78results about How to "Guaranteed forecast accuracy" patented technology

The invention relates to the technical field of intelligent evaluation and early warning methods, in particular to a nuclear radiation dose space distribution intelligent evaluation and early warning method and system, and the method comprises the following steps: obtaining multi-source monitoring data which comprises gamma radiation dose, alpha radiation dose and beta radiation dose measuring instrument data; performing fusion and enhancement processing on the multi-source monitoring data to obtain an enhanced radiation monitoring signal; based on the enhanced radiation monitoring signal, high-precision prediction of nuclear radiation dose spatial distribution is carried out through a deep probabilistic neural network model; according to the high-precision prediction result, in combination with a nuclear accident severity level standard, a radiation alert evaluation result and early warning information are generated, short-term radiation fluctuation can be captured, a long-term change trend can be identified, and the understanding ability of the system to a complex radiation field is greatly improved.

The invention relates to the technical field of lighting, in particular to a driving power supply control circuit, a control method and a lighting application method. Comprising the steps that an acquisition module acquires spectral data and temperature data of a current environment; the receiving module receives the target color rendering index, the target color temperature and the target illumination; the control module is connected with the acquisition module and the receiving module, and is used for calculating a PWM duty ratio needing to be compensated according to the temperature data, obtaining a target PWM duty ratio based on the current color rendering index Ra, the current correlated color temperature CCT, the current illumination and the control parameters, and outputting a PWM signal based on the target PWM duty ratio and the PWM duty ratio needing to be compensated; and independent and stable constant-current signals are finally output by using the voltage balancing circuit and the driving module. The problems that an existing LED driving power source is single in function and lacks self-adaptive control capacity and monochromatic light trailing are solved. The LED powder-free lamp can continuously and stably operate, and meanwhile, high color rendering index and dynamic light color balance are achieved.

The invention discloses a data-based XGBoost model construction method for evaluating the cerebral hemorrhage recurrence risk of a patient, and the method comprises the steps: S10, reading and processing the data of the cerebral hemorrhage patient, and obtaining the data of a training set; s20, receiving the training set data, processing the training set data by adopting LASSO regression and a step-by-step recursion program, and outputting a plurality of variable features; s30, judging whether the p value of each characteristic variable is less than 0.05 or not; s40, when the p value of the characteristic variable is smaller than 0.05, outputting the variable characteristic; s50, constructing and optimizing parameters of the first model, and outputting an XGBoost model; and S60, carrying out visual interpretation on an evaluation result of the XGBoost model by using an SHAP (Secure Hardware Application Program) and an LIME (Local Instrumentation Model Element) interpreter Compared with the prior art, the method provided by the invention has the advantages that the model resolution AUC is 0.92 and the prediction accuracy is 0.95 by taking nine characteristic factors as internal verification of the model; the model resolution ratio AUC tested by the test set is 0.93, and the prediction accuracy is 0.91.

The invention provides a geotechnical engineering settlement prediction method which comprises the following steps: S1, performing settlement prediction by adopting a traditional regression analysis method to obtain a regression curve; s2, calculating a historical regression value on each historical time node and a historical deviation value of the monitoring data; the historical regression value is a value of a regression curve on a historical time node; s3, based on the historical deviation value of each historical time node, performing Kriging interpolation on a future time node, and calculating a regression prediction deviation value; s4, the regression prediction deviation value on the future time node obtained through interpolation and the regression prediction value on the future time node are added, a final settlement prediction value is obtained, and the regression prediction value is the value on the regression curve on the future time node. Optimization is carried out on the basis of a traditional regression analysis method, and the settlement prediction precision is effectively improved.

The invention provides a wind power plant short-term power prediction method and system based on similar days of meteorological elements, and relates to the field of data processing, and the method comprises the steps: carrying out the data processing of basic information, operation data, meteorological data and landform data of a plurality of wind and light hydropower stations in a target region, carrying out the clustering of the plurality of wind and light hydropower stations, and carrying out the clustering of the plurality of wind and light hydropower stations; determining a plurality of wind power generation clusters, a plurality of photovoltaic power generation clusters and a plurality of hydroelectric power generation clusters, and establishing a plurality of candidate wind power generation prediction models, a plurality of candidate photovoltaic power generation prediction models and a plurality of candidate hydroelectric power generation prediction models; generating a plurality of candidate combination prediction models corresponding to the wind-solar hydropower station, and determining an optimal combination prediction model; obtaining weather forecast data and hydrological forecast data of the wind and light hydropower station to be predicted in the target time period; and the wind power generation power, the photovoltaic power generation power and the hydroelectric power generation power are predicted through the corresponding optimal combination prediction model. The method has the advantage of improving the accuracy of water-wind-light power generation power prediction.

