390results about How to "Fully understand" patented technology

The invention belongs to but not limited to the technical field of fault diagnosis and maintenance of power equipment, and discloses an intelligent diagnosis and maintenance system and method for power equipment based on a knowledge graph. A data acquisition and preprocessing module acquires real-time operation data and historical data of the power equipment; the knowledge graph construction module associates various kinds of information of the power equipment to construct a knowledge graph; the intelligent diagnosis module performs association analysis on power equipment faults based on a knowledge graph and a machine learning algorithm, identifies fault modes and potential faults, and predicts possible faults of the equipment; the maintenance suggestion module automatically generates a detailed maintenance scheme according to the diagnosis result; the operation and maintenance visualization module provides visual fault diagnosis results and maintenance suggestions, supports multi-level data display, and helps operation and maintenance personnel to clearly understand the equipment state. The operation safety of the power equipment is improved, the operation and maintenance cost is remarkably reduced, and the overall efficiency of equipment management in an intelligent power grid environment is improved.

The invention discloses a power load prediction method based on multi-source heterogeneous data feature fusion, system equipment and a medium, and relates to the technical field of power system load prediction, and the method comprises the steps: obtaining a power load prediction data set, and carrying out the preprocessing; performing input feature processing on the static data, the graph data and the time series data based on a multi-layer perceptron, a graph convolutional network and a gating loop unit network, and fusing the processed features based on a feature splicing method to obtain comprehensive feature information; based on full-connection neural network construction, taking the comprehensive features as input, inputting training set data into a model, and performing iterative optimization according to a model effect to adjust model parameters and hyper-parameters; and based on the parameter optimization adjustment model, using the evaluated optimal model to predict the power load in the future time period. The method is superior to a traditional method in precision and stability, can better adapt to complex and changeable load prediction scenes, and has wide application prospects and remarkable actual value.

The invention discloses a multi-model fusion parallelization reservoir parameter prediction method, system, equipment and medium, and the prediction method comprises the following steps: standardizing logging curve data and actually measured reservoir parameter data, constructing a BiGRU-TCN-Transform model, determining an optimal parameter combination, and training to obtain a final model; inputting the training set data into the model for training until the model converges, and obtaining a trained prediction model; predicting the test set data, and outputting a final prediction result; secondly, an improved VAE-Attention model is constructed, the prediction result in the first step serves as input, and a final operation result of the model is output through model processing; the invention further comprises a multi-model fusion parallelization reservoir parameter prediction system, equipment and a medium. The precision is improved, and good stability, reliability and applicability are achieved.

The invention relates to the technical field of ozone concentration prediction, in particular to an ozone concentration prediction method, device, equipment and medium based on a space-time diagram convolutional network model.The method comprises the steps that historical ozone concentration and meteorological data of a target area are obtained, and training and prediction samples are constructed after standardization; a space diagram structure is established, nodes represent monitoring points, and edges represent the monitoring points with correlation exceeding a threshold value; extracting a spatial dependency relationship by adopting GCN and GAT, and generating time sequence data; tCN or RNN is adopted to extract time-dependent features, and spatial-temporal feature comprehensive representation is obtained; training a model by utilizing the training sample and the spatial-temporal characteristics, adopting MSE (mean square error) as a loss function, adjusting parameters by an Adam optimizer, and optimizing hyper-parameters to minimize errors; newest meteorological and historical data are input, future ozone concentration is predicted, GNN extracts spatial features, RNN processes a time sequence, time sequence data can be processed in parallel through convolution operation, and high calculation efficiency is achieved.

The invention discloses a mechanical arm operation method based on a multi-mode large visual language model, which comprises the following steps of: firstly, acquiring image data of a mechanical arm operation scene through a visual sensor, and acquiring a user instruction in combination with a voice recognition module or a text input module; and inputting the natural language instruction and the visual scene data into a multi-modal large visual language model for multi-modal task analysis. The model identifies the category, position and attribute of a target object in a scene, and generates a specific operation sequence in combination with an input instruction. And then, according to the operation sequence generated through analysis, the system directly drives the mechanical arm to execute corresponding tasks including specific actions such as grabbing, moving and placing of the target object, and in the task execution process, the mechanical arm sequentially completes the actions according to the operation sequence generated in advance. The problems that in the prior art, a mechanical arm is inaccurate in complex task instruction analysis, poor in operation flexibility and adaptability and poor in real-time response capacity in a dynamic environment are effectively solved.

