138 results about "Predictive power" patented technology

A method for multivariate time-series long-term forecasting based on multi-scale temporal feature enhancements, includes a time-series forcasting model TFEformer. The model utilizes a multi-branch structure and a patch-series attention mechanism to extract global and local time-series features at multiple temporal scales, and designs an adaptive feature fusion mechanism to achieve adaptive fusion of multi-scale temporal features. It employs an variate-wise attention mechanism and a redesigned gated feedforward network to perform feature fusion among multivariate variables and within the time-series, respectively. The time-series forcasting model TFEformer proposed by the present invention significantly improves the prediction of long-term trends in time-series and enhances the fitting ability for short-term local fluctuations, comprehensively increasing prediction accuracy across different prediction time lengths in multivariate time-series forcasting tasks.

The invention discloses a unified framework for solving automatic driving track prediction and planning consistency based on a world model. According to the method, through cooperative work of the automatic driving domain controller and the vehicle-mounted sensing system, end-to-end joint optimization of track prediction and planning in a complex traffic scene is realized, time sequence dependence and interaction dynamics among intelligent agents are accurately captured, and the prediction capability and robustness of a model are remarkably improved. The method comprises the following specific steps: firstly, constructing a generative world model, and generating potential future state representation by utilizing a behavior conditional and backtracking expansion technology; secondly, in combination with global modeling and a local convolutional network, multi-scale features are extracted, adaptive fusion is carried out, and a multi-modal prediction trajectory is generated; then, a multi-target planning model is adopted to integrate various driving indexes, and a track with the minimum loss function is generated; finally, path planning parameters are dynamically optimized through real-time environment perception and decision feedback, and the problems of prediction uncertainty and planning consistency of the automatic driving track are effectively solved.

The invention provides a cross-domain equipment fault diagnosis method and system based on large and small model cooperation, and relates to the technical field of equipment fault diagnosis. According to the method, the causal field generalization structure is introduced into the small model, explicit decomposition is carried out on the stable causal law and the field specific difference, and meanwhile, the causal field generalization structure is corrected by using the large model, so that the small model can automatically identify and retain the causal relationship which is universally applicable to each device and each field; therefore, the influence of inter-domain distribution difference is effectively eliminated. Theoretical analysis shows that the generalization error of the model mainly depends on the accuracy of the stable causal item, and the structure can minimize error drift caused by distribution drift. Therefore, the robustness of health state evaluation and fault prediction can be remarkably improved in a cross-domain scene, and the fault diagnosis model can still keep the prediction capability close to the training domain level under the condition of no target domain annotation data.

The invention discloses a safety production standardized comprehensive management system and method, and relates to the technical field of safety production management, the system comprises the following components: a data acquisition module, a data preprocessing module, a model construction and training module, a prediction analysis module and a maintenance management module; according to the method, the time sequence data in the full life cycle of the equipment are continuously collected, the time sequence data comprise key parameters such as operation duration, start-stop times and maintenance records, the improved LSTM neural network is utilized to construct the equipment safety life prediction model, and the model can not only predict the overall life of the equipment, but also can predict the service life of the equipment. The method can accurately predict the residual safe use cycle of the easily-worn part, the accurate prediction capability enables an enterprise to plan a maintenance plan in advance, production interruption and safety accidents caused by sudden equipment faults are avoided, and the accuracy and foresight of equipment safety management are remarkably improved.

The invention provides a power distribution communication network fault management process optimization method, which is based on a fault evolution reverse deduction technology of historical repair knowledge, and realizes efficient fault root cause positioning and repair recommendation by combining a dynamic space-time atlas and a graph database technology. The method comprises the following steps: constructing a dynamic space-time atlas, and storing power distribution communication network topology and fault data; constructing a fault causal relationship model, and describing a fault state transition probability and an evolution causal chain; multi-path hypothesis testing is executed, possible fault sources and evolution paths are generated, and confidence scores are distributed; similar fault modes and repair measures are retrieved from a historical repair knowledge base, and a directional repair strategy is recommended; and space-time backtracking analysis is realized, and the whole fault evolution process is visually displayed. According to the method, the root cause positioning efficiency and accuracy are remarkably improved, the fault processing flow is improved, the fault prediction capability is improved, the preventive maintenance level is enhanced, and continuous accumulation and optimized application of maintenance knowledge are realized.

