367results about How to "Quick forecast" patented technology

The invention discloses a pollution research and water quality prediction method based on river segmented water environment capacity analysis. The method comprises the following steps: accurately dividing a drainage basin into water environment capacity accounting units; obtaining a river water body image through satellite remote sensing; identifying and quantifying the positions and the intensities of point source and non-point source pollution sources in the drainage basin by utilizing the water quality parameters of the river obtained by the remote sensing image; establishing a river water quality prediction model; calculating the water environment capacity of each river reach under different surface water environment quality standards; comparing the estimated pollutant discharge amount with the water environment capacity under each water quality standard, and judging whether the pollutant discharge amount exceeds the capacity or not; and predicting future water quality change by using the model. According to the method, remote sensing, GIS and machine learning technologies are utilized, rapid analysis of the river segmented water environment capacity and accurate prediction of the water quality are achieved, dependence on a large amount of field sampling and complex software simulation is reduced, and the working efficiency is greatly improved.

The invention belongs to the field of safety and stability analysis of tailing dams, and discloses a tailing dam stability analysis method based on a GCN-AGRU fusion model and a related device, and the method comprises the steps: obtaining the monitoring data of a multi-source heterogeneous sensor of a tailing dam; preprocessing the monitoring data to obtain multi-source heterogeneous time series data; performing feature analysis on the multi-source heterogeneous time series data to obtain key factors of the stable state of the tailing dam; constructing a heterogeneous hierarchical causal graph by using the key factors; determining a key factor stable state index system according to the heterogeneous hierarchy causal diagram; a multi-source GCN-AGRU fusion space-time prediction model is utilized to obtain space information of the state of the tailing dam; performing time correlation modeling on the multi-source heterogeneous time sequence data to obtain a time sequence of the state of the tailing dam; calculating a predicted value of the multi-source heterogeneous sensor according to the spatial information and the time sequence; and determining the stability state trend of the tailing dam according to the predicted value in combination with a key factor stability state index system. The intelligent sensing of the situation of the tailing dam can be realized.

The invention provides a cement decomposing furnace temperature prediction control method based on a KAN enhanced neural network, and belongs to the technical field of cement calcination. A CNN-LSTM-KAN network is established as an f-CaO and Nox prediction model, and a state parameter optimization objective function is established in connection with actual production requirements; and solving an optimal solution and an optimal solution interval of the state parameters of the decomposition furnace through a multi-target particle swarm optimization algorithm. A KAN-Enhanced-Informer network is established to predict the temperature of the decomposition furnace, a target function is established by combining a state optimization result with the more accurate prediction result, the less coal consumption and the reasonable alternative fuel quantity as targets, the optimal solutions of the coal feeding quantity and the alternative fuel quantity are obtained, and the optimal solutions are applied to a control model to ensure that the temperature of the decomposition furnace reaches the optimal set value. The problems that in the actual cement production process, energy consumption is too high, the yield is not ideal, and much sewage is discharged are solved.

The invention provides a method for training an end-to-end automatic driving strategy, and the method comprises the following steps: S1, data collection and preprocessing, S2, defining a loss function and an optimizer, and constructing a model, S3, data division, S4, model training, S5, evaluation on a verification set, S6, model adjustment and optimization, and S7, test set evaluation. S8, weight pruning and evaluation: calculating and evaluating the importance of each weight by adopting a gradient norm calculation mode, evaluating the model after weight pruning by using the test set in the step S7 according to an importance evaluation result, starting from a relatively small proportion, gradually increasing and removing unimportant weights, and if the performance of the model does not reach the standard at the moment, evaluating the model after weight pruning by using the test set in the step S7. According to the method, through weight pruning, the applicability of a model system is wider, and the training efficiency of the system is higher.

