669 results about "Predictability" patented technology

The invention relates to the technical field of real-time data processing, in particular to a distributed real-time data stream processing system and method, and the system comprises a state mapping modeling module, a load dynamic calculation module, a node task adjustment module and a resource scheduling execution module. According to the method, the dynamic identification and classification of the flow state are realized by analyzing the data flow change sequence and identifying the frequency and trend of state switching, the monitoring and understanding capability of real-time data is improved, the node resource pressure is accurately quantified, the predictability of load balancing is enhanced in the resource occupation and load range analysis, and the load balancing efficiency is improved. A scientific basis is provided for resource scheduling, task priority ranking and dynamic task adjustment are combined, the node computing resource utilization rate and the task processing response speed are optimized, flexible configuration of internal storage and computing resources is achieved, peak load pressure is relieved, and data processing efficiency and system stability are improved. And the adaptability and the reliability of processing large-scale data streams by the system are improved.

The invention relates to the technical field of electric power government affair management, and provides a cross-department government affair big data business co-processing method based on a computing power platform and data fusion, comprising the following steps: S1, deploying a heterogeneous computing power cluster, and establishing a computing power resource dynamic scheduling data processing platform comprising a CPU, a GPU, an FPGA and an edge computing node; s2, acquiring multi-dimensional data of power grid equipment data, power grid load data, industry power consumption data and government policy data; and S3, designing a cross-department data access system based on the zero-trust architecture, and developing a multi-dimensional model of a power grid risk prediction model, an industry energy consumption analysis model and a carbon emission accounting model in power management. The space-time joint probability prediction and conflict priority ranking algorithm is applied to power-government affair emergency disposal, intellectualization and precision of cross-department business collaboration are achieved, power faults with serious consequences are rapidly handled according to the priority level, and the harmfulness and unpredictability of the power faults can be reduced.

The invention relates to a coal mine safety risk intelligent management and control method, device and equipment and a medium. The method comprises the following steps: constructing a multi-source heterogeneous coal mine safety data system according to received data of a target coal mine park; the multi-source heterogeneous coal mine safety data system comprises static structure data, dynamic environment data, personnel behavior data and management data; constructing a coal mine three-dimensional space model according to the static structure data and the dynamic environment data; risk indexes in the dynamic environment data are extracted based on multi-algorithm fusion for evaluation, and a risk level is obtained; and if the risk level reaches a preset threshold value, triggering a corresponding linkage response mechanism, and forming a visual result in the coal mine three-dimensional space model. By the adoption of the method, closed-loop logic from sensing, evaluation to linkage treatment can be achieved, and the real-time performance, predictability and controllability of coal mine safety management are effectively improved through algorithm support of each stage and fine design of implementation details.

The invention discloses a lower limb motion data real-time analysis method and system applied to rehabilitation guidance, and the method comprises the steps: calling a personalized model according to user information, converting real-time motion data into historical motion feature vectors, inputting the historical motion feature vectors into a real-time prediction module, and generating multiple frames of future motion feature vectors through coding and decoding; calculating a deviation value between each future motion feature vector and a preset threshold range to obtain a motion deviation value, calculating a motion risk proportion of each future motion feature vector, performing weighted fusion to obtain a motion risk probability, and generating an equipment parameter adjustment value and pose adjustment information through reverse mapping; and finally, according to risk probability assessment training stage promotion, synchronously displaying correction information and adjusting equipment operation parameters. According to the method, the dynamic model is constructed through pre-training, real-time prediction of future motion features is realized in combination with the lightweight LSTM, and the predictability and personalized adaptation ability of rehabilitation guidance are improved.

The invention discloses an integrated collaborative management method for construction site risk operation data, and belongs to the technical field of data management. The method comprises the following steps: constructing and calibrating a space-time risk reference database; deducing a potential risk conduction link set, and generating an associated intervention knowledge base; during operation, through multi-sensor cooperative verification, an abnormal signal is confirmed as a risk event; matching the risk event with a conduction link to calculate a risk upgrade level and dynamically adjust an early warning threshold; and when the early warning is triggered, generating and issuing a dynamic collaborative response instruction, and feeding back a processing result to correct the reference database to form a management closed loop. According to the method, the technical means of constructing the space-time risk reference, deducing the conduction link, cooperatively verifying the risk and dynamically regulating and controlling the threshold are adopted, so that the predictability of project risk management and control, the efficiency of resource cooperative scheduling and the scientificity of overall management decision are improved.