The invention relates to the technical field of electric power prediction, discloses a photovoltaic power generation prediction system based on a multi-source analysis model, and aims to solve the problems that an existing photovoltaic power generation prediction technology is insufficient in individual and isomerized node prediction capability, and prediction is limited due to dependence on historical data in a newly-accessed and data-missing node scene. The system comprises a multi-dimensional node feature quantification module, a virtual historical data synthesis module, a power prediction module based on an enhanced data set, a closed-loop deviation traceability and correction module and a compensation strategy execution module oriented to a specific scene. Through adoption of the technical scheme, high-confidence prediction can be provided for blank or data missing nodes on the premise of not depending on historical data of the target node, and the precision, the coverage rate and the dynamic adaptability of distributed photovoltaic prediction are remarkably improved.

The invention discloses a super-long pile foundation resistance prediction method and model, and the method comprises the steps: firstly, improving the adaptability of the model to complex input and the precise prediction capability of the model to bearing capacity through the strong nonlinear expression capability of a CatBoost algorithm and the characteristics of native support class variables; then six intelligent optimization algorithms are introduced to automatically search hyper-parameters of the model, manual parameter adjustment errors are avoided, the stability and generalization ability of the model are improved, and finally an optimal fusion model is determined for prediction. By constructing the fusion optimization strategy, the multi-optimization algorithm and the CatBoost are deeply fused, the robustness and applicability of the system in the multi-scene and multi-sample environment are improved, and the problems that in the prior art, the resistance prediction precision of the super-long pile foundation is insufficient, and the model stability is poor are solved.

The invention relates to an intelligent thunderstorm weather forecasting method and system and a medium, belongs to the field of thunderstorm forecasting, and provides the intelligent thunderstorm weather forecasting method for solving the problem that an existing mode is insufficient in forecasting capacity, and the intelligent thunderstorm weather forecasting method comprises the following steps: collecting and processing initial data, and obtaining a multi-source fusion initial field of a unified format; constructing a short-time forecasting model based on a Swin Transform backbone network, and outputting a thunderstorm forecasting field in the future 6 hours by inputting a multi-source fusion initial field and underlying surface information; and a short-term forecasting model based on a FuXi large model is constructed, and by inputting the multi-source fusion initial field and underlying surface information obtained in the step 1, possible thunderstorm occurrence areas and intensity distribution data in more than six hours in the future are output. By introducing a Swin Transform backbone network and combining an autoregression prediction strategy and a multi-task loss function, the capturing precision and real-time performance of the thunderstorm time-space evolution process are effectively improved.

The invention discloses a power grid time frequency change trend prediction method and system based on intelligent electric meter data fusion, and the method comprises the steps: measuring the network transmission delay and time deviation between a master clock and a slave clock of a power grid, obtaining an initial time frequency deviation sequence, carrying out the error calibration, carrying out the data fusion after time domain alignment, and carrying out the prediction of the time frequency change trend. Generating a fusion time sequence; performing recursive estimation on the fusion time sequence, outputting an optimal time frequency deviation sequence, and performing synchronous estimation to obtain a clock phase error, a frequency error and frequency drift as state characteristic quantities; and training a preset machine learning prediction model by taking the optimal time frequency deviation sequence as a supervision signal and taking the state characteristic quantity of the corresponding moment as input, and obtaining a prediction result of the power grid time frequency change trend in the future period based on the trained time frequency change trend prediction model. According to the method, the power grid time synchronization precision and prediction capability are remarkably improved, and reliable technical guarantee is provided for stable operation of a power system.