The invention discloses a physically-guided critical region fluid nonlinear wave motion prediction method, system and device and a medium. The method comprises the steps that a critical region fluid computational domain and a nonlinear time-varying equation are established; constructing a sample data set; constructing a physical information neural network model; constructing a loss function including nonlinear time-varying equation residual error loss, initial point error loss and boundary point error loss; training the physical information neural network model to obtain a critical region fluid free water surface displacement wave envelope prediction model; weighting internal matching point samples in each iteration round by adopting a gradient-free nonlinear attention mask method based on residual accumulation in the training process; and inputting the time-space coordinates of the to-be-predicted point of the critical region fluid into the prediction model to obtain the free water surface displacement wave envelope of the to-be-predicted point fluid. According to the method, error accumulation is effectively reduced, the model convergence speed is increased, and the prediction precision of the nonlinear wave motion of the fluid in the critical region is improved.

The invention provides a machine learning optimization method and system for a low-temperature high-activity CO2 methanation catalyst, and relates to the technical field of machine learning. The optimization method comprises the following steps: collecting CO2 methanation reaction data, building a machine learning model for hyper-parameter optimization, selecting an optimal model, carrying out interpretable analysis on the optimal model, and screening the catalyst by combining the optimal model with a multi-objective optimization algorithm, so that the machine learning optimization of the CO2 methanation catalyst is realized, and the machine learning can analyze a large amount of data so as to realize the optimization of the CO2 methanation catalyst. The catalyst with high activity and selectivity is rapidly predicted and screened out, and the research and development efficiency is greatly improved; machine learning can reduce experiment times, optimize experiment design, significantly reduce research and development cost, and improve research and development success rate; machine learning can identify and utilize complex chemical and physical data, a catalyst with better performance is designed in a high-precision manner, and the conversion rate and selectivity of converting CO2 into CH4 are improved.

The invention discloses an amorphous forming ability prediction method based on Transform and table data conversion. The amorphous forming ability prediction method comprises the following steps: acquiring basic attributes and amorphous forming ability attributes of amorphous alloy composition elements in a metal alloy sample; the basic attributes refer to element properties simulated based on an experiment or a density functional theory and comprise atomic basic properties, chemical properties and physical properties, and the amorphous forming ability attributes comprise a critical casting diameter Dmax; generating an amorphous alloy data set; wherein the data set comprises an independent variable and a response variable, the independent variable comprises element composition and mixing enthalpy Hmix, and the response variable is a critical casting diameter Dmax; defining a relation model, wherein the relation model is an integrated model formed by three types of models; and training the integrated model by using an amorphous alloy data set, and constructing an amorphous forming ability evaluation model. According to the technical scheme, various relation models for predicting the critical casting diameter can be trained, and the various relation models are combined to construct the optimal amorphous forming ability evaluation model.

The invention provides a position prediction method and device based on an ultra wide band (UWB) digital key, and a position prediction system applied to the UWB digital key, the position prediction system comprises a UWB tag and a plurality of UWB anchor points arranged on a vehicle, and the position prediction method comprises the following steps: determining whether UWB positioning information of the UWB tag is invalid; when the UWB positioning information is invalid, acquiring a velocity vector and a historical track set of the UWB tag and orientation information of the vehicle; and inputting the velocity vector, the historical trajectory set and the orientation information into a position prediction calculation model to calculate a predicted position of the UWB tag. The technical problem that when an existing automobile digital key is blocked by an obstacle and the digital key cannot be connected to an automobile UWB anchor point, the function of the digital key is in a failure or invalid state, and the user experience is affected is solved.

The invention provides a high-performance material visual screening system based on machine learning and an application method thereof. The system comprises a feature extraction module, a machine learning model training module, a prediction module and a feedback optimization module. The data preprocessing module is used for performing primary processing on the original data, wherein the primary processing comprises removal of invalid data and abnormal values; the feature extraction module is used for converting a time domain signal into a frequency domain signal by adopting Fourier transform, extracting a frequency domain feature, and performing data dimension reduction by adopting principal component analysis (PCA); the machine learning model training module is used for ensuring the generalization ability of the model by adopting a cross validation technology in the training process, and accelerating the convergence process of the model through a gradient descent method; the prediction module is used for predicting the material performance of the high-performance material to obtain a prediction result; and the feedback optimization module is used for automatically optimizing model parameters by adopting Bayesian optimization, analyzing a prediction result of the model and dynamically adjusting the model parameters so as to improve the prediction precision. Through the scheme of the embodiment of the invention, the law can be learned from a large amount of experimental data by a machine learning technology, and the material performance can be quickly and accurately predicted.

