102 results about "Multiple linear regression model" patented technology

The invention discloses a disastrous big wave occurrence probability and wave height combined prediction method, device and equipment, a medium and a product, and relates to the technical field of meteorological ocean forecasting. The method comprises the following steps: acquiring an effective wave height measured value and environmental factor data of a to-be-predicted region in a historical time period; key environment factors are screened through spatial correlation analysis and linear fitting verification, a first training data set and a second training data set are constructed, and a multiple linear regression model and a logic regression model are trained respectively to obtain a wave height prediction model and a big wave occurrence probability prediction model; and inputting the key environmental factors of the current time period of the to-be-predicted region into the model to obtain the predicted occurrence probability of the disastrous waves and the predicted value of the significant wave height in the predicted time period. According to the method, the key environmental factors are screened and the two models are used for respective prediction, the influence of the various environmental factors on the big waves is considered, the occurrence probability and the wave height of the disastrous big waves can be predicted more accurately, and the ocean safety is effectively guaranteed.

The invention relates to the field of coal mine safety engineering, and provides a coal mine gas control extraction effect dynamic evaluation and closed-loop control method, which comprises the following steps: S1, based on a basic threshold value in an extraction design scheme, periodically and dynamically calibrating the threshold value through a multiple linear regression model, and generating a dynamic threshold value matrix of a core index; s2, collecting multi-dimensional data to construct a fusion data matrix; s3, a dynamic multi-scale convolutional neural network model is constructed and trained, and the model comprises a variable-length filter generator which is used for adaptively generating a variable-length filter according to the input data so as to extract multi-scale time sequence features; outputting a standard state judgment result of the current extraction effect and a core index prediction value of a future preset time period in parallel; and S4, starting a three-level linkage feedback management and control flow coordinated by the AI model and the expert rule base according to the standard reaching state judgment result and the predicted value output in the step S3. Accurate evaluation and short-term prediction of core indexes such as the extraction flow, the gas concentration and the attenuation coefficient are achieved.

The invention discloses a power load probability prediction method and system based on a neural network quantile regression model and multiple linear regression, and belongs to the technical field of power system load prediction. The method comprises the following steps: obtaining standardized data by using a longitudinal data analysis method; identifying key influence factors of the power load in the standardized data based on a Pearson correlation analysis method, and constructing a factor analysis model to quantify influence weights of the key influence factors on the power load; constructing a neural network quantile regression model based on seasonal trend decomposition to fit the key influence factors with different influence weights to obtain a quantile prediction result; based on the quantile prediction result, estimating a continuous probability distribution curve of a load common factor by adopting a non-parametric kernel density technology to obtain an interval prediction result; and constructing a multiple linear regression model to predict the load scale change, and adjusting the interval prediction result. According to the invention, the precision and calculation efficiency of load prediction are effectively improved.

This application discloses an elevator energy-saving control method and system, relating to the field of elevator control technology, aiming to solve the problem that the energy consumption of elevators remains high in practical applications under existing technologies. The method is used in a carrier frequency adjustment device and includes the following steps: acquiring elevator operating status information; acquiring a set of influencing factors based on the elevator operating status information; performing trend analysis based on the set of influencing factors to obtain the influence coefficient value corresponding to each influencing factor; inputting the influence coefficient values ​​corresponding to each influencing factor into a trained target multiple linear regression model, so that the target multiple linear regression model outputs a target carrier frequency value; adjusting the current carrier frequency value of the elevator to the target carrier frequency value to reduce the energy consumption of the elevator during operation. Based on the method described in this application, the energy consumption of the elevator under different operating states can be reduced, and the smooth operation of the elevator can be ensured.

The application discloses a dandelion medicinal material effective component rapid determination method based on color chroma values, and relates to the technical field of traditional Chinese medicinal material quality detection. * , a * , b * Chroma values of dandelion medicinal material powder are collected, and the contents of chicoric acid, monocaffeoyl tartaric acid and chlorogenic acid are determined by using high performance liquid chromatography as true values; subsequently, a neural network prediction model with chroma values as input and effective component contents as output is constructed and trained, preferably a single hidden layer structure; finally, the model is used for fast and nondestructive content prediction of unknown samples; compared with a traditional multiple linear regression model, the neural network model used in the application has significantly higher prediction accuracy, and the test set R² is all more than 0.86; the application realizes fast, nondestructive, low-cost and high-precision detection of dandelion medicinal material effective components, and is suitable for multiple scenes such as raw material quality control, market supervision, optimal harvest period judgment and fast pricing grading.