An agricultural harvesting system includes a control system. The control system identifies a predictive value of a power characteristic based on a relationship between the power characteristic and a characteristic. The control system generates a control signal to control a controllable subsystem of a mobile agricultural harvesting machine based on the predictive value of the power characteristic.

The invention provides a hybrid energy scheduling method and system. The method comprises the following steps: acquiring predicted meteorological information and predicted power generation demand in a preset future time period; adopting a multi-target mixed integer programming algorithm to obtain a planning scheme; starting preliminary power generation by each energy unit according to a preset power generation priority, acquiring actual output data in real time, and generating an updated power generation demand according to a preset condition; acquiring real-time meteorological information; dynamically adjusting the operation state of each energy unit by adopting a model prediction control algorithm, and adjusting the actual output according to a surplus scene / gap scene / normal scene subdivision adjustment strategy; calculating a deviation ratio between the adjusted actual output and the updated generating capacity demand, and judging whether the deviation ratio exceeds a preset threshold by adopting a dynamic trigger mechanism; and if the deviation rate exceeds a preset threshold value, starting graded energy complementation. The scheme has the characteristics of high-precision prediction capability, dynamic planning mechanism, subdivision scene adjustment strategy and flexible expansion.

The invention is suitable for the technical field of electrical equipment detection, and provides a ceramic insulator intelligent evaluation method based on spectral feature analysis, and the method comprises the steps: obtaining spectral data through 120 GHz high-frequency laser, synchronously collecting environment and electrical parameters, extracting 10-dimensional spectral features, and fusing the 10-dimensional spectral features into a multi-modal feature vector; training is carried out by using an integrated learning model fused by a random forest and XGBoost and an LSTM time sequence model, and a dynamic weight updating mechanism is combined to adapt to environmental changes; outputting aging grade probability distribution, aging trend prediction in the future 6 months and a risk value gt; and 3.0, early warning is triggered. According to the scheme, evaluation precision, dynamic adaptability and prediction capability are improved, and reliable support is provided for operation and maintenance of a power system.

The invention discloses a predictive maintenance method for an implicit state of power equipment, and belongs to the technical field of intelligent operation and maintenance of the power equipment. According to the method, multi-mode information such as SCADA data, acoustic vibration data, infrared thermal image data and partial discharge data is collected, space-time alignment and attention mechanism fusion are carried out, and a hidden state code representing the internal health state of equipment is extracted; a dynamic state deduction model combining physical constraint and data driving is constructed, and prediction of the future state evolution trajectory of the equipment is achieved; and performing a virtual maintenance experiment in the digital twin based on a prediction result, and generating an optimal maintenance strategy through multi-objective optimization. According to the method, the problems that the hidden state of the equipment cannot be sensed and the prediction capability is lacked in the prior art are solved, the conversion from passive maintenance to predictive maintenance is realized, and the accuracy and foresight of operation and maintenance of the power equipment are improved.

The invention discloses a data middle table design method and system suitable for a power system, and belongs to the field of power system data management. Comprising the following steps: constructing a quantitative mapping model through multi-source data acquisition, and outputting a standardized data value index; establishing a dynamic asset catalog and realizing accurate mapping from values to resources by adopting a multi-objective optimization algorithm; constructing an event-resource-carbon emission combined prediction model to generate an optimal regulation and control strategy; event regression is identified based on a value index, and stable recovery of the system is realized through a value recovery curve; a standardized value voucher is generated by using a block chain technology, and value conversion is completed by means of a cross-chain protocol. According to the method, the problems of data value evaluation deficiency, low resource scheduling efficiency, insufficient prediction capability and the like in the prior art are solved, accurate quantification of data values, intelligent resource scheduling, operation situation prediction, stable system recovery and value closed-loop conversion are realized, and the intelligent level of data management of the power system is remarkably improved.