The invention belongs to the technical field of automobile structure design, and particularly relates to a large-model-based automobile body structure generative design method and system, and the method comprises the steps: building a multi-source database containing automobile body structure pictures, design variables and mechanical properties; training a vehicle body structure performance evaluation model and generating an evaluation model according to the database; on the basis of the generated evaluation model, the general text graph large model is finely adjusted, and a vehicle body structure generative design large model is obtained; generating a structure picture by using the obtained large model, and establishing a structure size evaluation model in combination with a database; inputting design requirements and constraint conditions, generating a vehicle body structure, evaluating the performance and size of the vehicle body structure, and outputting a final result. According to the method, the problems of low efficiency and strong dependence on experience of engineers in an existing automobile concept design stage are solved, the efficiency and creativity of automobile body structure design are improved, the design period is shortened, manual intervention is reduced, and automation and intelligence of automobile body structure design are promoted.

The invention provides a focused ultrasound blood brain barrier opening effect prediction method and system, a medium and a terminal, and the method comprises the steps: carrying out the preprocessing of all collected acoustic signal data, and constructing a training data set based on the blood brain barrier opening effect data; training a deep learning model by using the training data set to construct a blood brain barrier prediction model; and on the basis of the blood brain barrier prediction model, predicting the permeability of the blood brain barrier according to the obtained real-time acoustic signal data of the ultrasonic contrast agent in the process of regulating and controlling the blood brain barrier so as to output a blood brain barrier opening effect prediction result. The acoustic signals are analyzed based on the artificial intelligence algorithm, a magnetic resonance contrast agent does not need to be injected, the long-time scanning process is avoided, safe and rapid prediction of the blood brain barrier permeability is achieved, and the test cost is reduced. Compared with a dye tracing method, the blood brain barrier permeability prediction method based on artificial intelligence does not need to perform any operation on a target test, is convenient and rapid, and has a living body detection capability.

The invention discloses an artificial intelligence chemical inverse synthesis route planning system, and belongs to the field of intelligent planning systems, the method comprises the following steps: a user interface, a Monte Carlo tree search and route search module, a reaction condition prediction and route optimization module, and an output and feedback module; wherein the user interface is used for providing a customized interface for a user, and the user performs molecule input based on the customized interface to determine a target molecule; the Monte Carlo tree search and path search module is used for constructing an inverse synthesis path of a target molecule based on a Monte Carlo tree search algorithm; the reaction condition prediction and path optimization module is used for predicting the reaction condition of each candidate reaction and optimizing the inverse synthesis path of the generated target molecule to obtain an optimal inverse synthesis path; and the output and feedback module is used for displaying the optimal inverse synthesis path to a user and receiving feedback of the user.

According to the multi-aircraft track credibility prediction method and device based on machine learning, track credibility prediction can be rapidly carried out on future surrounding aircrafts, and powerful support is provided for various tasks such as arrival and departure sorting, conflict detection, flight flow management and arrival time estimation in the field of air traffic flow management. The method comprises the following steps: (1) carrying out data preprocessing, constructing a multi-aircraft track data set, and converting data of a plurality of complex scenes into a fixed-length track vector sequence; (2) a multi-aircraft track credible prediction model is constructed, a framework is constructed based on spatial-temporal correlation of multi-aircraft tracks, the framework comprises a spatial feature extraction module and a long short-term memory (LSTM) network module, the spatial feature extraction module enables adjacent agents to interact in a modeling space, and the LSTM module captures a time dependency relationship of a 4D track of a single aircraft; and (3) training by using historical data, and carrying out credibility prediction on the multi-aircraft track by using the multi-aircraft track credibility prediction model.

