38 results about "Time model" patented technology

The invention provides a traffic flow prediction method combining a dynamic graph neural network and a Mama mechanism. The future short-term traffic flow is predicted by using historical traffic data. According to the method, firstly, normalization preprocessing is carried out on traffic state data collected by multiple sensors, a training sample is generated by adopting a sliding window, and a traffic flow value in the next one hour is predicted according to data in the past 24 hours. On the basis of the model structure, a time modeling module composed of multiple layers of MambaBlocks is constructed and used for capturing historical time sequence dependence; constructing a spatial modeling module of dynamic graph convolution, and combining a static adjacency matrix of the road network with a learnable adaptive adjacency structure to extract spatial association; and finally, the outputs of the modules are fused, and a prediction result is obtained through a prediction output module. In the training process, a Huber loss function is used as an optimization target, and evaluation indexes such as a mean absolute error (MAE), a root mean square error (RMSE) and a mean absolute percentage error (MAPE) are used for evaluating the performance of the model. According to the method, the traffic space-time dynamic characteristics are effectively mined, and the long-range dependence modeling capability and the prediction precision are improved.

The invention relates to the technical field of water, wind and light storage combination, in particular to a wind, light and water storage combination robust tracking method and device and a medium, which can uniformly represent the output uncertainty of extreme renewable energy sources and quickly map a safe feasible region in a high-dimensional scheduling space. And safe and economic cooperative control of the hydroelectric generating set, the reservoir water level and the energy storage resources within the resolution ratio of 15 minutes is realized. Firstly, unified normalization and sliding window time embedding are performed on wind speed, irradiation and load sequences in a data layer, and a dynamic error envelope is constructed through exponential weighting of a high-tail sample, so that a measurable extreme interval is directly obtained in an original observation space. And then introducing a physical climbing limit and a unit inertial constraint, sequentially expanding a disturbance scene tree on nodes at two ends of the error envelope, carrying out error envelope-physical mapping coupling modeling, enabling the system to keep time-space correlation, and directly incorporating unit dynamic characteristics in a scene generation stage.

The invention discloses a multi-scale time sequence prediction method based on an adaptive sparse expert selection strategy and a closed continuous time neural network. The method comprises the following steps: carrying out normalization and low-dimensional feature mapping based on RevIN; performing trend-seasonal structure enhancement processing on the feature sequence after linear mapping; constructing a multi-scale expert model based on the feature sequence after trend-season enhancement; self-adaptive sparse expert selection and load balancing loss calculation are carried out; carrying out weighted aggregation and residual fusion on multi-scale expert output; global modeling of a closed continuous time neural network based on channel weighting is carried out; and finally performing prediction generation and reverse normalization. The multi-scale time series prediction method has the multi-time-scale adaptive modeling capability, the sparse expert efficient selection mechanism and the global continuous time modeling capability, and can be applied to various multivariable time series prediction scenes such as power load prediction, weather prediction, industrial production monitoring, traffic flow prediction and financial price prediction.

This invention relates to the field of oil and gas field exploration technology, and discloses a method and equipment for predicting the time required for the first discovery of a low-exploration-level depression. The method includes: acquiring geological structural characteristic parameters of the target depression; determining the exploration efficiency of the target depression based on the structural characteristic parameters and pre-determined exploration effectiveness characterization parameters; and predicting the time required for the first discovery of the target depression based on the exploration efficiency of the target depression and a pre-generated first discovery time model. The first discovery time model is generated from the exploration efficiency and the time required for the first discovery of other depressions within the work area where the target depression is located. This invention provides a method for characterizing the time required for the first discovery of a depression based on exploration efficiency constructed from depression structural parameters, which can lay the foundation for medium- and long-term proven reserve prediction in low-exploration-level depressions.

