51results about How to "Improve predictive ability" patented technology

The invention provides an unmanned aerial vehicle multi-target tracking method and system based on visual and millimeter-wave radar information fusion, and the method comprises the steps: obtaining an association result through employing a space-time consistency model, constructing a self-adaptive joint probability data association frame according to the association result, so as to obtain an initial target state estimation value, and carrying out the recognition of the initial target state estimation value. A prospective path planning algorithm is adopted to obtain an optimal observation angle flight path, a Bayesian network is utilized to obtain initial observation data, based on the initial observation data, an optimal target state estimation value is generated, a distributed cooperative sensing mechanism is used to generate an initial tracking task allocation scheme, and an intelligent resource allocation strategy is implemented. Generating an initial resource allocation scheme, starting a quick response protocol to obtain an optimal tracking task allocation scheme, and generating an optimal resource allocation scheme based on the optimal tracking task allocation scheme; according to the technical scheme provided by the invention, the accuracy of target identification is improved, the environmental adaptability is enhanced, the tracking efficiency is improved, and the working time is prolonged.

The invention discloses a 5G message video monitoring method based on artificial intelligence. The method comprises the following steps: S1, constructing a multi-modal data set; s2, multi-modal fusion features are generated by using a joint convolutional neural network; s3, generating a target behavior trend curve and a target behavior prediction result by using the 3D convolutional neural network; s4, generating a target threshold curve by using an adaptive algorithm based on variable domain search, and obtaining behavior abnormality warning information; s5, generating comprehensive alarm information by using a method of combining fuzzy logic reasoning and a Bayesian network; s6, cross-camera data space-time correlation reasoning is carried out by using the graph neural network; s7, constructing an event severity curve by using a multivariable regression analysis and analytic hierarchy process, and generating a severity level and an event response measure of the event; and S8, pushing alarm information and event response measures to related personnel or equipment through the 5G message platform. The invention provides an efficient video monitoring method in combination with artificial intelligence and a 5G network.

The invention relates to the technical field of APT killing chain reconstruction and prediction methods, and provides an APT killing chain reconstruction and prediction method and system based on causal reasoning. The method aims at solving the problems that in an existing APT detection and defense technology, multi-source heterogeneous data integration is difficult, space-time dynamic feature modeling is insufficient, killer chain reconstruction is inaccurate, the predictive capacity is limited, interpretability is poor, and real-time performance and accuracy are difficult to consider at the same time. The method comprises the steps of obtaining multi-source heterogeneous data, constructing a causal graph, performing multi-modal feature extraction based on the causal graph, and obtaining node initial feature representation; inputting the node initial feature representation into a time sequence causal graph convolution learning algorithm to obtain an optimized node representation; executing killer chain reconstruction based on the optimized node representation to obtain a possible killer chain sequence; predicting a future attack path according to the possible killing chain sequence to obtain a potential attack path; and generating an interpretability analysis report based on the potential attack path.

The invention discloses a new energy automobile lithium battery thermal runaway monitoring system and method. The system comprises a vehicle-mounted lithium battery, a multi-source data acquisition module, a multi-source data fusion module, a hidden Markov model monitoring module, a state identification module, a data analysis processing module, an early warning module, an emergency processing module, a wireless communication module, a cloud platform and a battery management system. The multi-source data acquisition module is responsible for acquiring multi-dimensional data such as voltage, temperature and internal resistance of a lithium battery, and the data fusion module processes the data and then transmits the data to the hidden Markov model monitoring module for state monitoring and prediction. According to the system, a high-order hidden Markov model and a hidden semi-Markov model are adopted, state retention time modeling and a self-adaptive online learning mechanism are combined, and conversion of a battery from a normal state to an abnormal state and a thermal runaway state can be accurately recognized. The system remarkably improves the safety and stability of the lithium battery of the new energy automobile, and is suitable for power battery management of the new energy automobile.