The invention relates to a novel prediction method for the residual life of a power transmission tower, and belongs to the field of power system equipment monitoring and maintenance. The method comprises the following steps: firstly, arranging a permanent magnet on the surface of a power transmission tower body to magnetize the power transmission tower body by utilizing a magnetic flux leakage signal detection technology, collecting and normalizing magnetic induction intensity and extracting magnetic signal characteristic quantity by combining a Hall sensor, and establishing a function relationship with angle steel corrosion depth and width, so as to determine a corroded quantity; then, based on environmental factors, namely temperature, humidity, sulfur dioxide concentration and chloride ion concentration, predicting the corrosion rate through a multiple linear regression model; analyzing the relationship between the corrosion degree and the residual bearing capacity to obtain the corrosion degree of the power transmission tower under the ultimate bearing capacity; and finally, calculating the residual life of the iron tower by calculating the amount to be corroded and combining the corrosion rate. The residual life of the power transmission tower can be accurately predicted, and a reliable basis is provided for regular maintenance and reinforcement of the tower.

The invention relates to the technical field of down-the-hole drill ball tooth drill bit wear measurement, in particular to a down-the-hole drill ball tooth drill bit wear value real-time measurement method and system, and the method comprises the steps: carrying out the optimization and registration of an initial three-dimensional model, and obtaining an overall wear quantized value of a down-the-hole drill ball tooth drill bit; analyzing a mapping relation between the while-drilling parameters and the overall wear quantized value, and constructing a database between the while-drilling parameters and the drill bit wear value; deriving a dimensionless variable according to a dimensional analysis method and a database, and performing combined screening on the dimensionless variable, a second-order polynomial and a cross term to determine a dimensionless variable combination influenced by drill bit wear; and substituting the dimensionless variable combination into the multiple linear regression model to establish a mathematical relationship model between the drill bit wear and the while-drilling parameters, thereby realizing the real-time measurement of the wear value of the ball tooth drill bit of the down-the-hole drill. According to the method, non-dimensional variables of the while-drilling parameters are obtained based on dimensional analysis, the model is built in combination with multiple linear regression, the wear value of the ball tooth drill bit of the down-the-hole drill is measured in real time, and construction efficiency and safety are improved.

The invention discloses a Dst index monitoring method, system and equipment based on a Zhang-Heng-1 satellite, a medium and a program product. The method comprises the following steps: acquiring magnetic field data acquired by the Zhang-Heng-1 satellite; performing data preprocessing on the magnetic field data to obtain satellite-borne observation data; carrying out background magnetic field deduction processing on the satellite-borne observation data by utilizing a global geomagnetic field model to obtain a satellite-borne magnetic field residual error; and finally, the satellite-borne magnetic field residual error is input into a ground-based Dst index mapping model, a corrected satellite-borne Dst index is obtained, and the ground-based Dst index mapping model is a multiple linear regression model obtained through segmented fitting according to different magnetic storm grades based on historical Dst index data and solar radiation indexes. According to the method, high-timeliness and high-precision monitoring of the global geomagnetic storm index is realized through high-precision mapping from the satellite-borne original observation residual error to the foundation standard Dst index.

The application discloses a tobacco leaf baking management method and system, and the method comprises the following steps: pre-entering and storing baking stage judgment threshold values corresponding to different regions, varieties, parts and maturity tobacco leaves, and completing threshold value library initialization; in the baking process, multi-source data fusion judgment is realized by synchronously collecting multi-dimensional tobacco leaf state data in combination with a multiple linear regression model, and a judgment result of the current baking stage is output. The application effectively solves the problems of low accuracy and high misjudgment rate of the existing tobacco leaf baking stage judgment scheme, and realizes the effects of improving the judgment accuracy and reducing the baking quality loss.

The invention relates to the field of biological manufacturing, and particularly provides a hyaluronic acid fermentation process online monitoring method based on near infrared spectrum and multi-model fusion. The method comprises the following steps: constructing a near infrared spectrum online acquisition system in the hyaluronic acid fermentation process; collecting samples at different fermentation time points in different batches, and measuring the molecular weight; dividing a training set and a test set according to a certain proportion; preprocessing the original near infrared spectrum; constructing a plurality of basic models for prediction; optimizing hyper-parameters of the SVM model, the ANN model, the GBDT model and the RF model by using an improved umbrella-sky optimization algorithm which introduces a self-adaptive inertia weight strategy and an elite reverse learning strategy, and optimizing potential variable numbers of the PLS model by using a cross validation method; and a plurality of basic models are fused, and a multiple linear regression model is trained to generate final prediction. The method realizes real-time and accurate monitoring of the key parameter molecular weight in the hyaluronic acid fermentation process, and provides powerful technical support for optimizing the fermentation process and improving the yield and quality of hyaluronic acid.