The application discloses a new energy automobile energy supply and conversion facility site selection decision method and device, wherein the method comprises the following steps: acquiring traffic flow data and energy supply and conversion facility data in a planning area; dividing the planning area according to the traffic flow data and the energy supply and conversion facility data to obtain a plurality of traffic zones; establishing a nonlinear high-dimensional space-time model of the traffic flow data and the traffic zones, and obtaining a feasible decision set by using the nonlinear high-dimensional space-time model; performing dimension reduction processing on the nonlinear high-dimensional space-time model, and obtaining suitable strategy parameters in combination with a case data knowledge base; establishing a prediction model to evaluate the traffic flow state and the energy supply and conversion facility service level in the future planning area, and screening in the feasible decision set according to the evaluation result to obtain a plurality of optimal decisions; performing rank-sum ratio sorting on the plurality of optimal decisions to obtain a final decision result; and thus the operation efficiency of the regional traffic system and the energy supply and conversion infrastructure service level can be improved.

This invention relates to a lightweight adaptive method for testing non-stationary data streams. Its core lies in constructing a two-branch adaptive architecture that decouples observation and execution statistics. A shadow observation branch, independent of the main inference branch, maintains a statistical reference baseline in real time, unaffected by adaptive calibration actions, thus suppressing the closed-loop coupling problem between drift perception and statistical updates at the mechanism level. Based on this decoupled architecture, the shadow branch calculates a composite drift surrogate metric to map and generate dynamic momentum parameters, and introduces relaxed cumulative variables to perform three-level state determination and spatiotemporal misalignment mask generation. Finally, during inference forward propagation, dynamic online calibration of statistics is performed only for the key levels selected by the mask. This invention alleviates the adaptation lag and statistical oscillation risks caused by fixed momentum, suppresses the statistical degradation risk caused by single-sample noise, and is suitable for real-time model optimization in single-sample, non-stationary streaming scenarios.

The application provides a typhoon path prediction method and system fusing physical constraints and path mutation identification. The method comprises the following steps: obtaining a multidimensional input feature tensor through a feature tensor generation module; performing time modeling through a multiscale modeling module to obtain a deep feature tensor, performing significance guidance through a significance guidance module to obtain a prediction path sequence; identifying a bending angle through a bending event identification module to obtain a plurality of path anomaly scores, introducing a plurality of preset disturbances through a counter disturbance analysis module to determine path confidence, and outputting a path prediction result through a path prediction output module. According to the technical scheme of the embodiment of the application, the typhoon path can be finely modeled, the PINNs physical constraint module is used to ensure that the prediction result meets the physical conservation principle, the bending event identification is used to enhance the path mutation detection capability, the disturbance analysis is used to realize the uncertainty evaluation, and the typhoon path prediction result is more accurate, reliable and physically consistent.

The invention discloses a low-cost AI auxiliary forklift operation system which comprises a data collecting and preprocessing module used for collecting and processing video data in a forklift operation environment; the target detection module is used for identifying and positioning goods, obstacles and personnel by using an RT-DETR target detection model, and extracting a target category, position information and a motion trend; the space-time modeling module is used for carrying out space-time information modeling on the target features by adopting an improved TimeSform network and extracting a motion trail and an environment relation of the target; the operation scene understanding module is used for analyzing the operation state of the forklift, calculating the relative speed and distance change of a target, detecting an abnormal operation mode and predicting the operation risk; and the path planning and decision-making module is used for adjusting the driving route and the operation strategy of the forklift and generating navigation and safety early warning information. According to the method, AI visual perception and space-time modeling are utilized, the understanding ability of the forklift operation environment is improved, and intelligent auxiliary operation is achieved.

This invention discloses an intelligent correction method for the physical consistency of sub-seasonal model forecast products, belonging to the interdisciplinary field of artificial intelligence and meteorological forecasting. Multiple meteorological elements and multi-level circulation elements from the model forecast are used as raw data. First, the deviation field between atmospheric reanalysis data and model forecast data is calculated. Then, a multi-scale feature extraction network is used to obtain deviation features at different spatial scales. These multi-scale deviation features are concatenated with the original input features and input into a U-Net. The U-Net encoder and decoder are used for deep feature fusion and mapping. Simultaneously, a composite loss function that integrates numerical accuracy and physical consistency is designed to guide the collaborative training of the multi-scale feature extraction network and the U-Net network, resulting in a trained intelligent correction model. In application, the intelligent correction model outputs a physically consistent deviation correction amount to intelligently correct real-time model forecast data, ultimately obtaining a sub-seasonal forecast field that combines prediction progress and physical consistency.