The invention discloses a power business scheduling method and system based on 5G communication grading and classification, and relates to the technical field of 5G slicing. According to the method, the prediction model and the real-time monitoring data are utilized to effectively predict the network flow and dynamically adjust the network resources according to different levels of power business requirements, so that the power business with high priority can quickly obtain necessary network support, the quality of the overall network service and the operation efficiency of the power business are improved, and the network flow is effectively predicted. And network resource allocation is optimized through the graph neural network, accurate analysis and response to power service grading are realized, predictability and stability of the network are enhanced, and more complex power service application is supported.

The invention relates to the technical field of data processing, and discloses an operation state monitoring method and system for vibration test equipment. The method comprises the following steps: generating a multi-dimensional physical characteristic data stream through vibration waveform calibration according to operation parameters collected by each sensor of vibration test equipment; performing frequency response analysis on the data stream to obtain physical response characteristics; based on this, a test calibration reference is established; outputting test state parameters according to the reference analysis real-time response characteristics; and performing performance attenuation analysis according to the parameters, and generating an equipment quality evaluation report. According to the comprehensive analysis method based on the multi-dimensional physical characteristic data flow, real-time monitoring, multi-parameter joint evaluation and performance attenuation trend analysis of the dynamic characteristics of the equipment are achieved, so that the accuracy, comprehensiveness and predictability of state monitoring of the vibration test equipment are improved, and the reliability and consistency of test results are ensured.

The invention discloses an AI-NWP three-dimensional closed-loop bidirectional dynamic feedback coupling method for extreme rainfall event area simulation and a program product, and belongs to the technical field of meteorological numerical simulation and artificial intelligence fusion. According to the method, a mid-term forecast field is generated based on an AI global weather prediction model, an analysis field is constructed by assimilating multi-source observation data, and a high-resolution NWP region mode is driven to perform rolling simulation. Furthermore, a space-time residual field is constructed by using the difference between an NWP region simulation result and live data, a residual learning model is trained, an AI model prediction structure is fed back and corrected, and dynamic weight adjustment updating of the AI prediction model is realized. A closed-loop two-way feedback system among AI output, NWP simulation and residual evaluation is integrally formed, the space structure reduction capability and the area positioning precision of a medium-term heavy rainfall event are effectively improved, the continuous predictability and the simulation credibility of an extreme weather process are remarkably enhanced, and the method has good stability, universality and engineering expansion value.

The invention discloses a transaction platform management system based on cloud service, and relates to the technical field of cloud computing, and the system comprises the following components: a user behavior analysis module which collects the browsing track, purchase record and evaluation information of a user through an e-commerce platform, collects the attribute data and sales data of a commodity, and sends the collected data to a server; the user behavior analysis module accurately extracts key features of user activeness and commodity popularity to provide powerful support for personalized recommendation, the intelligent recommendation module generates a personalized recommendation list based on real-time behavior data and interests of the user, and the intelligent customer service module realizes personalized recommendation through natural language processing and machine learning technologies. According to the system, the user problem is quickly and accurately responded, the customer service efficiency and the user satisfaction degree are remarkably improved, the predictive maintenance module identifies potential risks and abnormities in advance through real-time data analysis, an early warning and solution scheme is provided for an administrator, and the system failure rate and the maintenance cost are effectively reduced.

The invention discloses a multi-warehouse intelligent distribution and scheduling method for raw material inventory. The method comprises the following steps: collecting raw material inventory data of each warehouse, and monitoring the inventory data of the warehouses in real time; based on the real-time inventory data and the historical inventory data, the inventory state of each warehouse is analyzed to determine the inventory balance state of each warehouse, and the warehouses with excess or short inventory are identified; and calculating an optimal inventory distribution scheme by adopting a dynamic adjustment algorithm based on an inventory state analysis result. According to the invention, by collecting the real-time inventory data and the historical inventory data, the inventory state of each warehouse is deeply analyzed, the inventory balance state is determined, and the warehouse with excess or short inventory is identified. Through the real-time data monitoring and analysis, the system can provide accurate inventory early warning, help managers to take measures in time, avoid production interruption and resource waste, and improve the accuracy and predictability of inventory management.