The application discloses a kind of painting workshop environment quality prediction method and prediction model establishment method, the present application selects volatile organic compound concentration as painting workshop environment quality reference index, selects volatile organic compound concentration as dependent variable, spraying time, spraying distance, air flow rate as independent variable, establishes the multiple linear regression model of painting workshop environment quality, establishes neural network model, the fitting value of multiple regression model is used as the input of neural network model together with independent variable, constructs multiple linear regression-neural network combination model, according to painting working condition, the air quality in workshop is predicted using combination model, when volatile organic compound concentration in painting workshop reaches certain threshold value, it is necessary to carry out alarm and forced ventilation.The present application can accurately and quickly calculate volatile organic compound concentration in workshop by painting workshop environment quality prediction model, so as to adjust ventilation volume in time, protect construction personnel, save energy consumption.

A comprehensive prediction method for drinking water taste based on sensory evaluation and physicochemical parameters is disclosed. The method includes the following steps: S3, sensory evaluation data collection, where the sensory evaluation data includes the following dimensions: odor and taste; and S4, physicochemical parameter detection, including at least the following indicators: pH value, total hardness, and free chlorine concentration. S5, data standardization and preprocessing; S6, construction of a comprehensive taste prediction model: based on the standardized sensory evaluation data and the processed physicochemical parameter data, a multivariate fusion data model is constructed; and S7, inputting each variable according to the pattern of a multiple linear regression model to obtain the final calculation result under the control of multiple variables in the model. This invention's prediction method features a unified evaluation scale, strong comparability of results, significantly improved objectivity of the evaluation, quantitatively correlates taste and physicochemical indicators, and ensures that different evaluators maintain consistency in their evaluation starting point and scale, thereby significantly reducing random fluctuations in the model input data and improving the stability of the prediction model.

The invention discloses a man-machine work efficiency quantitative decision-making method for a sandstone compound stratum of a shield tunneling machine. The method comprises the following steps: S1, defining a man-machine comprehensive work efficiency model based on a tunneling rate, a cutter loss rate and unit energy consumption; s2, screening independent variables; s3, establishing a multiple linear regression model by taking a man-machine comprehensive work efficiency model calculation value as a dependent variable and combining the screened independent variable; s4, performing variable standardization on the model and converting the model into a standard regression equation; s5, calculating the influence weight of each standardized variable; and S6, sorting the influence weights, and selecting a decision direction according to a result. According to the method, multi-dimensional parameters are integrated, the influence of the multi-dimensional parameters on the tunneling efficiency is quantified, a dynamic response scheme can be generated, the tunneling speed is increased, loss is reduced, and the continuous operation period of the shield tunneling machine is prolonged.

The application provides a wind turbine power curve prediction method and device, and relates to the technical field of wind turbines.The actual power, generator speed and environmental data of the wind turbine are collected first to provide a basis for subsequent analysis.Then, the power generation data of the previous moment is taken as input, the actual power of the next moment is taken as a label, a multiple linear regression model is trained, and the influence of the current moment is focused on, while the cumulative influence of the past is easily ignored.Then, the power generation data of the previous multiple moments is taken as input, the actual power of the next moment is taken as a label, a long short-term memory network model is trained, and the influence of the time series data is focused on, which is suitable for dynamic changing data.The generator speed, wind speed, temperature and humidity data are analyzed to obtain a fluctuation adjustment coefficient, the actual power predicted by the two models at the current moment is corrected by using the fluctuation adjustment coefficient, and the corrected actual power is applied to the wind turbine power curve.