The invention discloses a subtropical public welfare forest adaptive operation and carbon sink gain system and method based on dynamic monitoring. The system comprises a data standardization module, a space-time interpolation module, a parameter fusion module, a trend analysis module, a contribution quantification module, a factor extraction module, a standard optimization module and a trend prediction module. According to the invention, multi-source data are fused and a quantitative evaluation system is established, so that the monitoring and management efficiency of the subtropical public welfare forest carbon sink gain is comprehensively improved. According to the method, efficient integration of multi-source data and accurate evaluation of carbon sink are realized, and scientificity and predictive ability of ecological management decision are remarkably improved.

The invention discloses a tin-based metal material component-performance prediction model construction and application method based on machine learning, and the method comprises the steps: S1, collecting the component data and performance data of a tin-based metal material, and constructing a basic data set; s2, preprocessing the basic data set, extracting data features from the preprocessed data, and screening key features by using SHAP; s3, constructing a target prediction model based on a machine learning architecture, embedding physical constraints in the model, and training the target prediction model by using the key features; and S4, calculating the Pearson's index of the model, and selecting the current target prediction model based on the size of the Pearson's index. According to the closed-loop research and development system, the experimental verification result is fed back to the model, the data set is further enriched, the model performance is optimized, and the prediction capacity of the model and the material research and development efficiency are continuously improved.

The application discloses an improved short-term and nowcasting precipitation prediction method using random masks and a Transformer, and belongs to the field of precipitation prediction. The improved short-term and nowcasting precipitation prediction method using random masks and a Transformer comprises the following steps: S1, constructing a random mask spatiotemporal sequence image; S2, constructing a network model, and inputting the spatiotemporal sequence image marked with the mask into the network for model training; the network model comprises an encoder-decoder structure with a UNet as a core model, a SwinTransformer module is embedded in the encoder, and an SE-Net attention mechanism is introduced; S3, in the model training process, a prediction value is obtained through a forward propagation process of the input image, then the model is continuously fine-tuned according to a loss function, the loss function is minimized, and the accurate prediction capability of the model is realized; and S4, L1+L2 regularization is used in the training process to prevent overfitting. The high-order non-stationarity in the modeling spatiotemporal sequence is improved, and the short-term and long-term dependence information in the spatiotemporal sequence is learned at the same time, so that the prediction accuracy of the model is improved.

This invention relates to the field of stock index volatility prediction technology, specifically a LightGBM-LSTM hybrid model construction system for stock index volatility prediction. The system includes a data acquisition and processing layer, a multimodal feature processing layer, a dual-branch model training layer, a model fusion and prediction layer, and an application visualization layer. This LightGBM-LSTM hybrid model construction system for stock index volatility prediction employs an offline static screening mechanism to significantly reduce feature dimensionality while retaining effective information. It introduces a dynamic feature weighting and feature drift detection mechanism to dynamically adjust feature weights based on the predictive power of each feature group under different market regimes. A structured feature processor and a temporal feature builder are designed to generate optimal input feature sets for different model branches. A hybrid market regime identification algorithm identifies the current market regime and probability distribution in real time, training LightGBM and LSTM sub-models for the three market regimes and integrating a dynamic fusion strategy to achieve optimal model weight allocation.

This invention relates to a few-sample object detection method guided by dynamic latent features. The method involves acquiring a sample dataset, constructing a few-sample object detection model guided by dynamic latent features, training the model with base class data from the sample dataset, freezing the model's local parameters, and then adjusting the model with new class data from the sample dataset. The method then inputs the target data into the adjusted model to obtain the few-sample object detection results. This invention makes feature representation more refined and diverse, improving the overall detection effect and the model's generalization ability. Through multi-scale information extraction and multi-similarity guidance, it significantly enhances the model's effective utilization of supporting features, thereby strengthening the predictive ability of targets.