The invention provides a micro robot motion control method and system, and the method comprises the steps: carrying out the synchronization and calibration processing of environment state information, and generating a comprehensive perception image; analyzing the comprehensive perception image to obtain a classification result, and predicting the environment state change influencing the motion track of the micro robot in a period of time in the future based on the classification result; generating a candidate path by using a fast random tree, and optimizing the candidate path in combination with an AI search algorithm to obtain a target path; constructing a three-dimensional dynamic probability graph, continuously updating the three-dimensional dynamic probability graph by using a reinforcement learning algorithm, and generating an instant decision map; and estimating the position of the micro-robot by using a Kalman filter to obtain the position state of the micro-robot, adjusting the speed and steering angle of a driving motor of the micro-robot, and generating trajectory tracking of the micro-robot. According to the technical scheme provided by the invention, the flexibility and intelligence of the micro robot are improved.

The invention is suitable for the technical field of escalator monitoring, and provides an escalator health assessment method and device based on deep learning and terminal equipment. The method comprises the steps that escalator operation data are collected through a sensor, and the escalator operation data comprise escalator vibration signals, escalator temperature, escalator operation noise and escalator operation current; local features used for representing the health state of the escalator are extracted from the escalator operation data through a convolutional neural network, the local features of various escalator operation data are fused through an attention mechanism, and global health state representation of the escalator is obtained; and an escalator domain knowledge graph is constructed, when the global health state characterization shows that the escalator state is abnormal, the reason of the escalator state abnormality is reasoned based on the relationship between entities in the escalator domain knowledge graph, and the escalator health condition is evaluated according to the reason of the escalator state abnormality. The escalator operation data can be comprehensively analyzed, the health state of the escalator is effectively evaluated, and the safety, reliability and operation efficiency of the escalator are improved.

The invention discloses a data resource management system and method based on artificial intelligence, and belongs to the technical field of artificial intelligence. According to the method, multi-source data is collected, and data cleaning and preprocessing are carried out; selecting meteorological indexes and soil indexes, and calculating an influence factor of each index in the meteorological indexes and the soil indexes; determining the weight of each influence factor, and calculating to obtain an agricultural product external environment influence factor; historical data are collected, meteorological data and soil data are matched with agricultural product planting areas and planting periods, and a multi-dimensional data set is constructed; performing feature engineering on the multi-dimensional data set to obtain a feature set; training a long short-term memory network model, and performing model evaluation and optimization; after training is completed, meteorological data and soil data are input into the model, and agricultural product yield prediction data are output; and performing data alignment, comparing the predicted agricultural product yield value with the average value of the market demand, and analyzing a confidence interval to obtain a decision suggestion.

The invention provides an optical power prediction method based on WRF-Solar assimilation multi-source data and a related device, and belongs to the technical field of optical power prediction. According to the method, a WRF-Solar model and a WRF-DA 4D-Var assimilation framework are constructed, preprocessed historical meteorological data and historical radiation observation data are input into the WRF-Solar model and the WRF-DA 4D-Var assimilation framework, and a prediction result using the historical meteorological data and the historical radiation observation data is obtained; performing error analysis on prediction results of the historical meteorological data and the historical radiation observation data to obtain an optimal WRF-Solar model parameter and an optimal WRF-DA 4D-Var assimilation parameter; the optimal WRF-Solar model parameters and the optimal WRF-DA 4D-Var assimilation parameters are utilized to construct a final prediction model; predicting the solar radiation and meteorological elements to be predicted by using the final prediction model to obtain prediction results of the solar radiation and meteorological elements; and obtaining an optical power prediction result based on the prediction results of the solar radiation and the meteorological elements. According to the invention, the problem of low accuracy of optical power prediction is solved.