The invention relates to the technical field of intelligent traffic, and particularly discloses a pedestrian movement track prediction method and system based on visual information and a storage medium, and the method comprises the following steps: unmanned aerial vehicle data collection, coordinate projection transformation, coordinate sampling and track aggregation, sampling of pedestrian ground real coordinates after projection transformation, and tracking of the pedestrian movement track. Aggregating the coordinate data into trajectory information of a plurality of pedestrians; spatial visual information and time information are combined, a global interactive directed topological graph is constructed, then spatial feature and time feature extraction is performed, time sequence features of pedestrian motion are efficiently extracted through a multilayer convolution structure, and finally the future motion trail of pedestrians is predicted. Through the method, pedestrian motion in a complex scene can be predicted more efficiently, and the application effect of the model in pedestrian flow management and public safety planning is enhanced. The future movement condition is predicted by analyzing the historical trajectory of pedestrians, the pedestrian flow management is obviously optimized, and the safety level of public places is improved.

The invention discloses a trajectory prediction method and system for a bucket of a backhoe dredger, the next position of the bucket is predicted according to an observation model, a translational motion model and an external disturbance model, and the observation model, the translational motion model and the external disturbance model are introduced. The observation model, the translational motion model and the external disturbance model are integrally considered, and multiple interactions of the observation model, the translational motion model and the external disturbance model are realized, so that the prediction accuracy of the next position of the bucket is ensured. Further, the next position of the bucket is corrected, and a corresponding correction value is output; the translation transformation vector of the backhoe dredger is defined on the basis of real-time updating of the correction value, the position prediction value of the bucket is defined according to the translation transformation vector of the backhoe dredger and the bucket, the corresponding track is defined on the basis of the multiple position prediction values, association of the multiple position prediction values is achieved, and track prediction accuracy is guaranteed.

The invention relates to the technical field of database construction, in particular to a construction method and system based on an encephalopathy nursing knowledge base. The method comprises the following steps: acquiring multi-source encephalopathy data; performing structured data screening on the multi-source encephalopathy data to generate structured encephalopathy data and unstructured encephalopathy data; performing multi-modal data structured layering on the unstructured encephalopathy data to generate multi-modal structured layered data; performing data preprocessing on the multi-modal structured hierarchical data and the structured encephalopathy data to generate a standard structured encephalopathy data set; performing context analysis on the standard structured encephalopathy data set to generate an encephalopathy information semantic vector set; and performing redundant relation pruning processing on the encephalopathy information semantic vector set to generate spatial semantic optimization data. According to the method, by integrating multi-source data, structured and unstructured information, real-time dynamic mapping nursing requirements and system feedback evaluation, the construction quality and practicability of the encephalopathy nursing knowledge base are improved.

The invention belongs to the technical field of data processing, and particularly relates to an intelligent monitoring method for water pollution damage, and the method comprises the steps: dividing a sewage discharge region into a plurality of sub-regions, arranging a plurality of intelligent sensors in each sub-region, collecting sewage component data of the sub-regions, carrying out the data integration based on a collection timestamp, and obtaining combined data, the combined data belonging to different sub-regions are fused to generate fused data; after the remote server receives the fused data, the pollution degree of the fused data is judged based on the score and the weight value; the pollution degree of the fused data is judged based on the score and the weight value; and on the basis of the predicted sewage component data and abnormal conditions, generating corresponding countermeasures by using an artificial intelligence model, estimating the implementation time of the countermeasures, and evaluating the implementation effect of the countermeasures on the basis of newly collected fusion data. According to the invention, the pollution degree of water pollution can be rapidly and accurately judged.

The invention provides a battery remaining life prediction optimization method based on an interpretable deep learning algorithm. The method comprises the steps of constructing a data set and performing preprocessing; constructing model input data based on the preprocessed data set; constructing a battery residual life prediction model based on deep learning; interpreting decision behaviors of the model in the prediction process through an interpretable deep learning algorithm, and quantifying the contribution degree of input data in the prediction process; and optimizing aging data and a prediction process according to a model explanation result. According to the invention, the marginal contribution of the input features to the predicted value of the remaining life of the battery output by the neural network is calculated through the interpretable deep learning algorithm, and the contribution degrees of data of different working conditions, different feature parameters and different aging stages in the life prediction process are quantified, so that important aging features are identified and useless data are removed. The prediction process is optimized from the aspect of a data source, and the prediction speed is accelerated while the prediction accuracy is ensured, so that the consumption of computing resources is reduced.