The application discloses a kind of capacity limited vehicle path problem solving method based on timing coding and attention mechanism, first construct the initial feasible solution satisfying constraint condition, and current solution is expressed as containing node level, journey level and solution level hierarchical feature structure;By timing modeling to node sequence in journey and aggregation, journey level feature is obtained, and then the feature set of multiple journeys is modeled by attention, to generate solution embedding vector representing the overall structure of solution;The solution embedding vector is fused with environmental state features and input into the policy decision network, and the selection result of local optimization operator is output, and the solution is iteratively updated according to the selection result until the termination condition is met.The application can enhance the modeling capability of node order change and inter-journey correlation, and improve the stability and generalization performance of the algorithm under different scales.

The application discloses a high-water-pressure shield double-liquid slurry regulation method based on phase state transition, which comprises the following steps: S1, acquiring double-liquid slurry discrete proportioning measured data; S2, constructing a viscosity-intensity evolution prediction model based on the measured data, the model being used for predicting the evolution law of viscosity time variability and intensity of any proportioning combination within a limited interval; S3, acquiring a phase state curve and a defined physical interval based on the evolution law of viscosity time variability and intensity; S4, formulating a water pressure-space coupling phase state strategy based on the phase state curve and the defined physical interval, wherein the coupling phase state strategy is used for determining a target phase state; S5, constructing a comprehensive target function, and acquiring the optimal proportioning of the double-liquid slurry based on the comprehensive target function; and S6, constructing a two-stage time model based on slurry pipeline residence time and segment back filling diffusion time based on the target phase state and the optimal proportioning, wherein the two-stage time model is used for reverse regulation of the double-liquid slurry discrete proportioning.

The present application relates to the technical field of risk analysis, in particular to a supply chain risk quantitative evaluation method and system based on a dynamic matter graph, comprising collecting multi-source heterogeneous data, constructing a four-dimensional space-time model containing a time dimension, a geographical space dimension, a supply chain network space dimension and a risk influence space dimension, and representing a supply chain event as four-dimensional space-time data; identifying a supply chain entity from the four-dimensional space-time data, extracting a risk event and constructing a matter graph, the matter graph taking a risk event as a node and an evolution relationship between events as an edge, and calculating a relationship weight between events; calculating a probability quantitative value of a risk transmission path based on the matter graph, and obtaining a comprehensive risk score of a target entity; generating a risk mitigation strategy based on the comprehensive risk score; monitoring a deviation between an actual risk occurrence and a prediction result, and updating the matter graph and the risk mitigation strategy through an adaptive parameter optimization and an incremental learning mechanism, thereby forming a self-evolving risk evaluation system.

This application discloses a method, apparatus, device, and readable storage medium for predicting the load of a communication network. The method includes: determining whether a load prediction model for the target communication network is suitable for the current operating scenario of the target communication network based on the historical actual load and the historical predicted load of the historical actual load; if suitable, predicting the target predicted load of the target communication network at the next moment based on the recent load data of the target communication network and the load prediction model; if not suitable, updating the load prediction model based on the recent historical load data of the target communication network, and predicting the target predicted load based on the updated load prediction model and the recent load data, wherein the recent historical load data refers to the historical load data of the target communication network within a second preset time period prior to the current moment. Real-time model updates enable the prediction model to better adapt to the real-time operating scenario, thereby obtaining more accurate load prediction results.