The invention relates to the technical field of fault detection of the Internet of Things, in particular to a valve path detection method and system based on the Internet of Things, and the method comprises the following steps: based on real-time data of a valve path, removing high-frequency noise and high-frequency components in temperature fluctuation through a low-pass filter, carrying out the mean value smoothing operation of pressure data, and carrying out the mean value smoothing operation; and performing data standardization conversion on the flow data to obtain valve operation state data. According to the method, through real-time collection and optimization processing of valve operation data, it is ensured that the data is accurate and stable, weighted calculation is carried out in combination with the data influence degree, the representativeness of state feature extraction and the perception ability of working condition changes are enhanced, and based on state feature trend analysis, the sensitivity of fault detection is improved; and health state evaluation is carried out in combination with the operation duration and the load change, so that health state scoring is realized, the predictability of the long-term operation state is improved, false alarms and missing alarms are reduced, and the accuracy of fault early warning and the equipment health monitoring capability are enhanced.

The invention discloses a method for predicting the risk of a valley-type tailing pond being threatened by extra-small watershed flood, and the method comprises the steps: S1, obtaining the soil type, vegetation coverage and gradient of the extra-small watershed of the valley-type tailing pond, and the spatial distribution data of a valley, and building an underlying surface classification system; s2, non-linear key inducements in the rainfall-runoff process in the valley terrain extra-small watershed are extracted, dominant factors influencing runoff formation are recognized, and a non-linear confluence calculation numerical model is constructed; s3, constructing a distributed sensing monitoring network through a rainfall monitoring station and a runoff monitoring station; s4, constructing a layered three-dimensional hydrological hydrodynamic model of coupling rainfall runoff generation-slope confluence-shallow groundwater flow; and S5, establishing a risk assessment index system and a risk prediction model through a layered three-dimensional hydrological hydrodynamic model in combination with flood risk influence factors. According to the method, accurate prediction of the valley-type tailings pond under the flood threat in the extra-small watershed is realized, and the problem of poor predictability of the flood threat in the extra-small watershed is solved.

The invention discloses a boiler state monitoring and predicting system based on artificial intelligence, and relates to the technical field of state monitoring. The system comprises a data acquisition part, a combustion stability analysis part, a hydrodynamic instability risk assessment part, a steam drum water level fluctuation prediction part and a fault probability and service life assessment part. The data acquisition part is used for acquiring boiler operation data and water circulation data; the combustion stability evaluation part is used for calculating a thermodynamic efficiency index; the combustion stability analysis part is used for calculating a combustion stability coefficient; the hydrodynamic instability risk assessment part is used for calculating a hydrodynamic instability risk index; the steam drum water level fluctuation prediction part is used for calculating and predicting steam drum water level fluctuation; and the fault probability and service life evaluation part is used for evaluating the fault probability and the residual service life of the boiler. According to the invention, the comprehensiveness, accuracy and predictability of boiler operation state monitoring are obviously improved.

The invention discloses an intelligent production scheduling method and device for the equipment manufacturing industry, and relates to the technical field of production scheduling, and the method comprises the steps: building an equipment file information library, and recording the information of active equipment and standby equipment of a target production line; and constructing an equipment topology network according to the equipment file information base and the process information of the production line. Configuring a lightweight prediction model for each edge equipment node, and performing equipment health state analysis and short-term maintenance demand prediction in combination with real-time equipment operation data to obtain a preliminary evaluation state; and performing evaluation state correction through data sharing of edge equipment nodes to obtain a final evaluation result of the equipment. And according to an evaluation result and an equipment archive library, generating an equipment scheduling and maintenance decision, and realizing intelligent production scheduling. The technical problem that equipment maintenance and production scheduling are lack of flexibility and predictability in the prior art is solved, and the technical effects of improving the equipment utilization rate, optimizing production scheduling and ensuring efficient operation of a production line are achieved.

Systems and methods facilitate saving data storage through data storage management of storage location(s) by training, via an iterative training and testing loop, a predictive model using training data to detect data redundancies from two or more datasets stored to data storage location(s), the training includes testing the predictive model by predicting a target variable and iteratively adjusting weights and calculations during each subsequent iteration to improve predictability of the target variable, where the predictive model is trained to identify data similarities among the two or more datasets. Based on any error in predicting the target variable being less than a predetermined level, the predictive model is deployed and applied to at least two datasets to quantify a percentage of similarity among the at least two datasets. If it is determined the percentage of similarity surpasses a predefined threshold percentage, then an electronic notification is transmitted.