The present application relates to the technical field of power grid new energy consumption, demand side management, and specifically discloses a method and system for stimulating user energy consumption considering anchoring-ratchet effect, comprising collecting relevant parameters; constructing an IDR double-layer stochastic optimization model; and inputting the collected relevant parameters into the IDR double-layer stochastic optimization model to obtain an optimized incentive price; and adjusting the incentive price of the unit response quantity released by the current IESP to the user based on the optimized incentive price. The method comprehensively measures the influence of user subjective psychology on user response behavior based on "anchoring effect" and "ratchet effect", and improves the prediction ability of IESP for the to-be-estimated parameters representing user subjective psychology, so as to enable IESP to formulate effective and economic incentive strategies, thereby guiding users to adjust energy demand, reducing peak-valley difference, improving the accuracy and effectiveness of load regulation, and achieving a win-win situation between users and IESP.

The invention relates to the technical field of power transmission line monitoring, in particular to a power transmission line icing detection method, system and device based on deep transfer learning and a storage medium. According to the method, an insulator image data set containing five states of no icing, rime, mixed rime, glaze and snow coverage is constructed, a target detection model architecture comprising a feature extraction network and a prediction network is constructed, the feature extraction network adopts hierarchical feature learning to realize multi-scale feature extraction, and the prediction network adopts a task decoupling architecture to separate, classify, locate and predict; a transfer learning training strategy is implemented, a general target detection data set is used for learning general feature representation in the source domain pre-training stage, feature extraction network parameters are kept in the target domain fine tuning stage, a prediction network is initialized again, an icing data set is used for fine tuning training, and prediction capacity for icing recognition is established; and inputting a to-be-detected image into the trained model, and obtaining insulator position information and an icing type classification result.

The application discloses a lattice rainstorm prediction method, and the method adopts circulation background field, U\V prediction field, rainfall prediction field, model adjustment field and data of a ground meteorological station, and constructs a training sample set; an XGBoost method is used to establish a prediction model based on a mapping relationship of a circulation background field of an EC model and prediction products and future 0-24h precipitation; the training sample set is trained through the prediction model, so that the prediction model can perform medium and short-term precipitation prediction. The prediction ability (rainstorm TS) of the method for precipitation above rainstorm is improved by more than 17% compared with the EC model prediction, the prediction precision of medium and short-term precipitation can be effectively improved, and the method has a good application prospect and provides more accurate prediction services for disaster prevention and reduction.

Disclosed are techniques that generate a structural representation of a plurality of documents, the structural representation including a plurality of nodes and a plurality of edges, with the plurality of nodes being representations of the plurality of documents and the plurality of edges representing a feature in common between nodes of the plurality of nodes, with each node holding a vector of confidence values for weak models on a current optimization step and a weighted prediction for each of the weak models, generate a local ensemble model from the structural representation of the plurality of documents combined with the weighted prediction of the weak models, with the generated local ensemble model having a higher predictive power than any weak model individually, and generate a label for each node based on the local ensemble model.

Optical coherence tomography (OCT) data can be analyzed with neural networks trained on OCT data and known clinical outcomes to make more accurate predictions about the development and progression of retinal diseases, central nervous system disorders, and other conditions. The methods take 2D or 3D OCT data derived from different light source configurations and analyze it with neural networks that are trained on OCT images correlated with known clinical outcomes to identify intensity distributions or patterns indicative of different retina conditions. The methods have greater predictive power than traditional OCT analysis because the invention recognizes that subclinical physical changes affect how light interacts with the tissue matter of the retina, and these intensity changes in the image can be distinguishable by a neural network that has been trained on imaging data of retinas.

Aspects of dynamic control of minimum voltage for semiconductor devices are disclosed. For example, a method for minimum voltage dynamic control includes receiving a predictive temperature and a predictive power consumption from a predictive model. The predictive temperature and the predictive power consumption are determined based on power consumption, performance state residency, and temperature of an element of a semiconductor device operating at an operation point. A temperature-based minimum voltage, determined based on the temperature of the element, and a power-consumption-based minimum voltage, determined based on the power consumption of the element, are received at a logic. The logic generates a minimum voltage determined based on the temperature-based minimum voltage, the predictive temperature, the predictive power consumption, the power-consumption-based minimum voltage, and the performance state residency of the element. The voltage of the element may be adjusted based on the minimum voltage generated by the logic.