The invention relates to the field of intelligent medical treatment, and provides an epileptic seizure prediction method based on intra-channel and inter-channel feature mixing of an electroencephalogram, and the method comprises the steps: obtaining an original electroencephalogram, and carrying out the filtering and segmentation of the original electroencephalogram, and generating a plurality of electroencephalogram window images; inputting the electroencephalogram window image into a plurality of feature mixing modules for feature extraction to obtain mixed features; each feature mixing module comprises an in-channel feature extractor and an inter-channel feature extractor; the in-channel feature extractor is used for fusing mixed time dimension features in each channel; the inter-channel feature extractor is used for fusing inter-channel mixed spatial dimension features; and inputting the mixed features into a prediction head for classification, and determining an epileptic seizure prediction result. The epileptic seizure prediction method solves the defect of low accuracy of epileptic seizure prediction in the prior art, realizes efficient and accurate epileptic seizure prediction, and improves the performance and generalization ability of the prediction model.

The invention discloses a computing power resource optimization method and system based on an AI large model and a storage medium, and relates to the technical field of computing power resource regulation and control optimization, and the method comprises the following steps: collecting task sample data of a computer processing task, and carrying out the data preprocessing; constructing a task processing time prediction model, and predicting the time length required by the computer to process the task to obtain task time length information; constructing a task processing progress evaluation method, and calculating and evaluating the task processing progress of the computer to obtain task progress information; based on the task duration information and the task progress information, dynamically adjusting and optimizing the computing power resource in the process of processing the task by the computer; the method is used for solving the problem that the existing computing power resource regulation and control optimization technology cannot accurately evaluate the task progress by using multi-dimensional data while predicting the task processing time, so that the computing power resource is dynamically regulated and optimized according to the real-time change of the task progress in the task processing process.

The invention discloses a comprehensive method for analyzing compressibility of a shale reservoir based on machine learning, and relates to the technical field of oil and gas field development, the comprehensive method for analyzing compressibility of the shale reservoir based on machine learning mainly comprises the following steps: obtaining a training data set according to core experiment data and corresponding logging data; constructing a three-parameter prediction model by using a random forest regression model; training the three-parameter prediction model by using the training data set to obtain a trained three-parameter prediction model, and predicting the to-be-predicted logging data to obtain three parameters; and calculating a compressibility coefficient based on the three parameters so as to obtain compressibility classification evaluation of the shale reservoir in the research area. By implementing the comprehensive method for analyzing the compressibility of the shale reservoir based on machine learning provided by the invention, the compressibility of the shale reservoir can be efficiently, accurately and universally evaluated.

The present application relates to the technical field of bearing life prediction, in particular to a kind of gradient composite coating sliding bearing life prediction method, comprising the following steps: step one, build the coating degradation physical simulation model of multi-field coupling;Step two, adopt active learning algorithm to build high-precision proxy model, design active learning query strategy;Step three, online monitoring and multi-domain feature extraction;Step four, real-time state mapping and life prediction based on transfer learning;Step five, prediction result output and model updating.The present application can improve the accuracy of gradient composite coating sliding bearing life prediction, greatly reduce the operation load, and improve the prediction efficiency of bearing life.

The present invention relates to the technical field of database construction, and in particular to a method and system for constructing a knowledge base for brain disease nursing. The method comprises the following steps: acquiring multi-source brain disease data; performing structured data screening on the multi-source brain disease data to generate structured brain disease data and unstructured brain disease data; performing multi-modal data structured stratification on the unstructured brain disease data to generate multi-modal structured stratified data; performing data preprocessing on the multi-modal structured stratified data and the structured brain disease data to generate a standard structured brain disease data set; performing context analysis on the standard structured brain disease data set to generate a semantic vector set of brain disease information; performing redundant relationship pruning processing on the semantic vector set of brain disease information to generate spatial semantic optimized data. The present invention improves the construction quality and practicality of the brain disease nursing knowledge base by integrating multi-source data, structured and unstructured information, real-time dynamic mapping of nursing needs and system feedback evaluation.