The invention relates to the technical field of electronic equipment detection, and discloses an FPC connector quality detection method and system based on multi-modal fusion, and the method comprises the steps: obtaining an original multi-modal data set containing electrical stress, thermal stress and mechanical stress, and carrying out the filtering and time sequence alignment preprocessing, and obtaining a standardized multi-physical field sequence; constructing a three-dimensional space field intensity distribution diagram according to the sequence, and identifying field intensity anomalies by adopting density clustering so as to accurately position potential defects; learning a time sequence evolution mode based on the multi-moment observation value of the defect position to predict the future diffusion trend of the defect; and integrating the diffusion trend, the real-time field intensity and the environment variable to carry out comprehensive modeling and evaluation, and generating a dynamically updated and final quality detection report. According to the method, root cause diagnosis of connector defects, accurate three-dimensional space fault positioning and accurate prediction of future evolution trends can be realized, and the detection precision and predictability are remarkably improved.

The invention discloses a soft start control method for intelligent temperature control of a glue injection mold, and particularly relates to the technical field of glue injection molds. Initial temperatures of a plurality of heating areas of the mold are collected, and a temperature distribution model is constructed; calculating a thermal inertia score value based on the thermal response characteristic and the thermal capacity parameter of each region; predicting temperature rise rates of different areas by using the thermal inertia score value, the thermal coupling degree abnormal value and the historical control deviation frequency, and setting differentiated soft start amplitude limiting parameters; temperature feedback is collected in real time in the heating process, a temperature rise rate error sequence is constructed, and the future deviation risk is predicted through the dynamic Bayesian network; when the risk value exceeds the limit, the PWM duty ratio or the conduction angle of the heating area is dynamically corrected, and power output self-adaptive adjustment is achieved; the method has the advantages of being high in predictability, accurate in response and intelligent in control, and is suitable for high-precision heating control application of the complex glue injection mold.

The invention discloses a water plant whole-process intelligent optimization scheduling method based on adaptive feedback control. The method comprises the following steps: S1, dividing into a low-demand stage, a medium-demand stage and a high-demand stage; s2, in a low-demand stage, a medium-demand stage and a high-demand stage, correspondingly obtaining real-time monitoring data sets of the water plant; s3, constructing a multi-stage evolutionary game model based on the water plant real-time monitoring data set; s4, generating a local optimal scheduling scheme of the current stage; s5, outputting a local optimal scheduling scheme according to the evolutionary game model by utilizing a self-adaptive feedback control mechanism; s6, performing global optimal scheduling of the system in each stage; s7, optimizing a scheduling scheme in advance; and S8, displaying the optimized scheduling effect through the integrated intelligent scheduling platform, monitoring the operation state of the whole water plant, and generating a scheduling report for reference of management personnel. According to the invention, the predictability and perspectiveness of the scheduling system are effectively improved through the prediction model, and it is ensured that the water plant is always in an efficient and safe operation state.

The invention discloses a multi-agent task arrangement method and system, and relates to the technical field of artificial intelligence. In the method, firstly, a task demand is obtained, key information and intention of the task demand are extracted, and task semantics are obtained; secondly, based on the task semantics and the business standard process template, obtaining a business standard process template matched with the task demand; then, according to the matched business standard process template and task semantics, a task context is constructed, and a domain specific language DSL is generated through reasoning according to the task context; thirdly, the DSL is analyzed, a task dependency graph is constructed according to the DSL analysis result, and the task dependency graph is converted into BPMN process model data; and finally, sub-task scheduling and execution of the sub-agents are carried out according to the BPMN process model data. According to the method provided by the invention, the dependence on professional developers or process engineers can be reduced, the response and planning time of complex tasks can be shortened, the result predictability can be improved, and the reasoning cost can be obviously reduced.

The invention relates to the technical field of computer vision, in particular to a station platform limit intelligent detection method and system based on computer vision, and the method comprises the following steps: obtaining platform region spectral data, extracting spectral features according to the reflectivity value of each pixel point in a spectral range, comparing the spectral features with the features of various known materials, and determining the position of the station platform limit. And identifying material types of the plurality of positions, detecting obstacles, and generating material classification feature values. According to the method, the spectral data is acquired, the reflectivity value is compared to identify the material type, the texture feature and the gradient change rate are utilized to analyze and identify the fuzzy area, misjudgment on the fuzzy identification is reduced, point cloud data and image information are adopted for time-space synchronous registration, and multi-source sensing information is fused to enhance the time-space consistency of three-dimensional reconstruction. The size accuracy of the platform model is improved, the space offset is calculated by comparing engineering design drawing data, a structure deviation quantitative evaluation system is established, a deformation area is positioned, and the foreseeability of safety early warning is improved.