The invention discloses a pea seed quality detection method and device based on an infrared spectrum, and relates to the technical field of seed screening. Comprising the following steps: collecting infrared spectrums of sample pea seeds with known quality parameters, carrying out denoising and enhancement pretreatment, and establishing a neural network prediction model in combination with the quality parameters (water content, crude fiber, fat, starch and protein); after same preprocessing is carried out on the infrared spectrum of a seed to be detected, the trained model is input, and quality parameters of the seed to be detected are predicted; calculating a nutrition richness factor and a water swelling potential factor based on the predicted value, and combining with a surface image to extract a color channel value to calculate a color uniformity coefficient; the quality index is calculated by integrating information such as color uniformity, circularity and surface roughness, and is compared with the preset judgment threshold to generate the quality judgment result, so that the quality judgment result of the pea seeds can be quickly output, and the requirements of modern agriculture on high-throughput detection and intelligent analysis are met.

The invention provides a deep sea convergence area position determination method and device, equipment and a medium, and the method comprises the steps: judging whether a target sea area meets the necessary conditions of a convergence area or not according to the hydrological environment parameters of the target sea area; if yes, establishing a sound velocity profile of the target sea area in each vertical layer and a sound ray trajectory equation of the target sea area; in a horizontal uniform hydrological environment, based on the hydrological environment parameters and the sound ray trajectory equation, constructing an expression of a sound ray span of a zero-degree emergence angle; and based on the sound velocity profile of the target sea area in each vertical layer and the expression of the sound ray span of the zero-degree emergence angle, obtaining the sound ray span of the zero-degree emergence angle, and determining the sound ray span of the zero-degree emergence angle as the convergence area position of the target sea area. In this way, the zero-degree-angle emergent sound ray span is used as the initial position of the convergence area, and the purposes of reducing the calculated amount and rapidly forecasting the position of the sound field convergence area can be achieved.

The invention discloses a coating impact damage prediction method based on a material genome. The method comprises the following steps: (1) obtaining original data through an impact experiment and related tests; (2) taking the residual stress, the maximum shear stress, the shear stress depth and the porosity of the amorphous coatings with different thicknesses as material gene characteristic data, taking the crack length caused by impact damage as performance data, constructing a corresponding relation database, and obtaining an original data set of a machine learning model; (3) establishing eight machine learning models, and dividing a training set and a test set through 5-fold cross validation; (4) performing grid search on the training set to obtain an optimal hyper-parameter of the model, and further obtaining an optimal machine learning model; (5) evaluating the data in the test set to obtain a machine learning model with the highest accuracy; and (6) predicting impact damage according to the machine learning model with the highest accuracy. According to the method, impact damage prediction is realized based on machine learning, and the precision of performance prediction in coating service performance research is improved.

The invention discloses an underwater vehicle hydrodynamic noise rapid forecasting method and system based on a BP neural network optimized by a genetic algorithm, belongs to the field of ship underwater radiation noise forecasting, and designs a BP neural network noise forecasting model optimized by the genetic algorithm. Key variables including shell thickness, molded line parameters, surface roughness, material attributes, sailing speed, fluid medium density, viscosity coefficient and underwater vehicle diving depth are selected as characteristic conditions. The initial weight and threshold value of the optimal BP network are obtained through the global optimization capacity of the genetic algorithm, the optimal initial weight and threshold value obtained through the optimization algorithm serve as the initial weight and threshold value of the BP neural network, and training is carried out to avoid falling into the local minimum value. And performing iterative training on the optimization model by using the training set to obtain a high-precision noise forecasting model, thereby realizing forecasting of the hydrodynamic noise of the underwater vehicle. The method is suitable for forecasting the hydrodynamic noise of the underwater vehicle, and has the advantages of wide applicability, high accuracy and the like.