The invention discloses an AI-based printing success rate judgment method and system. The method comprises the following steps: firstly, collecting real-time data of a printer; then preprocessing the real-time data of the printer to generate feature data used for judging the printing success rate; constructing an initial multi-dimensional feature prediction model for judging a printing success rate, and training the initial multi-dimensional feature prediction model by using historical printing job data to obtain a trained target multi-dimensional feature prediction model; and finally, inputting the feature data into a trained target multi-dimensional feature prediction model, and outputting a success rate prediction result of the printing job. By providing an AI-based target multi-dimensional feature prediction model, the pre-judgment capability of printing faults is improved, printing parameters can be adjusted in real time, the printing process can be optimized, and the printing failure rate can be reduced.

The invention relates to the technical field of traditional Chinese medicine detection, in particular to a UPLC-Q / TOF-MS identification method for pinellia ternate and rhizoma typhonii flagelliformis. According to the method, on the basis of pinellia ternate, ginger processed pinellia, rhizoma typhonii flagelliformis and ginger processed rhizoma typhonii flagelliformis component information collected by ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry, characteristic substances are screened in combination with a multivariate statistics method, OPLS-DA analysis is carried out on pinellia ternate and rhizoma typhonii flagelliformis, differential compounds of the pinellia ternate and the rhizoma typhonii flagelliformis are found, and The method comprises the following steps: determining five key compounds for identifying pinellia ternate, rhizoma typhonii flagelliformis and processed products thereof by combining a plurality of analysis methods such as box plot analysis, thermogram analysis and correlation analysis, then determining the relative content of the five key compounds, and establishing a pinellia ternate and rhizoma typhonii flagelliformis identification model by utilizing OPLS-DA analysis. The detection of rhizoma typhonii flagelliformis doped in the rhizoma pinelliae and processed products thereof can be realized.

The invention discloses a drug-target affinity prediction method based on a multi-scale convolutional neural network, and the method comprises the steps: obtaining a sequence of a target protein and an SMILES character string of a drug small molecule, obtaining an SMILES word vector matrix and a protein word vector matrix based on word vector coding, constructing the multi-scale convolutional neural network, carrying out the learning of the SMILES word vector matrix, and carrying out the prediction of the drug-target affinity. The method comprises the following steps: constructing a protein word vector matrix, acquiring chemical and structural characteristics of drug small molecules, constructing a one-dimensional convolutional neural network, learning the protein word vector matrix, acquiring structural characteristics of target protein, and fusing the chemical and structural characteristics of the drug small molecules and the structural characteristics of the target protein through splicing operation. And expressing the fused features as drug-target features, constructing an affinity prediction model, and learning the drug-target features to obtain an affinity prediction value between the drug and the protein. Compared with a graph neural network and a pre-training model, the drug-target affinity can be better predicted, and the prediction capability is better than that of a deep learning method.

The invention belongs to the field of geological engineering construction, discloses a TBM (Tunnel Boring Machine) real-time slag monitoring system and a TBM real-time slag monitoring method for advanced geological forecast, solves the problems of lack of real-time performance, continuity and accuracy and low processing efficiency of surrounding rock state analysis in the traditional technology, and provides powerful support for advanced geological forecast in a TBM tunneling process. The method comprises the following steps: continuously acquiring rock slag images to obtain a rock slag acquisition video; cutting the rock slag acquisition video into an image sequence with a fixed time interval, and aligning the timestamp of each rock slag image with the recording time of the TBM tunneling parameters; then the cut rock slag image is preprocessed; then identifying the preprocessed rock slag image to obtain a classification result of the rock slag; and finally, the classification result of the rock slag and TBM tunneling parameters are fused to serve as input, and the surrounding rock grade is calculated through a surrounding rock grade grading model so that the tunneling strategy can be adjusted in real time. The method is suitable for underground engineering construction scenes.