The invention discloses a junior high-school English reading understanding test question intelligent analysis method based on a large model, which comprises the following steps: S1, obtaining answer texts of junior high-school English reading understanding test questions completed by students and corresponding test question texts, and answers are from published standard answers; s2, preprocessing the input test question text and answer text; s3, performing semantic level deep analysis on the test question text by utilizing a deep context understanding large model; s4, according to the test question text and the standard answer, in combination with understanding of the logical reasoning large model on text context and logical relationship, analyzing the semantic matching degree of the student answer, and based on the standard answer and the logical reasoning large model, generating a semantic and logic comprehensive evaluation result of the student answer; s5, dynamically generating learning suggestions based on the semantic and logic comprehensive evaluation result of the student answers; and S6, outputting a test question analysis report. According to the invention, the answers of the students are analyzed more accurately, and targeted learning guidance is provided, so that the teaching effect and the learning experience of the students are remarkably improved.

The invention relates to the technical field of image classification, in particular to a fetal heart monitoring image classification method and system based on a neural network. The method comprises the following steps: acquiring a fetal heart monitoring image, generating a standard monitoring image through Gaussian noise elimination and spatial domain normalization, then performing multi-channel convolution on the standard image to extract image convolution features, obtaining standard fetal heart features through compression regular optimization, performing morphological reconstruction to generate a fetal heart dynamic model, and finally obtaining the fetal heart dynamic model. Baseline features are identified based on historical data, then periodic variation quantification and safe floating range derivation are performed on fusion features, and finally a multi-scale discrimination framework is constructed to realize intelligent classification of fetal heart monitoring images. The fetal heart monitoring image classification method is more stable and more accurate.

The invention discloses an equipment fault natural language question and answer method and system based on cue word optimization. The method comprises the steps of obtaining an equipment fault problem which is input by a user and described by a natural language; performing natural language understanding and semantic representation on the question, and identifying core words and dimensions; generating an optimal query prompt word based on the core word and the dimension; combining the optimal cue word with a pre-constructed equipment fault knowledge base, and constructing query and retrieving related knowledge; generating a natural language reply based on the retrieval knowledge and presenting the natural language reply to the user; obtaining user feedback information; and based on the feedback information, optimizing the cue word generation strategy and updating the knowledge base to form a closed-loop learning process. The system further comprises innovation modules such as cross-modal fusion and understanding, intelligent labeling and knowledge extension. According to the method, innovative mechanisms such as cue word optimization generation, knowledge base closed-loop optimization, multi-source heterogeneous data fusion utilization and low-cost knowledge extension are provided, and highly-intelligent fault diagnosis and guidance are realized.

The invention provides an intelligent interaction method and device and an agent-based interaction system, relates to the technical field of artificial intelligence such as natural language processing, large language models and deep learning, and can be applied to multi-agent collaboration based on a large language model (LLM). Comprising the steps of determining a knowledge base, a positive sample reference statement and a negative sample reference statement which are currently associated with a dialogue system; performing intention recognition on the current first input statement based on the knowledge base, the positive sample reference statement and the negative sample reference statement, and determining a first intention of the first input statement and the definition of the first intention; outputting an intention inquiry statement under the condition that the definition of the first intention is smaller than a first threshold value; obtaining a second input statement, and returning to execute the step of performing intention recognition on the first input statement based on the second input statement until a target intention with the definition greater than a first threshold value is obtained; and outputting a reply statement corresponding to the first input statement based on the target intention.

The invention discloses an unmanned aerial vehicle cluster intelligent formation method, device and equipment based on distributed reinforcement learning, and the method comprises the steps: constructing a reinforcement learning environment of an unmanned aerial vehicle cluster, which comprises a state space, an action space and a reward function, according to a flight task scene of an unmanned aerial vehicle, the flight mission scene comprises execution of air refueling and long-distance formation transportation; building a training framework of a distributed reinforcement learning algorithm based on the reinforcement learning environment, and performing network construction based on the training framework to obtain a neural network including an action decision network and a formation target evaluation network; performing distributed training on the neural network by adjusting hyper-parameters of the neural network and utilizing large-scale distributed data to obtain a formation model; and loading the formation model to an actual unmanned aerial vehicle cluster system, and adjusting a formation strategy according to a real-time flight environment and task requirements to carry out a formation flight task. The cooperative flight capability and the self-adaptability of the unmanned aerial vehicle cluster can be improved.