The invention discloses a trapezoidal sail multi-target size optimization method based on an NSGA-II algorithm, and relates to the technical field of unmanned sailboats. According to the method, the aspect ratio, the taper ratio and the camber ratio are selected as design parameters of the trapezoidal sail, after multiple sets of design parameters are sampled and obtained within the value range of the design parameters based on a Latin hypercube sampling method, a trapezoidal sail model is constructed by utilizing fluid simulation analysis software for each set of design parameters, and the trapezoidal sail model is obtained. Performance evaluation parameters corresponding to each group of design parameters are obtained through simulation, a database is constructed, the database is utilized to train a Kriging agent model for replacement, multiple groups of trapezoidal sail design parameters to be preferentially selected are selected, and multi-target size optimization is performed on the trapezoidal sail design parameters based on an NSGA-II algorithm and the trained Kriging agent model. And determining optimal trapezoidal sail design parameters and verifying the accuracy of the optimal trapezoidal sail design parameters. According to the method, trapezoidal sail design parameters and performance evaluation parameters are combined, and a basis is provided for optimization of a sail design scheme by using an NSGA-II method and a Kriging agent model.

The invention discloses a typhoon storm surge beach overtopping intelligent forecasting method based on a Vision Transform neural network. The method comprises the following steps: firstly, acquiring continuous storm surge water level data and wind field data at a forecasting moment; preprocessing the acquired continuous storm surge water level data and the wind field data at the forecast moment, performing extreme value normalization on the preprocessed data, scaling the data to 0-1, and dividing a training set and a test set; based on the training set, the normalized numerical value is input into a Vision transfer neural network for training, and a typhoon storm surge beach overtopping intelligent forecasting model is obtained; and inputting the test set into the typhoon storm surge beach overtopping intelligent forecasting model to adjust and optimize the typhoon storm surge beach overtopping to obtain a final typhoon storm surge beach overtopping intelligent forecasting model, and performing typhoon storm surge beach overtopping forecasting by using the typhoon storm surge beach overtopping intelligent forecasting model. According to the method, the two-dimensional distribution and the water increasing height of the typhoon storm surge overtopping beach can be accurately and quickly predicted.

The invention discloses an energy regulation and control method and system of a virtual power plant based on a prediction model, and relates to the field of virtual power plant operation control, and the method comprises the following steps: collecting energy data in the virtual power plant; integrating the wind power data and the load data, and performing feature extraction on the integrated data; based on the predicted power generation data of the distributed energy and the predicted demand data of the load side resources, constructing a preliminary energy dynamic balance prediction model; performing day-ahead load demand and distributed energy prediction by using the preliminary energy dynamic balance prediction model; according to the preliminary scheduling strategy, using a niche grey wolf optimization algorithm to adjust the operation state of each distributed energy and load side resource in real time; dynamically adjusting and optimizing the energy dynamic balance prediction model; and carrying out load demand and distributed energy prediction again by using the optimized energy dynamic balance prediction model. According to the method, the change of the power system can be responded in advance, and more efficient energy scheduling is realized.

The invention discloses a method and system for predicting electrical characteristics of a CMOS device, and belongs to the technical field of computer aided design and simulation, and the method comprises the steps: obtaining technological parameters and electrical parameters based on the production line data of the CMOS device; based on the data dimensions of the process parameters and the electrical parameters, the number of convolution layers of the model is adjusted, an activation function is set between each layer of convolution, and by setting a full connection layer and constructing a prediction model, the process parameters are used as model input, the electrical parameters are used as model output, and the electrical characteristics of the CMOS device are predicted. According to the method, the problem of weight difference caused by the sequence of the process parameters during convolution is avoided, the problem of overfitting is solved, and the electrical parameters of the CMOS device are predicted more accurately and more quickly based on the process parameters.