The invention provides an earthquake nucleation seismic phase identification and model training method and system, a terminal and a medium, and particularly relates to the technical field of data monitoring. Performing feature extraction on the to-be-measured seismic data to obtain a plurality of types of data features; a trained machine learning model is utilized to process the data features, a plurality of seismic phase moments are determined, and the machine learning model is obtained through training based on seismic simulation data of different roughness; and based on the seismic phase moment, obtaining an earthquake nucleation seismic phase identification result. According to the scheme, the seismic simulation data with different roughness are used as training samples to train the model, so that the diversity and data volume of the training samples are increased, and the generalization ability of the model is improved; and meanwhile, the machine learning model has the capabilities of processing complex time sequence data and multi-target prediction, so that the trained model has good prediction capability, can process the complex time sequence data and is beneficial to improving the nucleation seismic phase recognition precision.

The invention discloses a ship lift accident safety mechanism thread pair gap detection device and method. The device mainly comprises a laser gap detection assembly and a via hole electric slip ring assembly. The laser gap detection assembly comprises a hoop, the hoop is mounted on the spherical trunnion, a spiral support supporting surface is mounted at the upper end of the hoop, a plurality of laser sensors are mounted on the spiral support supporting surface, and the plurality of laser sensors are spirally distributed and have the same lead angle as the safety screw; the plurality of laser sensors emit laser in the radial direction of the spherical trunnion, and the laser is vertically emitted to the nut column 1; the via hole electric slip ring assembly comprises a stator and a rotor. The lower end of the stator is mounted on the upper guide frame, the rotor is mounted on the spherical trunnion, a conducting ring is embedded in the rotor, an electric brush is embedded in the stator, and the conducting ring is in contact with the brush; the via hole electric slip ring assembly is used for supplying power to the laser sensor; meanwhile, the via hole electric slip ring assembly transmits data detected by the laser sensor to the PLC. According to the ship lift accident safety mechanism thread pair gap detection device and method, real-time detection of the thread pair gap can be achieved.

This invention discloses a liquid-cooled collaborative intelligent temperature control method and system based on layered PCM, relating to the field of liquid-cooled temperature control technology. The method includes: S1, real-time acquisition of liquid-cooled temperature control data and preprocessing of the data; S2, calculation of the sensible and latent heat absorption of the outer and inner PCM layers, assessment of the risk of thermal imbalance on the conductor surface, determination of the current thermal state, and output of a thermal risk warning signal; S3, assessment of the degree to which cable bending weakens thermal coupling performance, triggering a layered PCM collaborative temperature control strategy, and execution of graded control commands for the power pump, circulation pump, and three-way valve; S4, assessment of the deviation between the current heat exchange capacity and the heat load demand, correction of the cooling response coefficient, and optimization of the PCM collaborative temperature control strategy. This solves the problems of unstable thermal coupling between the coolant pipe and the heating conductor in existing liquid-cooled cables, and the tendency for pipe deformation and contact thermal resistance and localized hot spots to occur with frequent bending.

The invention provides a method for rapidly characterizing physical parameters of traditional Chinese medicine powder. The method comprises the following steps: (1) determining a near infrared spectrum of a traditional Chinese medicine powder sample; (2) measuring physical parameters of the traditional Chinese medicine powder sample, and constructing a physical fingerprint spectrum; and (3) preprocessing the near infrared spectrum, screening spectrum variables related to the physical parameters, establishing a spectrum variable model, and representing the physical parameters. According to the method for rapidly representing the physical parameters of the traditional Chinese medicine powder, rapid conversion of the physical fingerprint spectrum of the powder is achieved; after model performance investigation, the GRNN model is found to have the best prediction capability on the physical parameters of the medicinal material powder, and the GRNN simplified model still shows satisfactory prediction performance; the result proves that the NIR spectrum technology has great application potential in the rapid characterization of the physical properties of the pellet core and the traditional Chinese medicine powder.