The invention discloses an energy industry intelligent algorithm optimization method based on federal learning. The method comprises the following steps: S1, collecting and preprocessing local energy data; s2, acquiring and preprocessing expert example data; s3, constructing a comprehensive data set by applying a multi-modal data fusion technology; s4, each participant performs local model training on the comprehensive data set by using a federal adaptive optimization algorithm to generate local model update; s5, the central server aggregates local model updating to generate a global model; s6, applying a soft actor-commentator algorithm in the global model, and combining an inverse reinforcement learning technology to optimize an energy distribution strategy; s7, creating a virtual simulation environment by using a digital twinning technology, and testing and optimizing the performance of the global model; and S8, providing a user-friendly interface and a natural language processing interface, and realizing personalized energy management suggestions. According to the method, federated learning, a soft actor-commentator algorithm and an inverse reinforcement learning technology are combined, and an energy distribution strategy is optimized.

The invention discloses a platform vehicle abnormal behavior intelligent analysis method based on deep learning. Accurate early warning is realized through multi-dimensional data acquisition and deep learning model construction. Firstly, platform passenger flow, road conditions, video monitoring and vehicle operation information are integrated, and a comprehensive data basis is provided for anomaly analysis. After the data is preprocessed, a graph neural network and an attention mechanism are utilized to perform preliminary fusion, and an AGCN adaptive graph convolutional network and a DTW dynamic time warping algorithm are introduced to optimize a fusion result, so that time sequence data are effectively aligned, and the feature extraction accuracy is improved. And then, a deep learning model is constructed to automatically identify abnormal behavior modes of the platform vehicles, such as illegal parking and passenger congestion. Early warning information is issued in real time, potential risks can be timely handled, and safe and efficient operation of a public transportation system is guaranteed. According to the scheme, the identification and early warning capability of the abnormal behavior of the station vehicle is enhanced, and powerful support is provided for bus safety management.

The invention discloses a nucleic acid binding protein recognition method based on a protein map and a protein language model, and belongs to the technical field of bioinformatics. According to the method, high-dimensional sequence features extracted by a ProtT5 protein language model are combined with three-dimensional structure information predicted by AlphaFold, so that the model can capture rich feature information of the sequence and the structure at the same time, and the accuracy and reliability of NABPs prediction are remarkably improved; the expression capability of sequence features is enhanced through the GAT network and the BiLSTM network, so that the constructed prediction model can more comprehensively understand important information in the protein sequence, and the prediction performance of NABPs is improved; the features output by the GAT network and the BiLSTM network are fused, and NABPs classification is performed through the MLP, so that the prediction model can make full use of feature information from different sources, and the NABPs prediction performance and robustness are improved.

The invention discloses an automatic driving planning method based on a visual language large model and related equipment, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring driving scene information; the driving scene information is input into a visual language large model, automatic driving planning parameters output by the visual language large model are obtained, and the visual language large model is obtained through sample driving scene information and thinking chain fine adjustment; the thinking chain is used for assisting the visual language large model to perform multi-stage reasoning analysis on the sample driving scene information. By implementing the technical scheme provided by the invention, the reliability of model output is improved through multi-stage reasoning, and the safety of the automatic driving system is enhanced.

The invention provides a multi-course fusion knowledge graph construction and real question analysis method and system. The method comprises the following steps: carrying out knowledge point alignment operation, carrying out integration operation on cross-course knowledge points, carrying out quantitative analysis on the knowledge points by utilizing a cosine similarity calculation method, and identifying and aligning synonymous knowledge points in different courses; performing knowledge point disambiguation operation, and processing the situation that the knowledge points have the same name and different meanings to obtain integrated knowledge points and an integrated knowledge graph; performing knowledge point standardization operation on the integrated knowledge points, and performing standardization processing on the integrated knowledge graph to obtain standardized knowledge points and a standardized knowledge graph; converting the normalized knowledge points, extracting to obtain key features, and processing according to the key features to obtain semantic feature sequence vectors; matching to obtain a matching result and applicable knowledge points; and verifying and evaluating the matching result, and mounting the question to the corresponding knowledge point. The technical problems that the knowledge point extraction result is one-sided and the result accuracy is not high are solved.