The application discloses a carbon material graphitization degree detection method and a carbon material specific capacity detection method. The carbon material graphitization degree detection method comprises the following steps: constructing a multiple linear regression model, the multiple linear regression model is the correlation between the graphitization degree of the carbon material and multiple Raman characteristic parameters, the graphitization degree is equal to the sum of a constant term and the product of each Raman characteristic parameter and a respective weight coefficient; determining the graphitization degrees of selected samples, collecting Raman spectrum data of the samples and extracting Raman characteristic parameters; substituting the determined graphitization degrees of the samples and the extracted Raman characteristic parameters into the multiple linear regression model to obtain the values of the weight coefficients through fitting; and collecting Raman spectrum data of a to-be-detected sample, extracting Raman characteristic parameters of the to-be-detected sample and substituting the Raman characteristic parameters into the multiple linear regression model to calculate the graphitization degree of the to-be-detected sample.

The invention discloses a 10-35kV neutral ungrounded system electric shock identification method, system, device and medium, and belongs to the technical field of power system fault identification, and the method comprises the steps: collecting asymmetric components in a circuit, and carrying out the preprocessing of a relay protection-oriented three-phase asymmetric fault feature signal in the asymmetric components; extracting the pre-processed asymmetric component through an empirical mode decomposition method to obtain an intrinsic mode function; the fault moment relevance of the high-frequency IMF is judged through the amplitude break variable; performing electric shock current amplitude quantitative detection on the low-frequency IMF by constructing a multiple linear regression model based on an energy ratio; and single-phase electric shock and two-phase short-circuit electric shock are distinguished through a multi-feature joint criterion. Hilbert-EEMD transformation reveals that the residual current and the electric shock current have high similarity in spectral characteristics in the electric shock transient process, and an accurate fault moment identification mechanism is constructed based on the amplitude abrupt change characteristic of the high-frequency IMF component.

The present application provides a pore pressure prediction method based on rock physics modeling and multiple linear regression, relates to the oil and gas exploration and development technical field, and the method comprises the following steps: S1: selecting a plurality of reference wells in a secondary structural unit for overpressure analysis; S2: pre-processing the logging data of the reference wells and analyzing the overpressure causes; S3: performing fluid replacement by using the Gassmann equation, and performing solid replacement by using the Brown-Korringa theory, and calculating the rock elastic modulus; S4: selecting an anisotropic soft pore model to calculate the rock effective velocity; S5: performing sensitivity analysis on the elastic parameters and the pressure coefficient; S6: constructing a multiple linear regression model with the elastic parameters having the best correlation with the pressure coefficient, and predicting the pore pressure; and S7: comparing and verifying the prediction result with the Eaton method. The present application avoids the problem of errors caused by relying on the normal compaction trend line, fits the elastic parameters having good correlation with the pressure by using multiple linear regression, comprehensively considers the influence of multiple variables, and has higher prediction and interpretation capability.

The present application relates to a method for judging rock mass drillability based on rotary speed, torque and axial pressure of a drill bit, characterized by the following steps: Step 1, conducting drilling test, collecting drilling axial pressure, rotary speed and torque in real time, and drawing time history curve; Step 2, establishing multiple linear regression model; Step 3, solving the multiple linear regression model in Step 2 to obtain regression coefficient a i ; Step 4, establishing standard normal distribution function of drillability, defining drillability index as f(K z ); Step 5, dividing drillability grade, and evaluating rock mass drillability according to drillability grade. The present application is based on the above method. By extracting relevant data of drilling blast hole of a field drilling machine, the formula is analyzed, drillability grade is divided, and rock mass drillability is evaluated according to drillability grade.

The invention provides an annealing furnace strip steel temperature prediction method and system based on a self-adaptive box separation mechanism. The method comprises the steps that target variables and related variable data are collected, and a training set and a verification set are divided; training a global multiple linear regression model, and calculating the loss Loss on the verification set as a base; based on an adaptive binning mechanism, initializing binning points and performing iterative optimization to obtain an optimal binning point, segmenting the training set by using the optimal binning point, establishing a new multiple linear regression model on a left sub-training set and a right sub-training set, calculating loss (Sp) on a verification set, and comparing the loss (Sp) with a base to determine whether the binning succeeds or not; s3, regarding the successfully binned sub-training set as a new independent training set, and recursively repeating S3 until loss (Sp) after binning of all the sub-training sets is greater than or equal to base, so as to obtain a segmented multiple linear regression model; and predicting the to-be-predicted strip steel temperature of the annealing furnace by using the segmented multiple linear regression model. The method can predict the strip steel temperature of the annealing furnace.