The application provides a trajectory planning method and system of a mechanical arm continuous synchronization mode, relates to the technical field of mechanical arms, and collects operation related data of a mechanical arm system; a digital twin model is used to simulate the movement process of the mechanical arm along a preset movement trajectory, so as to generate a risk prediction result of the mechanical arm in the continuous synchronization mode; the preset movement trajectory is divided into multiple trajectory segments, and each trajectory segment is optimized and integrated based on the risk prediction result and the movement characteristic parameters of the mechanical arm, so as to form a preliminary operation trajectory; operation state data is analyzed, so as to generate data anomaly information; each mechanical arm in the mechanical arm system is controlled to move according to the preliminary operation trajectory, and real-time operation data of each mechanical arm is collected; the preliminary operation trajectory is adjusted in combination with the data anomaly information and the risk prediction result, so as to obtain a target operation trajectory of the mechanical arm, and safe, efficient and self-adaptive target trajectory operation of the mechanical arm in the continuous synchronization mode is realized.

The invention discloses a biomarker for predicting severe treatment of a COVID-19 patient and application of the biomarker, and relates to the technical field of biomedicine. According to the invention, a COVID-19 patient severity prediction or auxiliary prediction model which is obtained by constructing biomarkers including Gemelia haemolsos, Streptococcus sp.A12, Fusobacterium pseudoperiodontium, Gemelia mollissima, Villonella atypical, Villonella rogosa and Prevotella melaninogena is found, the model is used for predicting and evaluating the severity of the COVID-19 patient, the sensitivity is high, and the specificity is good. The invention further discloses a method for predicting and evaluating the severity of the COVID-19 patient severity, and the method has the advantages that the method can be used for predicting and evaluating the severity of the COVID-19 patient severity;

This invention relates to the field of fall detection technology, specifically disclosing a millimeter-wave radar fall detection method that integrates HRV nonlinear features and 3D point clouds. The method includes the following steps: acquiring the inter-heart rate interval (IBI) sequence and 3D point cloud of the target in parallel based on millimeter-wave radar signals; extracting HRV nonlinear physiological features based on the IBI sequence and biomechanical features based on the 3D point cloud; aligning the HRV nonlinear physiological features and biomechanical features to a unified time axis and performing feature coupling calculations to obtain four types of coupling indices; the four types of coupling indices are: cardiac-gait phase coupling, exercise-induced heart rate response, cardiac-triggered motion variability, and joint entropy; and performing fall detection on the target based on a three-level progressive detection mechanism. This invention effectively captures abnormal physiological signals before a fall by establishing an intrinsic correlation between physiological and motion signals, effectively solving the problem of warning lag caused by traditional methods relying solely on motion features.

The invention relates to a reliability method for an IED (Intelligent Electronic Device) of a transformer substation under an intrusion intention, which is characterized by comprising the following steps of: constructing a Petri diagram by taking a firewall state, an isolation state of IPS on IED equipment and an IED equipment state as positions and taking state change of a safety system of the transformer substation in an intrusion process as conversion; simulating the Petri diagram to obtain the intrusion success probability of the substation safety system, and constructing a Markov state transition matrix of the substation safety system in combination with the natural failure rate of the system; and solving the steady-state probability of each state in the state transition matrix when the substation safety system reaches a steady state, and constructing a reliability index of the substation safety system by using the steady-state probability of each state so as to evaluate whether the substation safety system reaches a system safety requirement or not.

The application provides a quantitative analysis method for ore pore and fissure structure, and relates to the technical fields of mineral processing, metallurgical engineering and geological analysis. The method realizes accurate prediction of effective porosity of heap leaching ore by coupling on-site particle size investigation, laboratory "quasi on-site" heap construction, ore block fissure microscopic measurement and mass-volume conservation calculation. The method effectively overcomes the technical problems that the traditional method depends on single scale measurement, and is difficult to comprehensively reflect the complex structure from nanoscale pore to centimeter scale fissure, especially for the mixed ore of coarse and fine particle grades. The existing technology often processes different particle grades respectively, lacks systematic integration and mutual correction, and causes deviation of the calculation results of porosity and fissure rate.