The invention discloses a blind person auxiliary interaction method and system based on a multi-modal large model and a storage medium, and belongs to the technical field of artificial intelligence, and the method comprises the steps: obtaining a user voice instruction and an external environment image, and carrying out the preprocessing of the obtained data; inputting the preprocessed data into a pre-trained multi-modal large model to obtain an environment understanding result; integrating the environment understanding result into feedback information for voice output, and adjusting output intonations according to environment characteristics; the acquisition method of the pre-trained multi-modal large model comprises the following steps: acquiring historical multi-modal data; and inputting the historical multi-modal data into a pre-constructed multi-modal large model, carrying out fine tuning on the trained multi-modal large model through a base layer in a frozen model to obtain a trained multi-modal large model, fusing multi-modal information through the multi-modal large model, accurately identifying a user intention and a corresponding object in a complex environment, and carrying out real-time identification on the user intention and the corresponding object in the complex environment. Therefore, accurate scene description is fed back.

The invention belongs to the technical field of computer vision and artificial intelligence, and discloses a combined zero sample image classification method based on hierarchical feature fusion. According to the method, a solution based on hierarchical feature fusion is innovatively provided for the technical difficulty of combined zero sample image classification. According to the method, features of different depths in a CLIP model visual encoder are selected and subjected to hierarchical fusion, multi-level and multi-scale visual information is effectively extracted, the fused features are further used for visual and semantic cross-modal interaction, and the understanding ability of the model for complex combinations is remarkably improved. In addition, by introducing a specific loss item, attributes and object features are effectively separated, and the feature aliasing problem is avoided. In a combined zero sample image classification task, the method shows excellent generalization ability and recognition effect, and can significantly improve the classification performance of the model for unknown complex combined images.

The invention provides a multi-platform collaborative data acquisition and analysis method, and relates to the technical field of data processing and analysis. The method comprises the following steps: receiving and converting data from different platforms into a unified intermediate format through a data adaptation intermediate layer, and processing a data structure and an interface of a specific platform by using a plurality of adapters; the data abstraction layer defines a standardized interface, so that multiple platforms request data through a unified interface, and a mapping rule converts the request into calling of the adaptation layer; the data integration engine preprocesses the data to ensure data consistency and accuracy; the data warehouse adopts a distributed architecture to ensure the service continuity through load balancing, data partitioning, a caching mechanism and a redundant storage and fault transfer mechanism; and the data processing and analysis module converts the data into decision suggestions by using a visual tool and a data mining algorithm. According to the method, manual intervention is reduced, the data processing speed and accuracy are improved, the data mode and trend are deeply mined, and service decision is promoted.

The invention relates to the technical field of knowledge base construction, in particular to a knowledge service system construction method, which comprises a knowledge acquisition step for collecting knowledge data from knowledge sources of multiple fields; the preprocessing step is used for preprocessing the knowledge data; the classification step is used for classifying the obtained knowledge data; the similarity calculation step is used for calculating the similarity between different categories of knowledge bases; the first judgment step is used for judging whether the similarity is greater than a first threshold value or not; the association step is used for establishing association between the knowledge according to the similarity; the second judgment step is used for judging whether the similarity is greater than a second threshold value or not; the labeling step is used for constructing a new upper label; a label directory generation step for establishing a label directory; a query demand acquisition step for acquiring a knowledge retrieval demand of a user and generating a query request; the knowledge retrieval step is used for conducting knowledge recommendation according to the query request. The method has the effect of improving the searching efficiency.

The invention provides a semiconductor equipment data acquisition method and a semiconductor equipment data acquisition system supporting standardized communication transformation, which are used for realizing data acquisition and process control on different types of semiconductor equipment, formulating corresponding communication rules for equipment conforming to a communication standard, and converting an equipment communication protocol by a cluster system so as to match a standard interface. For equipment which does not completely meet the standard but can realize communication through programming, the trunking system directly acquires required data and converts the required data to meet the standard interface specification. For equipment without communication capability, data is acquired through an equipment interface by adopting an image recognition technology, data input and function operation are realized through simulating keyboard and mouse operation, and finally, protocol conversion from data to a standard interface is completed by a cluster equipment control system. And the equipment data is converted into a message data packet conforming to a standard semiconductor equipment communication interface, so that docking with a semiconductor equipment data analysis and process control system is realized, and the production efficiency and quality of the semiconductor equipment are improved.