The application discloses an airport operation situation dynamic prediction method and device, equipment, a storage medium and a program product, and relates to the technical field of airport management. The airport operation situation dynamic prediction method comprises the following steps: acquiring multi-source heterogeneous data of airport operation; performing multi-modal fusion based on the multi-source heterogeneous data to obtain multi-modal space-time data; inputting the multi-modal space-time data into a pre-trained airport operation situation space-time model to perform situation dynamic prediction, and obtaining a situation prediction result. Since the multi-modal fusion is performed based on the multi-source heterogeneous data, the complementarity and synergy between the data can be mined, so that all-round dynamic prediction of the airport operation situation can be realized. The situation dynamic prediction is performed through the pre-trained airport operation situation space-time model, the instantaneity of the processing of the abnormal time is improved, and the application scenarios are increased.

A priori georeferenced vegetative index data is obtained for a worksite, along with field data that is collected by a sensor on a work machine that is performing an operation at the worksite. A predictive model is generated, while the machine is performing the operation, based on the georeferenced vegetative index data and the field data. A model quality metric is generated for the predictive model and is used to determine whether the predictive model is a qualified predicative model. If so, a control system controls a subsystem of the work machine, using the qualified predictive model, and a position of the work machine, to perform the operation.

The invention discloses a dynamic traffic prediction method, device and equipment, a medium and a product, and the method comprises the steps: obtaining traffic matrixes of a plurality of time steps among all nodes in a communication network, and forming a traffic matrix sequence; according to the traffic matrix sequence, determining an initial spatio-temporal representation feature and a dynamic mode sensing adjacency graph; and obtaining a flow prediction result according to the dynamic mode perception adjacency graph, the initial space-time representation characteristics and the stacked space-time modeling module. The method comprises the following steps of: constructing a dynamic mode sensing adjacency graph based on a traffic matrix sequence under a plurality of real-time time steps, describing local spatial correlation between adjacent traffic, mining global spatial correlation between non-adjacent traffic, and carrying out space-time modeling on the dynamic mode sensing adjacency graph and the traffic matrix sequence based on a stacked space-time modeling module to obtain a space-time model; and the contribution of different flows to the prediction result is adaptively adjusted according to the real-time change of the flows, so that the prediction stability and generalization ability in a non-stable or burst flow scene are improved.

The invention provides a scene demand peak-valley prediction method and system based on a space-time large model, and relates to the technical field of intelligent traffic, and the method comprises the steps: obtaining multi-source heterogeneous space-time data of a target region, carrying out the non-uniform space grid division, calculating an instantaneous supply-demand ratio and a settlement price rolling fluctuation ratio, and constructing a settlement risk enhanced space-time feature tensor; inputting the settlement risk enhanced spatio-temporal feature tensor into a peak-valley prediction spatio-temporal large model, extracting features through a spatio-temporal diagram attention network fusing settlement price fluctuation features, and respectively outputting a transportation order demand prediction value and an uncertainty quantitative index by a double-branch decoder; inputting the transportation order demand predicted value and the uncertainty quantitative index into a strategy to generate a cellular automaton, simulating supply and demand propagation by dynamically adjusting state transfer resistance between cells, and generating a non-uniform distributed settlement parameter adjustment instruction; and directly outputting a settlement parameter configuration result of each sub-region in the target region based on the distributed settlement parameter adjustment instruction.

This invention proposes a method, system, electronic device, and storage medium for offline-real-time fusion feature prediction of big data based on parameter transfer. It belongs to the field of big data processing and artificial intelligence technology. The method includes: an offline model generating offline analysis results using historical data and extracting transfer parameters; transferring the transfer parameters to a real-time model; the real-time model using the transfer parameters to expand features of real-time data and make predictions; after the offline model completes its analysis of data from the same period, fusing the offline and real-time prediction results using a confidence model fusion method; and finally, the offline model updating the transfer parameters for use by the real-time model in the next period. This invention proposes a parameter transfer mechanism and "three principles," constructing an offline-real-time collaborative closed loop. Without compromising real-time performance, it utilizes the deep features of the offline model to improve the accuracy of real-time predictions, solving the problem of balancing timeliness and accuracy in big data analysis.