The invention provides an analysis and management system for telecontrol information data, and relates to the technical field of electric power and data analysis, and the system comprises a configuration inspection module which carries out the differential analysis of a master station telecontrol information point table and a substation telecontrol information point table, and generates a telecontrol configuration file; the data acquisition module acquires test telecontrol information data and a telecontrol configuration file, and performs communication link matching to obtain a communication link check code; the data preprocessing module preprocesses the test telecontrol information data to obtain test telecontrol information preprocessing data; the data verification and analysis module performs exception verification and analysis on the test telecontrol information preprocessing data to obtain an integrity index; the abnormal data prediction module constructs a telecontrol information data abnormal prediction model and performs abnormal prediction on test telecontrol information data collected in real time; and the real-time management module manages the test telecontrol information data in real time according to the prediction result. According to the scheme, the operation efficiency of the system and the accuracy of telecontrol information data analysis can be improved.

The invention discloses a drainage basin flood control scheduling scheme optimization method based on a neural network. The method specifically comprises the following steps: S1, constructing a hydrological and hydrodynamic coupling model of a drainage basin; s2, performing simulation calculation by using the hydrological and hydrodynamic coupling model constructed in the step S1, and preparing training data required by a neural network model; s3, constructing a neural network model containing scheduling information, and performing neural network model training and verification by using the training data in the step S2; s4, based on the neural network model trained in the step S3, according to a preset target water level, selecting an optimal flood control scheduling scheme by using an optimization algorithm; according to the method, the neural network technology is introduced, the computational efficiency bottleneck of a traditional hydrological hydrodynamic model in flood control scheduling is overcome, the scheduling scheme optimization speed and accuracy are remarkably improved, and a new solution is provided for coping with extreme flood events.

The invention belongs to the technical field of water environment monitoring and simulation analysis, and aims to provide a water pollutant flux prediction method and system, electronic equipment and a product. According to the invention, the final water pollutant flux prediction model is obtained based on historical data, and the real-time water monitoring data obtained by water quality monitoring is used as the input data of the final water pollutant flux prediction model, so that the real-time water pollutant flux prediction value is obtained. And finally, performing correction processing on the real-time water body pollutant flux prediction value to obtain a corrected real-time water body pollutant flux prediction value of the target water body area, thereby realizing pollutant flux prediction of the target water body area. According to the method, the flux of the water pollutants can be predicted, and the prediction precision is high.

The invention provides a building group seismic loss probability assessment method based on seismic oscillation intensity random field simulation, and belongs to the technical field of seismic loss assessment. The method provided by the invention comprises the following steps: firstly simulating seismic oscillation of a region by adopting a seismic oscillation intensity random field simulation technology to obtain seismic oscillation intensity spatial distribution of different sites, and then estimating loss probability distribution of each building by utilizing a building anti-seismic performance model and a building vulnerability database. According to the method, the occurrence probability of extreme earthquake loss of a large-scale building group can be evaluated more quickly and accurately, and important support is provided for earthquake risk evaluation and post-disaster emergency decision, so that the disaster resistance and recovery capability of a city are enhanced, and the overall toughness of the city is improved.

A long-term intelligent prediction method for a wind-induced response of a super high-rise building structure during a typhoon transit period is characterized by comprising the following steps: calculating a structure wind-induced response corresponding to a target super high-rise building; a Levenberg-Marquardt algorithm is adopted to carry out training on the cascade forward BP neural network model; predicting the incoming flow wind field information of the target super high-rise building by using an AI-driven Transform weather prediction model and a linear typhoon model; and inputting the incoming flow wind field prediction information and the corresponding structure modal parameters into a trained cascade forward BP neural network model, and rapidly predicting the structure wind vibration response of the super high-rise building at different positions under different time conditions in the future. According to the method, the long-term prediction of the wind vibration response of the super high-rise building structure during the typhoon transit period can be quickly, flexibly and reliably realized, and a reference is provided for the prediction and early warning of the vibration state of the super high-rise building structure during the typhoon, so that the related emergency response capability of a city is improved, and the transition from passive emergency to decision forward is realized.