The invention discloses a high-precision flow monitoring and reckoning system based on side-sweeping rada, and relates to the technical field of hydrological monitoring. A side-sweeping rada monitoring module is used for carrying out spatial intensive scanning on a river section, and surface flow velocity data of dozens of vertical lines are synchronously obtained; a high-resolution surface flow velocity field space sample set is constructed, then a data processing module performs median filtering and data alignment preprocessing on the sample set, and a key vertical line group which is most closely associated with the overall flow state dynamics is intelligently screened out by calculating a correlation coefficient between the flow velocity of each vertical line and the average flow velocity of an ADCP actually measured section; on the basis of a multiple linear regression model, the surface flow velocities of the key vertical lines serve as independent variables, a calculation equation of the average flow velocity of the section is constructed, and the equation quantifies the contribution degree and the influence mode of each key vertical line to the average flow velocity of the section; according to the mechanism, the complex synergistic effect of different flow zones such as a main flow zone and a shoreside attenuation zone in the section on the overall average flow velocity can be carefully captured.

The invention discloses a CO2 flooding minimum miscible pressure prediction method, system and equipment for in-situ crude oil and a storage medium, and the method comprises the steps: S1, obtaining multiple pieces of CO2 flooding minimum miscible pressure data of in-situ crude oil with high intermediate component content, and carrying out the data preprocessing to remove extreme values; s2, establishing a multivariate linear regression model for predicting the minimum miscible pressure based on the preprocessed data, and removing the collinearity of the multivariate linear regression model through ridge regression to obtain a minimum miscible pressure prediction formula; and S3, predicting the minimum miscible pressure of the in-situ crude oil by adopting a minimum miscible pressure prediction formula. The method is suitable for CO2-driven formation crude oil of formation crude oil with high intermediate component content, and can accurately predict the minimum miscible pressure value.

The invention discloses a forage soybean yield measuring method, and belongs to the technical field of crop yield detection, and the method comprises the following steps: S1, obtaining core basic information of a target land parcel, and carrying out homogeneous yield measuring unitization division on the land parcel; s2, dynamically adjusting the side length of the grid according to the planting density, setting center-edge double-region sampling points in a stepped manner according to the area of a homogeneous yield measurement unit, and arranging a sampling frame; s3, respectively collecting morphological parameters and spectral parameters, and measuring grain potential parameters; s4, respectively constructing multiple linear regression models of the fresh grass yield and the grain yield; s5, after the model measurement values of all the homogeneous yield measurement units are corrected, the total yield of the fresh grass and the grains of the target plot is obtained through summarizing; and S6, enabling the core parameter variation coefficient of all sampling points in each homogeneous yield measurement unit to be not higher than 15%, and generating a color graded yield distribution diagram and an Excel report. The technical problems that a traditional yield measuring method is weak in pertinence, high in destructiveness, large in error and low in efficiency are solved.

The invention provides an infrared temperature measurement curve model construction method, a medium and equipment, and relates to the technical field of infrared temperature measurement, and the method comprises the steps: obtaining the focal plane temperature of an infrared thermal imaging equipment detector and AD values of a plurality of radiation calibration sources collected by the equipment; taking the focal plane temperature of the detector and the AD values of the radiation calibration sources at the lowest temperature point, the middle temperature point and the highest temperature point as input, taking the AD values of the radiation calibration sources at the other temperature points in the plurality of radiation calibration sources as output, and constructing a multiple linear regression model; acquiring the focal plane temperature of the infrared thermal imaging equipment detector to be constructed and the acquired AD values of the radiation calibration source at the lowest temperature point, the intermediate temperature point and the highest temperature point, inputting the multiple linear regression model to obtain the predicted AD values of the radiation calibration source at the other temperature points, and performing curve fitting based on the acquired AD values and the predicted AD values to obtain the radiation calibration source of the infrared thermal imaging equipment. And obtaining an infrared temperature measurement curve model. According to the method, the construction efficiency of the infrared temperature measurement curve model can be improved.

The invention provides a rail transit ticket business clearing method and system based on dynamic path selection. The method comprises the following steps: screening effective paths of target starting and ending stations to form an effective path set; calculating the path selection probability of the effective path; in combination with the path selection probability, the proportion of the key influence factors and the weight coefficient of the key influence factors dynamically adjusted by a multiple linear regression model, calculating the distribution proportion of the operation main body in each path interval in the effective path; and performing aggregation calculation on the distribution proportion of each path interval of each effective path in the effective path set to obtain the clearing proportion of the operation main body in the ticket revenue of the target start-stop station. And multi-dimensional splitting in the section is carried out to final aggregation of the operation main bodies, so that reasonable clearing of tickets among the operation main bodies is realized, and legitimate rights and interests of all parties can be guaranteed.