The invention relates to the technical field of computers, and discloses a cross-region advertisement putting analysis and decision method and system based on artificial intelligence, and the method comprises the steps: responding to a cross-region advertisement putting instruction, carrying out the analysis of the cross-region advertisement putting instruction, and determining a target advertisement putting region in the cross-region advertisement putting instruction; obtaining multi-dimensional data of the target advertisement putting area; performing semantic analysis and code conversion on the multi-dimensional data, and constructing a structured regional data matrix; feature extraction is performed on the regional data matrix based on a preset deep learning algorithm, and potential features related to the advertisement putting effect are mined; and inputting the potential features into an advertisement putting decision model to obtain an advertisement putting strategy output by the advertisement putting decision model for the target advertisement putting area. According to the method, the problems that the dynamic change and potential trend of the market cannot be accurately predicted and the advertisement putting strategy is lack of accuracy are solved, and the accuracy and predictability of cross-region advertisement putting analysis decision are improved.

The invention discloses a terrestrial heat development risk monitoring method and device combined with multidimensional data analysis, and relates to the technical field of terrestrial heat development, and the method comprises the steps: laying an Internet of Things sensing topology network of a target terrestrial heat development area, and the Internet of Things sensing topology network comprises K topology nodes; historical abnormal monitoring data of the K nodes are collected, and an abnormal data set is obtained. And according to the monitoring time nodes, performing neighbor data interval centralized identification, and determining the abnormal monitoring scale of the nodes. And constructing a self-adaptive convolutional network layer, and extracting monitoring data in sequence to obtain a global perception monitoring topological data set. And performing feature recognition on the set to obtain a global perception monitoring topological feature set. And performing neighbor matrix interaction analysis to obtain an interaction perception monitoring feature set. And calling the risk monitoring network to carry out cross-scale feature fusion analysis to obtain a risk monitoring result. Therefore, the technical effect of improving the risk monitoring accuracy and predictability is achieved.

The invention relates to the technical field of data testing, in particular to an intelligent test analysis system and method for an integrated circuit. The method comprises the following steps: acquiring circuit monitoring data; performing data preprocessing on the circuit monitoring data to generate circuit monitoring preprocessing data; performing multi-dimensional feature extraction on the circuit monitoring preprocessing data to generate a circuit monitoring multi-dimensional feature data set; performing feature set fusion on the circuit monitoring multi-dimensional feature data set to generate a circuit monitoring feature fusion data set; therefore, the problems of insufficient precision, response delay and high maintenance cost in traditional circuit monitoring are solved through an accurate data processing and intelligent analysis method, and the real-time performance, the accuracy and the predictability of intelligent testing of the integrated circuit are improved.

The invention relates to a virtual power plant load prediction and dynamic response method and system integrated with vehicle network interaction. The method comprises the following steps: acquiring and processing historical charging data to generate a user data set; extracting and analyzing the user data set to generate a charging behavior prediction result; collecting traffic flow data in real time and matching to generate a traffic influence coefficient; constructing a regional charging demand prediction model, and generating regional charging demand distribution; acquiring external environment event data and processing the data to generate a sudden load correction value; a comprehensive load prediction value is generated by fusing multi-level factors, the power distribution and the charging priority of the charging pile are adjusted, and a scheduling instruction is output to complete closed-loop operation, so that the problem of load prediction deviation caused by user behavior randomness, traffic dynamic coupling and unpredictability of emergencies is solved; accurate modeling and collaborative optimization of multi-level dynamic factors are realized, and the scheduling efficiency of the virtual power plant to the vehicle network interaction resources is improved.

The invention discloses a dynamic water quality monitoring system based on an unmanned ship and a data acquisition and distribution algorithm. Comprising an unmanned ship platform module, an initial navigation path setting module, a water quality data acquisition module, a dynamic adjustment and optimization module, a control interaction module and a dynamic adjustment and optimization module, wherein the dynamic adjustment and optimization module dynamically adjusts the navigation strategy and sampling frequency of the unmanned ship in a target to-be-detected water area through a second branch of a first neural network model; the unmanned ship can sail autonomously, continuous and global water quality monitoring is achieved, through an efficient data processing algorithm, the accuracy and reliability of data are improved, a scientific basis is provided for water quality management, a distributed algorithm model is achieved to conduct predictive analysis on each specific target sampling point, the accurate predictability of unmanned ship monitoring is improved, and the water quality monitoring efficiency is improved. The monitoring strategy is dynamically adjusted according to actual requirements, optimal configuration of monitoring resources is achieved, the monitoring cost is reduced, the water quality change trend is found in advance through prediction analysis of the neural network model, and powerful support is provided for early warning and emergency response.