The invention provides a frequency conversion series resonance withstand voltage detection device and method, and the device comprises a comprehensive control box, an excitation transformer, an electric reactor group, a voltage divider, and a compensation capacitor group, and the output end of the comprehensive control box is electrically connected with the low-voltage side of the excitation transformer. The high-voltage side of the excitation transformer is electrically connected with the reactor group, the voltage divider and the compensation capacitor group in sequence, the compensation capacitor group is electrically connected with a to-be-detected device, the comprehensive control box is directly and electrically connected with the voltage divider, and the comprehensive control box is electrically connected with the low-voltage side of the excitation transformer through the shunt resistor. And the comprehensive control box is used for collecting the sampling voltage at the voltage divider and the sampling current at the low-voltage side of the excitation transformer, is also used for predicting a resonant frequency interval where a resonant point is located by combining the sampling voltage and the sampling current, and is also used for outputting variable-frequency alternating current to the low-voltage side of the excitation transformer according to the resonant frequency interval. The method has the effect of improving the accuracy of a voltage withstanding test result.

The invention discloses a rusted high-strength bolt connection pretightening force intelligent self-optimization prediction method and system, and belongs to the technical field of bolt pretightening force detection.The method specifically comprises the steps that a finite element database of bolt connection pretightening force loss is established, and a machine learning prediction model based on a neural network is trained; when the recognition accuracy of the machine learning prediction model reaches a preset value, pre-tightening force loss of the target bolt is monitored in real time, and real-time geometric feature information of connection of the target bolt is obtained; when the similarity between the real-time geometric feature information of the target bolt connection and a finite element model in the finite element database is greater than or equal to a preset value, calling the pre-tightening force loss of the existing finite element database; when the similarity is lower than a preset value, the trained machine learning prediction model is utilized to output actual bolt connection pre-tightening force loss in real time; the bolt looseness detection device realizes automatic and marker-free bolt looseness detection, is high in bolt image acquisition angle and distance tolerance, and can meet bolt looseness detection at different positions.

The invention relates to the technical field of composite material performance prediction, and discloses a composite material flame retardant property and thermal stability prediction method based on thermogravimetric analysis and infrared spectrum data fusion. The invention aims to solve the problem that it is difficult to effectively fuse multi-modal data to realize quantitative prediction in the prior art. The method comprises the following steps: firstly, collecting thermogravimetric and infrared spectrum combined data, calculating gas transmission lag time, and carrying out forward correction on infrared spectrum data to realize time sequence alignment; then solid-phase thermogravimetric feature vectors and gas-phase infrared feature vectors are extracted, and a fusion feature matrix is generated; extracting solid-phase and gas-phase characteristics respectively by using a dual-channel LSTM network, and performing weighted fusion based on a self-attention mechanism; and finally, synchronously outputting a flame retardant index predicted value and a flame retardant grade classification result through numerical mapping and classification judgment. According to the method, multi-dimensional feature mining of the whole pyrolysis process of the material is realized, and a rapid and accurate digital evaluation means is provided for screening of the flame-retardant material.

The invention provides an automatic focusing method and device of an image, a storage medium and electronic equipment, and the method comprises the steps: cutting out a knife edge image from a target image, determining the feature data of the target image according to the knife edge image, the feature data at least comprising one of the following: an ESF radius, a PSF radius and an extreme radius of the target image, the cutting edge image is an image comprising a clear cutting edge; adopting multiple groups of training data to train based on a FousNet network to obtain an over-focus and under-focus classification neural network; inputting the target image into the over-focus and under-focus classification neural network to obtain a target out-of-focus direction of the target image; and determining a defocusing amount of the target image according to the feature data of the target image and the target defocusing direction, and performing automatic focusing on the target image according to the defocusing amount. According to the method, three characteristic indexes in linear relation with the defocusing amount are adopted, the defocusing amount is reversely deduced through a linear model or a high-order model, and the quasi-focusing position of the chip is rapidly predicted.