The present application relates to the technical field of photovoltaic power prediction, and provides a distributed photovoltaic power prediction method and system fusing dynamic regions and space-time. The method comprises the following steps: according to historical power data and corresponding key meteorological characteristics representing power stations, a time model is used to extract time series trend characteristics, and preliminary time prediction values and time prediction errors of sub-regions of the power stations are obtained; the power stations are taken as nodes to construct a graph structure, and a space model is used to extract space correlation characteristics therein, and preliminary space prediction values and space prediction errors of the sub-regions of the power stations are obtained; the contribution degrees of the time model and the space model are dynamically adjusted by using an information entropy method, and combined prediction powers of the sub-regions of the power stations are obtained; data quality evaluation scores of the sub-regions of the power stations are determined, a scale-up weight is obtained through a proportion of the data quality evaluation scores, and the combined prediction powers of the sub-regions of the power stations are weighted and summed, and a total power of a region is predicted, so that the super-short-term prediction accuracy and robustness are effectively improved.

The application discloses a kind of multi-working condition industrial control method and system based on integrated space-time model prediction, method includes: the typical working condition feature extraction method based on orthogonal test, constructs working condition identifier;The time dynamic model and spatial distribution model under each working condition are constructed, and the optimal parameters of the space-time model corresponding to each working condition are obtained using data-driven and integrated training method;The current working condition of multi-working condition industrial system is identified in real time using working condition identifier, and the space-time model under the current working condition is used as prediction model, and the current optimal control input of system is obtained by rolling optimization method.The application learns the space-time correlation of observable point and unobservable point, establishes the prediction model of unobservable point, and integrates it into the prediction control framework, to realize the accurate control of unobservable point.

The invention relates to the technical field of low-altitude navigation and flight safety, in particular to a low-altitude navigation position credible early warning method and system. Resolving to obtain a spatial reference position and a credible level of a corresponding position, and acquiring external standard time by integrating a satellite clock and a time service center; carrying out real-time modeling on risk or anomaly labeling to generate a dynamic quality field; generating a multi-dimensional credible factor; inputting the multi-dimensional credibility factors into a fusion decision engine based on a weight or a machine learning model, outputting a comprehensive credibility score, and triggering a graded early warning response according to the comprehensive credibility score and the over-limit condition of each independent credibility factor; and generating a structured early warning message according to the judged early warning response. A multi-dimensional cross validation and flight environment situation awareness model is constructed, comprehensive credibility evaluation and graded early warning are achieved, upgrading from passive receiving correction to active verification credibility is achieved, and therefore the safety and robustness of low-altitude flight are remarkably improved.

The invention discloses a combustible gas leakage source positioning method and system for an underground pipe gallery. The method comprises the following steps: acquiring actual measurement data corresponding to a plurality of sensors deployed along a gas pipeline; obtaining expressions of predicted values corresponding to the sensors through a time model and a concentration model; obtaining an expression of a time residual error and an expression of a concentration residual error of each sensor based on the expression of the predicted value corresponding to each sensor and the corresponding actually measured data, and respectively carrying out weighting processing on the expressions of the residual errors by adopting corresponding dynamic weighting coefficients; and taking the sum of the weighted time residual sum of squares and the weighted concentration residual sum of squares of all the sensors as an optimization objective function, and obtaining the final leakage point position and the alarm number of the sensor closest to the leakage point position by minimizing the optimization objective function. By means of the scheme, the globally optimal solution most matched with all the measured data can be found, and therefore more stable and more credible leakage source positioning is achieved.

The application discloses a digital twin water conservancy super-fusion algorithm model and a hardware system performance optimization method, and belongs to the technical field of a digital twin water conservancy system. Real-time interaction data of a model is extracted, and a docking configuration system is built. The docking configuration system has an adaptive interface for external service and an interaction mechanism for internal processing. The adaptive interface is adapted to a heterogeneous platform, realizes dynamic acquisition and analysis of external data of the model to obtain input data, and the interaction mechanism is embedded with a hydrology-hydrodynamic river network multi-scale hierarchical model to obtain real-time model parameter values and broadcast focus point information according to the input data. Forecasting and early warning parameter calculation is performed based on the real-time interaction data of the model to obtain current output data, and the current output data is converted into display data for rolling display. The docking configuration system with the adaptive interface is built to adapt to data formats of the heterogeneous platform and break data islands, and a parameter self-adaptive adjustment mechanism combining particle swarm optimization and reinforcement learning is adopted to solve the problem of insufficient adaptability of static parameters.