The invention provides a new energy day-ahead prediction power correction method and system for a load peak period, and relates to the technical field of new energy power generation, and the method comprises the steps: obtaining the new energy initial day-ahead prediction power of a prediction object in the whole period of a prediction day, and the historical electric power before the prediction day; constructing a new energy output deviation power set in the load peak period according to the historical electric power, and calculating a new energy output expected deviation value; the load peak period part in the initial day-ahead predicted power of the new energy is corrected according to the output expectation deviation value of the new energy, and then the day-ahead predicted power of the daily load peak period and the non-load peak period part of the initial day-ahead predicted power of the new energy are combined; and obtaining the new energy target day-ahead prediction power of the whole time period of the prediction day. According to the method, the prediction power before the new energy initial day is corrected through the obtained new energy output expected deviation value, the credibility of the power prediction result of the prediction day is improved, and the risk of a power supply gap is reduced.

The invention discloses a BMS management system and management method of a flow battery. A BMS management system of a flow battery comprises an external information collection module, an internal information collection module, a battery information collection module, an information preprocessing module, a historical SOH information collection unit and an SOH prediction device, and a current SOH value of the battery is generated based on a feature set; the SOH prediction device includes: a prediction unit that predicts an SOH value of a generated battery based on an input feature value; and the correction unit is used for generating influence information of each element in the feature set and the SOH value based on the historical SOH value, and sending the influence information to the prediction unit so as to guide the generation of the SOH value of the battery. According to the technical scheme provided by the invention, the convergence rate is higher during training, and the prediction unit can be continuously corrected according to the actual conditions of different flow batteries during use.

The invention belongs to the field of ship viscous resistance evaluation, and particularly discloses a marine vehicle viscous pressure resistance acquisition method based on considerable ellipsoid hypothesis, which comprises the following steps: establishing a geometric model of a marine vehicle, gridding the geometric model, and determining centroid geometric coordinates, surface element area and surface element external normal vectors of a surface element grid; dividing a flow separation area of the geometric model according to a velocity vector of a remote unit incoming flow and a normal vector outside a surface element; calculating a projection perimeter, a projection area and a shape factor according to the flow separation area, constructing a considerable ellipsoid of the marine vehicle, and determining the geometric dimension of the considerable ellipsoid; the characteristic length of the marine vehicle is determined according to the geometric dimension, the Reynolds number of the marine vehicle is calculated, and the viscosity-pressure resistance coefficient of a considerable ellipsoid is obtained through an interpolation method; and acquiring the viscous pressure resistance of the marine vehicle according to the viscous pressure resistance coefficient, the fluid density, the incoming flow velocity and the wet projection area. According to the invention, the accuracy and efficiency of acquiring the viscous pressure resistance can be improved.

The invention discloses an alpha-arylation reaction enantioselectivity prediction method based on machine learning. The alpha-arylation reaction enantioselectivity prediction method comprises the following steps: acquiring a reaction data set; calculating a physical and chemical descriptor based on a ligand molecular structure, and carrying out feature screening, standardization and dimension reduction processing on the descriptor to construct low-dimensional feature representation of the reaction; constructing and training a supervised machine learning regression model based on low-dimensional feature representation; and inputting a to-be-predicted ligand structure into the trained machine learning model to obtain a prediction result of the enantioselectivity of the alpha-arylation reaction, and applying the prediction result to virtual screening and optimization of candidate ligands. The method can effectively describe the influence of the ligand structure on the reaction enantioselectivity without depending on complex reaction mechanism assumption, realizes prediction of the enantioselectivity of the asymmetric catalytic reaction, and shows good generalization ability. According to the method, molecular structure characteristic engineering and machine learning modeling are combined, so that the accuracy of enantioselectivity prediction and the ligand screening efficiency are improved.

The application discloses a kind of sepsis multiple complications prediction method, system, storage medium and equipment, belong to intelligent medical technical field, method includes the following steps: S1. acquisition sepsis patient multidimensional time series physiological data;S2. medical perception feature vector is constructed to the time series physiological data;S3. sepsis multiple complications prediction model is constructed and introduced physiological logic space-time composite mask matrix;S4. the output multiple complications prediction probability using hierarchical expert hybrid network H-MoE;S5. the model is trained using dynamic weighted loss function based on task uncertainty;S6. risk output and feature contribution degree explanation;The model of the application is introduced by physiological logic mask and H-MoE architecture, realizes the accurate, synchronous prediction of sepsis complex complications, can effectively filter redundant features, and reveal the complex correlation therebetween, provide quantitative decision support for clinician, to realize early intervention and individualized treatment.