The application relates to the technical field of intelligent construction and automatic construction, and is a multi-levelling robot cooperative operation method and system based on laser point cloud data. In order to solve the problems of lack of prediction ability and low cooperative efficiency of traditional centralized scheduling, the application proposes a new cooperative paradigm based on predictive digital twinning and market game. First, a predictive digital twinning system capable of deducing the future material state is constructed by using the real-time fused laser point cloud data of multiple robots. Then, the system decomposes the global task into dynamic contracts and publishes them to the robot market, and each robot as an independent agent wins the contract through bidding game, thereby spontaneously forming the most efficient temporary operation chain. The application replaces rigid instruction control with prediction and game, realizes the fundamental change from passive response to predictive adaptive cooperation, and significantly improves the efficiency, flexibility and robustness of large-area levelling operation.

The invention relates to prediction-based power reservation regulation for data center scale. In various examples, systems and methods related to prediction-based power reservation regulation of a data center scale are disclosed. One or more circuits may receive power consumption data for a first time period from a plurality of components of a data center. The one or more circuits may generate predicted power consumption for the plurality of components for a second time period after the first time period using at least one prediction model and based at least on the power consumption data. The one or more circuits may determine a power policy for the plurality of components based at least on the predicted power consumption and a state of the data center, and cause the plurality of components to limit power consumption for a second period of time according to the power policy.

The invention discloses a lightweight long-time target tracking method based on an information entropy online updating template, and belongs to the field of computer vision. The method comprises a lightweight backbone network for feature extraction and relation modeling, a tracking prediction head for outputting target bounding box coordinate statistical distribution, and a novel tracking quality self-evaluation mechanism based on information entropy. And the prediction capability of the teacher model is distilled to a lightweight student model, probability distribution of a target bounding box output by the tracking prediction head is directly used for tracking quality evaluation, and whether online template updating is carried out is determined based on an evaluation result. The template updating mechanism almost does not introduce extra computing overhead, and the finally obtained lightweight target tracking network model with the template updating mechanism can be conveniently deployed to an edge computing platform with limited computing power, so that long-time stable target tracking is realized.

The invention discloses an ultra-low orbit satellite electric propeller two-stage cascade power overall planning control method, and relates to the technical field of satellite electric propellers, and the method comprises the steps: constructing a thrust-radio frequency power matching model; utilizing an extended Kalman filtering algorithm to carry out radio frequency source power dynamic estimation technology research, and constructing a state observer; based on a thrust-radio frequency power matching model and a state observer, constructing a self-adaptive predictive power control framework based on measured data and dynamic state estimation; determining a power coupling mechanism between the spiral wave radio frequency source and the ion cyclotron resonance heating radio frequency source, and constructing a front-stage and back-stage power linkage relation mathematical model; based on a front-stage and back-stage power linkage mathematical model, a dynamic power optimal distribution method with optimal efficiency, thrust and specific impulse under power constraint is established, and a feedback controller based on delay compensation linear quadratic Gaussian control is constructed. And the performance of the ultra-low orbit satellite electric propeller is optimized.

The invention provides a fire hazard wind field and spread dynamic prediction method and system based on radar echo evolution characteristics. The method comprises the following steps: step 1, carrying out ground clutter suppression, radial high-frequency oscillation filtering, speed deblurring and time-space consistency check on reflectivity Z, radial speed V, spectral width W and dual polarization quantity of a plurality of weather radars; 2, under a three-dimensional variational 3DVAR framework, taking a numerical mode background field as prior, and performing inversion to obtain a boundary layer three-dimensional wind field; 3, establishing an anisotropic cellular automaton (CA) and live wire particle swarm coupling model under the constraints of a digital elevation model (DEM), combustible material types, water content and meteorological elements; 4, performing real-time correction and short-term and temporary extrapolation on the fire point and live wire states; and step 5, result output and early warning. By applying the technical scheme, the minute-level short-term and temporary prediction capability can be improved, and the safety of key assets such as power transmission channels is ensured.