The present invention discloses a ship motion response prediction method based on Gaussian interpolation, which relates to the technical field of ship motion response prediction, and includes the following steps: S1, data acquisition and preprocessing, including data cleaning, coordinate system conversion, data standardization and data set division; S2, constructing a Gaussian interpolation model, including setting an initial σ value and selecting interpolation points; S3, optimizing model parameters by a random search algorithm, including calculating the weight of each point in a training set, model verification, and model parameter optimization; S4, ship motion response prediction, substituting the ship to be predicted and sea state data into the Gaussian interpolation model to predict the ship motion response. The present invention uses a Gaussian interpolation model to accurately simulate the motion response of a ship under complex sea conditions. Compared with traditional methods, the Gaussian interpolation model can more accurately depict the complex relationship between sea state parameters and the ship motion state with its excellent fitting ability for multidimensional and nonlinear data.

The application discloses an energy storage power station intelligent optimization method and system based on deep coupling learning, relates to the technical field of energy storage power station optimization, and is used for solving the problems of maximum demand capture failure caused by production plan fluctuation and invalid charging and discharging loss caused thereby, extracting strong correlation characteristics of production procedures and power mutation, constructing a time-space matrix anchor peak value inducement, outputting a latent space feature vector through an encoder by using a deep coupling multi-task network, explicitly quantifying a demand refresh risk probability, introducing the peak value risk and a battery health soft measurement quantity into a target function based on an MPC-DRL rolling optimization framework, dynamically adjusting charging and discharging sensitivity, constructing a composite feedback signal by extracting extreme samples of missed or false peak values, and realizing closed-loop evolution of attention reconstruction and decision threshold prediction; the robustness of demand management under a complex production environment is improved, the power consumption cost is reduced, and the whole life cycle optimal balance of battery life and economic benefits is realized.

The invention belongs to the field of power grid data prediction and processing methods, and particularly relates to a multi-source data comprehensive prediction processing method for a power grid system. A sensor device is deployed at each node of a power grid system, power grid operation data are collected in real time, verification and standardization processing are carried out, and global and local standardized data sets are formed. On this basis, a global guidance model and a local learning model are constructed, and efficient prediction of data is realized through knowledge transmission. And meanwhile, a trust function and a trust threshold mechanism are introduced, so that the reliability of prediction data is ensured. And finally, according to the characteristics and trends of the global and local data, making a corresponding decision rule, and realizing real-time monitoring and optimal control of the power grid system. According to the method, the accuracy and the real-time performance of power grid data prediction are improved, the adaptability to special conditions of a power grid system is enhanced through the combination of the global model and the local model, and powerful support is provided for efficient, stable and safe operation of a smart power grid.

The invention relates to the technical field of food biology, in particular to a method for predicting the activity of amylase subjected to an inactivation process or inactivated amylase. The invention provides a method for predicting an inactivation process based on required amylase activity, which comprises the following steps: providing the required amylase activity, presetting an inactivation temperature, and calculating the inactivation time by adopting an amylase activity-inactivation time prediction model according to the required amylase activity. The invention also provides a method for predicting the activity of the inactivated amylase based on the inactivation process, which comprises the following steps: providing the inactivation temperature and the inactivation time, and calculating the activity of the inactivated amylase by adopting the amylase activity-inactivation time prediction model according to the inactivation temperature and the inactivation time. The method disclosed by the invention has relatively good predictive capacity, and is beneficial to accurately controlling the inactivation process conditions; by optimizing the inactivation process, the production time and cost can be reduced, and the production efficiency is improved; the enzyme activity is strictly controlled or the enzyme is inactivated, so that the food quality safety is ensured.