The embodiment of the application discloses a grid water depth prediction method and device based on an LSH attention mechanism, and relates to the field of hydrological monitoring. The method disclosed by the application first collects B batches of boundary flow time series data and grid water depth time series data of a target basin output by a two-dimensional IFMS model; then, standardizes the collected data, constructs a first feature tensor and transposes the first feature tensor into a second feature tensor; then, a weighted feature tensor is calculated based on an attention algorithm by using a Reformer model; finally, the weighted feature tensor is decoded by using a pre-trained iTransformer to obtain prediction values of grid water depths at future time points. The application can realize cross-time period prediction, improve prediction efficiency and accuracy, and provide a reliable basis for flood control and disaster reduction.

The invention discloses an electric power company e-commerce purchasing method, relates to the field of e-commerce purchasing, and solves the problem of low accuracy and reliability in a traditional prediction method, and the method comprises the steps: S1, obtaining historical purchasing data and historical inventory data of an electric power company; s2, analyzing historical purchase data and historical inventory data of the electric power company to obtain a fluctuation coefficient corresponding to a selected year; s3, calculating the difference order and the lag order of the selected year corresponding to the monthly purchased commodity amount; s4, predicting the predicted monthly purchase quantity corresponding to the current year through the monthly purchase commodity quantity corresponding to the selected year; and S5, calculating a predicted accurate value corresponding to the predicted monthly purchase quantity, receiving the predicted monthly purchase quantity by the user through the user terminal, and judging the predicted accurate condition through the predicted accurate value, thereby eliminating the influence of abnormal data in the historical data on the prediction result, and ensuring the prediction accuracy and prediction reliability of the purchase quantity corresponding to the purchased goods.

The application discloses a high-density residential form optimization method based on height difference ventilation quantity quantitative guidance, and belongs to the technical field of building ventilation design. A building design parameter set is acquired, and a category identification vector and a numerical vector are respectively extracted through a parameter dimension encoder and a parameter numerical encoder; the ventilation coupling calculation network is input, the coupling strength between each two of the building width, the front row building height and the building spacing is calculated, and a three-dimensional coupling feature tensor is generated; then low-rank compression is carried out along three dimension directions respectively, and the compressed feature tensor is fused through ring product operation; the influence weight coefficients of each parameter are dynamically calculated through the adaptive gating unit, and the comprehensive feature vector is generated by weighted summation; the height difference prediction layer is input, and the minimum north-south building height difference layer meeting the ventilation requirement is output. Through the construction of the end-to-end framework of parameter decoupling, coupling extraction, low-rank compression, adaptive weighting and height difference prediction, the application provides quantitative guidance for high-density residential form optimization.

The invention discloses a big data prediction method based on sub-matrix factorization, which comprises the following steps: constructing a data matrix, selecting sub-matrixes, determining the size and the quantity of the sub-matrixes, constructing a linear equation set, solving a weight matrix, and performing prediction output. According to the method, the sub-matrixes are used for constructing corresponding linear equations, so that a final prediction result is obtained, the weights of the equations are averaged, the obtained model can adapt to the requirement of big data processing, the prediction accuracy and stability can be remarkably improved, and the prediction efficiency is improved. The dimensionality of the original data is reduced by using the linear independent sub-matrix, and the calculation amount is reduced; by selecting a full-rank sub-matrix, the prediction accuracy of the model in different sub-spaces can be ensured; the method is suitable for high-dimensional data processing in various fields, and can be used in combination with other machine learning algorithms to further improve the performance.

The invention discloses a price prediction method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring basic information of a subject matter, wherein the basic information comprises a subject matter name and a target area; a space matrix corresponding to the target area and a fluctuation vector corresponding to the subject matter name are determined, the space matrix comprises a plurality of elements, one element corresponds to one geographic range comprising the target area, the geographic ranges are nested layer by layer, and the fluctuation vector is used for representing the influence degree of objective factors associated with the subject matter on the price; inputting the name of the subject matter and the spatial matrix into a first preset model, and determining a price interval of the subject matter in each geographical range; respectively inputting each price interval and the fluctuation vector into a second preset model, and determining the region price of the subject matter in each geographical range; and calculating the predicted price of the subject matter in the target region according to the prices of all regions. According to the scheme, the price of the subject matter can be quickly and accurately predicted, and reference is provided for scenes such as purchase and transaction.