The invention provides a heterogeneous complex data set-oriented multi-scale time entropy evaluation method, which belongs to the technical field of time sequence analysis, and comprises the following steps of: S1, performing down-sampling on an initial time sequence, dividing the initial time sequence into subsequences on different time scales, and then processing the subsequences by applying a time window method, obtaining a plurality of clustering windows; s2, calculating an entropy value in each clustering window, and averaging the entropy values of all the clustering windows to obtain a multi-scale time entropy; and S3, according to the obtained multi-scale time entropy, carrying out comprehensive evaluation on the initial time sequence. According to the multi-scale time entropy evaluation method for the heterogeneous complex data set, the method has extremely high generalization ability, so that the evaluation of the model performance is more accurate and reliable, and the method is suitable for wide practical application scenes.

The invention discloses a target voice extraction method based on multi-reference clue fusion, and the method comprises the steps: enabling a to-be-extracted multi-channel mixed voice to pass through a feature extraction module, calculating the complex spectrum and spatial features of the mixed voice, and obtaining a mixed voice feature map; enabling the azimuth information and the reference voice of the target speaker to pass through an embedded feature extraction module, and combining the direction features and the voiceprint features to obtain embedded features of the target speaker; inputting the mixed speech feature map and the embedded features of the target speaker into a main extraction network based on a dual-path Fastform neural network to obtain high-dimensional feature representation of the speech to be extracted; the high-dimensional feature representation is converted into a plurality of masks for estimating the separated voice through a mask estimation module, and the masks act on a plurality of spectrums of the mixed voice signals; estimating and extracting a time domain waveform of the speech through inverse short-time Fourier transform reconstruction; and the extracted voice of the target speaker is input into the comparison learning module, so that the correct extracted voice can be output when the mixed voice is absent by the target speaker.

The invention provides a multi-mode user intention identification method and system based on eye movement tracking, the method identifies a gazing object based on an eye tracker, and combines modes such as text and image to estimate the user intention to execute a specific task, the eye movement tracking is that pupil positions are positioned through an event camera to obtain coordinates, and the gazing point of eyes is calculated; according to the method, the user intention is identified and estimated, then the gaze information graph is constructed through picture element segmentation, the user intention is identified and estimated, the robustness and accuracy of user intention identification are enhanced, the system can more intelligently adapt to complex requirements of different users in different scenes, and therefore more personalized service experience is provided.

The invention relates to the technical field of video source switching, and discloses an intelligent video source switching system and method for a director switching station. The method comprises the following steps: acquiring an input video stream of a current video source through a director switching station, and performing N-gram feature vector calculation and inter-frame difference analysis to obtain video frame segmentation data; performing scene analysis on the video frame segmentation data to obtain a scene feature signature; performing spatial-temporal feature extraction and fusion on each video frame in the input video stream according to the scene feature signature to obtain a scene spatial-temporal fusion feature set; inputting the scene space-time fusion feature set into an LSTM-based lightweight time sequence convolutional network model to carry out video switching behavior prediction so as to obtain video switching behavior prediction data; and inputting the video switching behavior prediction data into a double-depth Q learning model based on a Markov decision process to carry out video source switching decision to obtain a video source switching decision result, thereby improving the intelligent video source switching accuracy of the director switching station.

The invention discloses a short-time traffic flow prediction method and system based on a multi-feature map, and relates to the technical field of traffic flow prediction, and the method comprises the steps: respectively constructing an occupancy rate feature map, an average speed feature map and a traffic flow feature map through three kinds of data, including traffic flow, occupancy rate and average speed data, collected by a road sensor at the same time; node feature information of three feature maps is aggregated by using a GCN, supervised learning training of occupancy features and average speed features is realized by using an improved Transform encoder, and feature interaction fusion of obtained feature vectors of occupancy and average speed is realized in a multi-feature fusion traffic flow prediction model. The occupancy and average speed features are utilized to assist a multi-feature fusion traffic flow prediction model to better capture traffic flow features, and more accurate traffic flow prediction is provided.