The application discloses a bridge structure self-adaptive sampling monitoring method and device suitable for an extreme road environment, and belongs to the technical field of bridge structure health monitoring.The core target of the application is to solve the problems of repeated arrangement of existing monitoring equipment, road-bridge data fragmentation, and low self-adaptive sampling regulation precision.Through innovative design of a road-bridge multiplexing integrated sensing node, single-node multi-parameter and full-scene monitoring are realized, and the arrangement cost and construction difficulty are reduced; through construction of a self-adaptive sampling strategy based on fuzzy logic control, a feasible algorithm model is given, continuous dynamic regulation of sampling parameters is realized, monitoring precision and low power consumption are taken into account, deep collaborative linkage of road environment monitoring and bridge structure monitoring is realized, and the safe operation of the bridge structure under the extreme road environment is ensured.

The application provides a content review model training method and device, equipment, a storage medium and a computer program product; the method comprises the following steps: predicting compliance of at least one piece of content to be reviewed by a content review model to obtain a predicted review result of each piece of content to be reviewed; for each piece of content to be reviewed, determining a predicted difficulty score of the content to be reviewed based on the predicted review result; from at least one piece of content to be reviewed, a target review content is selected whose predicted difficulty score meets a difficulty score condition; the content review model is trained by taking the target review content as a sample and taking the actual review result of the target review content as a sample label of the sample, so as to update the content review model; by the application, the training cost of the content review model can be reduced, and the training speed and content review accuracy of the content review model can be improved.

The present invention discloses a soft sensing method for effluent ammonia nitrogen based on a dual-correlation pruning strategy neural network. First, a high-dimensional mutual information method is used to extract input features to reduce input redundancy and computational complexity. Then, a self-construction strategy consisting of a growth scheme and a pruning scheme is proposed to dynamically optimize the fuzzy neural network model to obtain a neural network SCFNN-DCP based on a dual-correlation pruning strategy. In the growth stage of fuzzy rules, the system error judgment criterion and the containment boundary criterion are used to generate fuzzy rules to enhance the generalization ability of the network. In the pruning stage of fuzzy rules, the dual-correlation criterion is used to remove fuzzy rules from the rule base to reduce the complexity of the network. At the same time, a dynamic adjustment scheme is used to update the network width. Finally, a recursive least squares algorithm with a stage forgetting factor is used to estimate the network weights online to obtain a strong prediction ability. The present invention can ensure the stable and safe operation of the water quality treatment process.

The application provides a disaster-causing similar typhoon retrieval model acquisition method, a retrieval method, a terminal and a medium. The method comprises the following steps: acquiring spatial distribution information and meteorological field information of each path observation point in a target typhoon and related typhoons, and dividing the target typhoon into a first target typhoon and a second target typhoon, and dividing each related typhoon into a first related typhoon and a second related typhoon; based on the path similarity and the meteorological field similarity between the first target typhoon and the corresponding first related typhoon, and in combination with the fitting similarity between the second target typhoon and the corresponding second related typhoon, the weight parameters in the disaster-causing similar typhoon search model are optimized and configured to obtain a final disaster-causing similar typhoon search model; and the configured disaster-causing similar typhoon search model is used to obtain the corresponding disaster-causing similar typhoon of a typhoon to be retrieved. The application improves the prediction ability of the disaster-causing similar typhoon for the path abnormal typhoon and improves the accuracy of the disaster-causing similar typhoon search.

The present application provides a method and device for converting surface wind speed. The surface wind speed conversion model trained by the present application can convert wind speeds, for example, convert the wind speed at a height simulated by the offshore wind energy resource coupling model to obtain the wind speed at another height. The surface wind speed conversion model trained by the present application is combined with a random forest model, which can balance the amount of calculated data, the accuracy of the calculation, and the efficiency of the calculation, effectively reducing the number of features while retaining the important information required for wind speed conversion. Combined with the powerful predictive ability of the random forest model, the efficiency and accuracy of wind speed conversion are improved. In this way, the surface wind speed conversion model trained by the present application has higher efficiency and accuracy in converting surface wind speeds.