This invention discloses a wellbore trajectory identification method and device based on marker layer vertical depth, belonging to the field of oil and gas drilling technology. The method integrates actual drilling data and seismic data, formation data, and logging data from the surrounding work area to predict and invert marker layers that may be encountered during drilling, and reconstructs a planar structural map of the marker layer's burial depth. By capturing the marker layers encountered in the actual wellbore trajectory and calculating and comparing them with the vertical depth of the marker layers encountered in the preset wellbore trajectory, the downhole drill bit position is calculated and inferred, thereby achieving the identification and correction of the actual wellbore trajectory. This invention also discloses a wellbore trajectory identification device based on marker layer vertical depth. The technical solution of this invention can be used for wellbore trajectory identification during re-entry operations of old wells in new drilling operations, greatly reducing the drilling difficulty of re-entry and exploration of old wells, providing a reference for drilling process adjustments, and ultimately achieving effective plugging of old wells.

The invention discloses a data prediction method based on real-time adaptive signal decomposition. The method comprises the steps of adaptive signal dynamic adjustment, original data matrix decomposition, linear equation set construction, linear equation set solving, prediction result obtaining and sub-matrix decomposition method combination model. According to the method, the signal parameters are dynamically adjusted according to the input signal change through the parameter adjusting model, the purpose of adjusting the parameters is to enable the signal decomposition algorithm to better adapt to the signal change, the adaptive signal decomposition can adaptively adjust the decomposition parameters according to the dynamic change of the signal under the real-time requirement of signal processing, and the real-time performance of signal processing is improved. Accurate decomposition of the signal is realized; the dimensionality of the original data is reduced by using the linear independent sub-matrix, and the data volume is reduced; the prediction accuracy of the model in different subspaces is ensured by selecting the full-rank sub-matrixes, noise in original information is effectively removed by selecting the full-rank sub-matrixes, and the prediction performance of the model is improved.

This invention discloses a real-time control method for electron cyclotron resonance heating power based on model predictive control, belonging to the field of electron cyclotron resonance heating control technology. The method includes a piecewise modeling step, dividing the operating range of the electron cyclotron resonance heating system into multiple power intervals according to the output power, and establishing a linear system dynamic model describing the dynamic characteristics of the system within each interval; and a real-time control loop step, acquiring the real-time output power and state estimate in each control cycle, determining the target model to be used based on the real-time output power and a hysteresis switching strategy, constructing a quadratic programming problem based on the target model and a preset power reference trajectory, including an objective function and constraints that minimize the power tracking error, solving for the optimal control input and applying it to the system. This invention aims to solve the problem that existing control methods struggle to simultaneously consider adjustment speed and multivariable constraints, achieving rapid and accurate control of electron cyclotron resonance heating power, and improving the safety and stability of system operation.

The invention discloses an underwater image depth-of-field prediction method and system based on the Lambert-Beer law, and the method comprises the steps: carrying out the preprocessing of a to-be-predicted underwater image, and obtaining a noise reduction image; according to the Lambert-Beer law, calculating a first prediction distance and a second prediction distance corresponding to each pixel point in the noise reduction image; wherein the first prediction distance and the second prediction distance are dual-channel light propagation distances corresponding to pixel points; and according to the first prediction distance and the second prediction distance corresponding to each pixel point, respectively calculating to obtain a depth value of each pixel point. According to the invention, the efficiency of underwater image depth information prediction can be improved.

The embodiment of the invention provides a click rate prediction method and device, equipment and a computer program product. The prediction method comprises the following steps: acquiring a click rate prediction request for a preset object; based on the click rate prediction request, determining a neural network model for realizing click rate prediction operation, the neural network model corresponding to model parameters; mapping the model parameters to a topological space, and performing Riemannian optimization on the model parameters to obtain optimized model parameters; updating the neural network model based on the optimized model parameters to obtain an optimized model; and performing a click rate prediction operation based on the optimized model to obtain a click rate prediction result corresponding to the preset object. In the embodiment of the invention, the optimization operation of the neural network model is realized in a Riemannian optimization mode, and then the click rate prediction operation is performed by using the optimized model, so that time and space resources required by the click rate prediction operation can be saved, and the prediction accuracy of the click rate can be ensured.