The invention relates to the technical field of numerical simulation, in particular to a grid model generation method for the fluid displacement process of well cementation engineering. The method comprises the following steps: acquiring geological data of a well cementation project area, and constructing an initial geological model; well cementation engineering design data are obtained, well cementation engineering topological structure modeling is carried out, and a well cementation engineering topological structure model is obtained; according to the initial geologic model and the well cementation engineering topological structure model, fluid dynamic simulation is carried out, and a fluid characteristic matrix is obtained; performing fluid displacement process simulation according to the fluid characteristic matrix to obtain fluid displacement process data; performing fluid displacement process optimization on the well cementation engineering design data based on the fluid displacement process data to obtain optimized fluid displacement process data, and performing resource allocation redistribution on the well cementation engineering design data according to the optimized fluid displacement process data to obtain optimized well cementation engineering design data. According to the invention, the accuracy of fluid displacement simulation can be improved.

The invention relates to the technical field of breach image processing, in particular to a breach rapid situation awareness method and system based on multi-source data. The method comprises the following steps: acquiring dam monitoring station data; performing hydrological sensor network setting on the dam monitoring station data to obtain dam hydrological sensor network data; performing ultrasonic water level change trend analysis on the dam hydrological sensor network data to generate an ultrasonic water level change trend chart; performing breach node sensing on the dam hydrological sensor network data according to the ultrasonic water level change trend chart to generate breach node data and breach prediction node data; and according to the breach node data, macroscopic disaster risk analysis is carried out on the breach area, and a breach area macroscopic disaster risk schematic diagram is generated. According to the invention, through real-time data integration, comprehensive risk assessment, unmanned aerial vehicle monitoring, dynamic prediction model and multi-party cooperative response, the accuracy and response speed of breach rapid situation awareness are improved.

The invention discloses a speech emotion recognition method based on multi-modal feature fusion, and the method comprises the steps: designing a multi-modal speech emotion recognition model, enabling an audio to serve as a main mode, employing the function of wav2vec2.0, and combining a text mode with a visual mode to enhance the performance of emotion recognition based on the audio; the self-attention mechanism, the GRU and the LVC are combined to extract global and local information from a visual mode, the global information is extracted only from text features, visual clues and text clues related to emotion can be comprehensively understood, the extracted text and visual features are integrated into a wav2vec2.0 model, multi-mode representation of multi-level audio information is formed, and the multi-mode representation of the multi-level audio information is realized. And meanwhile, the discrimination capability of the model is enhanced by utilizing multi-modal homogeneous feature difference learning, and the proposed model is evaluated on two data sets, so that the model is proved to have excellent robustness and generalization compared with the existing most advanced model.

The invention discloses a wind speed prediction method based on improved ensemble empirical mode decomposition and LSTM model series connection, and the method comprises the steps: adding grid point historical meteorological data into adaptive noise through singular spectrum analysis, carrying out the ensemble empirical mode decomposition, and carrying out the processing of an intrinsic mode function component and a residual error obtained through the decomposition; performing standardization processing on the topographic data, and extracting topographic spatial features through different spatial feature extraction schemes; and constructing the fused features into time sequence format data, inputting the time sequence format data into an LSTM model for prediction, obtaining a predicted intrinsic mode function component and a residual error, adding the intrinsic mode function component and the residual error to generate a short-time wind speed prediction result, comparing the short-time wind speed prediction result with the real-time wind speed data to calculate a model loss function, and training the LSTM model by adopting back propagation iteration to obtain a short-time wind speed prediction result. And performing rolling prediction to obtain a future wind speed prediction result. The prediction result of the method is highly consistent with the observation value in numerical value, and the method also shows high accuracy in trend prediction.

The invention provides a wind power wind speed prediction and optimization arrangement method, system and device. The wind power wind speed prediction and optimization arrangement method comprises the steps of obtaining wind power wind speed basic data of a target area; constructing a neural network model based on physical information based on the wind power wind speed basic data; acquiring a wind speed at a specific position in a speed field from a neural network model based on physical information, and creating a training data set; training the neural network model based on the physical information based on the training data set; reconstructing wind speed distribution in a speed field through the trained neural network model based on physical information, and predicting to obtain a wind speed prediction value of the wind power plant in the next time period; and optimizing the arrangement of the wind turbine group according to the wind speed distribution in the reconstructed speed field. According to the method, the two-dimensional flow velocity field of the wind power plant can be reconstructed by training one PINN, and accurate wind speed ultra-short-term prediction is realized based on the trained model.