The invention provides a parameter transfer-based big data offline-real-time fusion feature prediction method and system, electronic equipment and a storage medium. The method belongs to the technical field of big data processing and artificial intelligence. The method comprises the following steps: an offline model generates an offline analysis result by using historical data and extracts transmission parameters; transmitting the transmission parameters to the real-time model; the real-time model performs feature extension and prediction on the real-time data by using the passing parameters; after the offline model completes the analysis of the data of the same period, the offline and real-time prediction results are fused through a confidence model fusion method; and finally, the off-line model updates the transmission parameters for the real-time model in the next period to use. According to the method, a parameter transmission mechanism and'three principles' are provided, an offline-real-time collaborative closed loop is constructed, on the premise that the real-time performance is not reduced, the accuracy of real-time prediction is improved by utilizing the depth characteristics of an offline model, and the problem that timeliness and accuracy are difficult to consider in big data analysis is solved.

A device and method for analyzing a technical system or process that includes a first function and a second function. A first model models a functional relationship of input and output variables of the first function and an influence of a first disturbance variable. The second model models a functional relationship of input and output variables of the second function and an influence of a second disturbance variable. An observation is provided that includes input variables and output variables of the first function and of the second function. A prediction regarding an effect of changing the first function under the influence of the first disturbance variable and / or changing the second function under the influence of the second disturbance variable is determined, and of these functions, that function whose change has a higher likelihood of achieving a desired effect is selected.

A priori georeferenced vegetative index data is obtained for a worksite, along with field data that is collected by a sensor on a work machine that is performing an operation at the worksite. A predictive model is generated, while the machine is performing the operation, based on the georeferenced vegetative index data and the field data. A model quality metric is generated for the predictive model and is used to determine whether the predictive model is a qualified predicative model. If so, a control system controls a subsystem of the work machine, using the qualified predictive model, and a position of the work machine, to perform the operation.

The invention discloses a remote sensing end-to-end ocean three-dimensional temperature forecasting method based on artificial intelligence. The method comprises the following steps: acquiring multi-source remote sensing data and ocean reanalysis data, preprocessing the acquired multi-source remote sensing data and reanalysis data, constructing a space-time sequence forecasting model, training the space-time sequence forecasting model to be qualified, inputting the preprocessed multi-source remote sensing data into the space-time sequence forecasting model which is trained to be qualified, and forecasting the ocean reanalysis data according to the preprocessed space-time sequence forecasting model. And obtaining a forecast result, and carrying out visual display on the forecast result. According to the method, a remote sensing end-to-end forecasting mode is used, a generative space-time model architecture and a sliding window attention mechanism are provided, a space-time problem is disassembled into a space problem combined with a time dimension, and a global attention mechanism is simplified into a window attention mechanism by performing window segmentation in space, so that the calculation overhead is reduced, and the calculation efficiency is improved. And high-resolution ocean three-dimensional environment forecasting can be realized without inputting a three-dimensional numerical mode background field.

The present application relates to a kind of wet desulphurization system prediction control method considering uncertainty compensation, comprising: establishing wet desulphurization object prediction model, which is constructed by using residual model based on Gaussian process theory to predict the unmodeled dynamics and process noise of the labeled uncertainty in the discrete time model of wet desulphurization system;Based on the time delay stage of prediction model in the prediction time domain, and the stage after passing time delay, the uncertainty propagation of wet desulphurization process is carried out respectively, and the corresponding uncertainty propagation formula is obtained;Set the constraint set of outlet sulfur dioxide concentration and circulating pump frequency, and construct desulfurization control process probability constraint based on the constraint set;With prediction model, uncertainty propagation formula and probability constraint, construct prediction controller, realize prediction control.The present application realizes the prediction control considering uncertainty compensation, and has good uncertainty prediction and evaluation ability.