitored in real time in the transportation process; for different types of fruits and vegetables, the fresh-keeping time, namely the fruit and vegetable aging time, is fully considered. In the route selection and transportation process, the fruit and vegetable aging duration is always used as an important reference index, and it is ensured that the selected route and transportation scheme can complete the transportation task within the fruit and vegetable fresh-keeping period.

The invention relates to the technical field of numerical control lathe control, discloses a high-precision numerical control lathe Y-axis dynamic adjustment drilling reamer holder control method, and aims to improve the precision and efficiency of Y-axis motion control. According to the method, Y-axis motion data, machining parameters, tool abrasion and workpiece material information in the historical machining process are collected, and a motion prediction model is constructed. And acquiring current processing state data in real time, and inputting the model to predict a Y-axis movement track. Meanwhile, Y-axis real-time motion data are obtained and compared with a predicted track, and deviation points are marked. And then, a Y-axis dynamic adjustment decision model is constructed, and the adjustment process is modeled as a Markov decision process. And performing real-time control based on the trained decision model, selecting an optimal adjustment action according to the current processing state and strategy, and automatically executing the optimal adjustment action. According to the method, multiple factors are comprehensively considered, the prediction accuracy and the adjustment flexibility are improved, and the machining quality and the production efficiency are guaranteed.

The invention discloses a water conservancy and hydropower project ecological scheduling optimization method, and relates to the technical field of water conservancy and hydropower ecological scheduling, and the method comprises the steps: carrying out the fine ecological region division of a river channel, and dividing the river channel into a plurality of regions with definite ecological functions through combining a plurality of factors. By monitoring the ecological water demand of different ecological regions in the river channel in real time, the actual demand condition of each ecological region is accurately mastered, so that the ecological scheduling scheme is individually formulated according to the characteristics of different ecological regions, the scheduling accuracy is improved, the ecological model comprehensively considers multiple factors such as hydrology, water quality, ecology and meteorology, and the ecological scheduling efficiency is improved. According to the method, the change trend of the ecological water demand of different ecological areas can be predicted, a suitable living environment is provided for various aquatic organisms by accurately meeting the ecological water demand of different ecological areas, protection of rare and endangered species is facilitated, stable development of biocenosis is promoted, and the biodiversity level of a whole river ecosystem is improved.

The invention relates to the technical field of green intelligent building planning and design, and discloses a green intelligent building multi-factor fusion planning and design method. The method comprises the steps of firstly obtaining geographic information data and ecological environment parameters of a building site, collecting peripheral multi-dimensional environment data and performing fusion processing, constructing a three-dimensional space topology model, and generating an initial space planning model; and then climate simulation and resource circulation conditions are loaded, a comprehensive planning analysis model is formed, and dynamic collaborative evaluation indexes are output. A multi-factor coordination model and key coordination parameters are obtained by constructing a multi-objective optimization pre-training framework for iterative learning, and then a system collaborative simulation result is generated. And establishing a configuration model to optimize facility deployment parameters to obtain an optimal configuration scheme. And finally, dynamically adjusting parameters according to the efficiency evaluation framework to realize planning control of the green intelligent building. According to the method, multiple factors are integrated, optimization of building planning and design is achieved, and the sustainability and the intelligent level of the building are improved.

The invention provides a factory power load monitoring and early warning method based on digital twinning, and belongs to the technical field of digital twinning and power load prediction, WTConv and i Transform are combined for power load prediction, and the challenge that a traditional power load prediction method is difficult to deal with in a complex dynamic factory environment is solved; in the actual factory operation process, the power load is influenced by multiple factors such as the production rhythm, the equipment operation state and the external environment, so that load data have high time-varying characteristics and nonlinear characteristics; through the combination of WTConv and iTransform, the method can better capture a complex space-time mode and multi-scale features, thereby remarkably improving the accuracy and robustness of power load prediction, providing more reliable load prediction support for a power system, helping enterprises to optimize production scheduling, reducing energy loss, and effectively preventing the load risk of the power system.

The invention relates to a high-dimensional statistics and hydrodynamic force coupled flood risk assessment method and system. The method comprises the following steps: acquiring historical observation data covering river flow, tide level and rainfall events; constructing a univariate marginal distribution function; connecting the univariate marginal distribution functions of the river flow, the tide level and the rainfall event, and simulating a joint probability distribution function; coupling the joint probability distribution function with the mechanism model to obtain a probability-mechanism coupling model; inputting the determined joint probability boundary condition into a probability-mechanism coupling model, simulating to obtain composite flood scenes in different return periods, and comparing the composite flood scenes with single variable flood scenes in different return periods; and quantizing the contribution of each variable to the change of the submerging area by using the obtained submerging area and the corresponding rainfall event, river flow and coastal tide level data. According to the method, the defect that traditional single-factor evaluation cannot capture multi-factor interaction is overcome, and the accuracy of flood disaster evaluation is greatly improved.

The invention discloses an ice condition prediction method based on multi-feature fusion and physical constraint, and belongs to the technical field of hydrological forecasting. Firstly, various types of data of a target area are collected and preprocessed to serve as a data set, features of the various types of data are extracted, feature fusion is conducted on obtained image features, time sequence features and environment features through a multi-head attention mechanism, and a fusion feature vector is generated. Secondly, adopting a physical information neural network model, taking the fusion feature vector as input, taking ice thickness and ice stress as output layers, carrying out constraint by using a composite loss function, carrying out model optimization by using a verification set, carrying out processing through a full connection layer in the network, and carrying out end-to-end training and regularization of a prediction model; and finally, evaluating the final prediction model obtained by training through the test set, and verifying the effectiveness and generalization ability of the final prediction model. The method combines multi-source and multi-mode observation data, can effectively capture complex relations between ice surface changes and various factors, can improve prediction precision, stability and reliability, and can improve model training efficiency and generalization ability.