The invention discloses a crop growth real-time monitoring system and method, and relates to the technical field of crop monitoring. According to the crop growth real-time monitoring method, crop image data time sequence and soil state time sequence data are continuously acquired in real time, data analysis is performed respectively to obtain an initial agricultural healthy flow index and a soil conditioning index of each monitoring time period, comprehensive analysis is performed to obtain a comprehensive agricultural healthy flow index of each monitoring time period, and prediction analysis is performed to obtain a real-time crop growth monitoring result. According to the method, the comprehensive agricultural healthy flow index of the next monitoring time period is compared and analyzed with the preset comprehensive agricultural healthy flow index interval, and corresponding early warning measures are taken based on the comparison and analysis result, so that the growth state of crops and the change of the soil environment are tracked in real time; therefore, the health state of the crops and the influence of the soil on the growth of the crops are obtained, and the monitoring accuracy and predictability are improved.

A state-space model which includes one or more neural networks. The state-space model is configured to stochastically model a technical system by modelling uncertainties both in latent states of the technical system and in weights of the one or more neural networks. Thereby, the state-space model may be able to capture both aleatoric uncertainty (inherent unpredictability in observations) and epistemic uncertainty (uncertainty in the model's parameters or weights. During the training and during subsequent use for model-predictive control, moment matching across neural network layers is used, which may ensure that the model's predictions are consistent and close to real system behavior.

The invention discloses an intelligent mold part machining cutting tool control method and system. The method comprises the following steps that S1, a multi-source sensing information collection layer is established; s2, constructing a cutter health knowledge graph; s3, generating a dynamic decision control instruction; s4, executing a bimodal control response; and S5, realizing closed-loop control optimization. Through quadruple mechanism coupling of global coverage of a multi-mode sensing layer, dynamic deduction of a knowledge graph decision-making layer, risk isolation of a double-track execution layer and intelligent evolution of a closed-loop optimization layer, the method is realized in an industrial control domain for the first time: raising a sensing dimension, and converting a physical world fragmentation signal into a cutter full life cycle digital twinborn body; reconstructing decision logic, replacing traditional threshold judgment with topological correlation, and foreseeably inhibiting ill-conditioned failure; the system is ecological and self-consistent, and the control strategy continuously evolves in operation to form the anti-interference capability; and finally, normal form transition of tool wear control from passive remediation to active immunity is achieved.

The invention discloses a forest fire risk assessment method combining microwave remote sensing and fire spreading simulation, and the method comprises the following steps: integrating microwave remote sensing data, and selecting a driving factor which affects the occurrence probability of a forest fire; establishing a time series data set based on the selected driving factor, and constructing a forest fire occurrence probability prediction model based on deep learning; simulating a fire spreading process through a cellular automaton-Monte Carlo simulation method, and calculating a potential combustion probability; and comprehensively evaluating the forest fire risk by combining the fire occurrence probability and the potential combustion probability. The forest fire risk comprehensive assessment method provided by the invention has the advantages of high accuracy, strong predictability and the like, the model can capture trends and changes of factors such as meteorology and vegetation within a period of time before a fire occurs, important features in driving factors can be better concerned, and the risk of the forest fire can be comprehensively assessed. The method for comprehensively evaluating the forest fire risk is more accurate in identification of high-risk fire areas, and more scientific and accurate information is provided for forest fire management.

The invention discloses a task scheduling method based on predictable resource state atlas modeling, a platform is oriented to a heterogeneous computing environment, a unified resource state atlas is constructed by collecting multi-dimensional resource state parameters of computing nodes, and performance characteristics and communication topological relations among the nodes are comprehensively described. On the basis, a bidirectional time sequence model and an attention mechanism are fused, and the load trend of each node in a future short time is predicted. The platform constructs a multi-factor scheduling scoring function based on a task feature vector and resource state prediction map, integrates parameters such as resource matching degree, prediction load, communication delay and energy consumption cost, dynamically evaluates the adaptability of tasks and resources, and realizes adaptive scheduling and optimal resource allocation of the tasks. Compared with the prior art, the method has the advantages of being high in resource state predictability, high in task allocation intelligence degree, outstanding in platform evolution capability and the like, and is suitable for intelligent task scheduling application in a large-scale heterogeneous resource environment.