This invention relates to a method for predicting plant element content based on the interannual variation of element content in jujube tree rings. The method includes: (1) cutting off the main trunk of a jujube tree, cross-dating the annual rings and slicing them, measuring the nitrogen, phosphorus, and potassium content in each year of the trunk's annual rings, determining the correlation coefficients between the nitrogen, phosphorus, and potassium content in each year of the annual rings and the elemental indicators for the predicted year, and screening based on the obtained correlation coefficients to obtain screening indicators; (2) using the obtained screening indicators to establish multiple linear regression models for nitrogen, phosphorus, and potassium elements respectively, obtaining the optimal multiple linear regression model, and predicting the interannual variation of nitrogen, phosphorus, and potassium content in the annual rings. This invention's method for predicting plant element content based on the interannual variation of element content in jujube tree rings constructs a multiple linear regression model to predict the interannual variation of nitrogen, phosphorus, and potassium content in the annual rings, enabling balanced fertilization in jujube orchards.

The application relates to the technical field of aircraft design and cost optimization, and discloses an aircraft low-cost factor verification method based on counterfactual reasoning, which comprises the following steps: based on the historical design scheme of an aircraft and the corresponding total cost, constructing an influence factor dataset and a total cost dataset; calculating the Pearson correlation coefficient between the cost influence factors and the total cost, so as to determine the first key cost influence factor; constructing a multiple linear regression model by using the influence factor dataset, and determining the second key cost influence factor by solving the regression coefficient vector of the multiple linear regression model, and then performing consistency judgment on the first key cost influence factor; then, a benchmark scheme is constructed, and a counterfactual reasoning verification method is adopted to further perform reliability verification on the key cost influence factors, so as to determine the final key cost influence factors, thereby providing support for aircraft design.

This invention discloses an intelligent tea brewing method based on OpenMV machine vision, belonging to the field of intelligent tea brewing technology. It includes acquiring real-time image data of tea infusion in a brewing container using an OpenMV camera at a preset frequency. The image data includes RGB color channel information. This invention acquires tea infusion images in real time using OpenMV, utilizes adaptive threshold segmentation and multi-dimensional color feature extraction, and combines a multiple linear regression model to transform visual information into accurate concentration estimates, replacing traditional human sensory judgment and solving the quality instability problem caused by open-loop control. It introduces a PID control algorithm combined with a water temperature compensation factor to dynamically adjust the heating power according to the concentration change trend. By dynamically time-warping and comparing the real-time curve with the standard leaching curve, it automatically adapts to different tea varieties and abnormal conditions, achieving a leap from fixed programs to data models, significantly improving the intelligence of home tea brewing machines.

The invention discloses a big data platform service management system based on an AI algorithm, and relates to the technical field of big data platform service management, and the big data platform service management system comprises a data sensing module, a task priority evaluation module, a resource allocation module and a resource scheduling execution module. The data sensing module collects platform data in real time and monitors data quality; and the task priority evaluation module performs quantitative evaluation on the task priority by utilizing a machine learning algorithm and a multiple linear regression model in combination with the real-time requirement, the data size, the business importance and other characteristics, and optimizes an evaluation result by correcting parameters. Through intelligent priority evaluation and resource allocation, the resource utilization rate and task execution efficiency of the big data platform are remarkably improved, the reliability of data transmission is enhanced, and an efficient and flexible solution is provided for big data processing.

The invention relates to the technical field of automatic medical treatment, provides an interventional nursing process control scheme, and particularly relates to an interventional nursing process monitoring method and device and electronic equipment. According to the method, the basic medical data related to clinic is constructed, and the basic medical data is subjected to linear regression processing twice through the screening model pair formed by combining the unit linear regression model and the multiple linear regression model, so that the associated medical data highly related to the bleeding risk is obtained; and carrying out risk probability calculation on the associated medical data through the prediction model to obtain a final risk value, so that the bleeding risk in the nursing stage is determined. Compared with the prior art, according to the embodiment of the invention, the hemorrhagic high-risk patient can be automatically identified in the early stage of the nursing stage, and the occurrence of hemorrhage conditions is reduced.

The application provides a multi-feature lithium battery health state online estimation method and device, comprising the following steps: step S1, acquiring first feature F1, second feature F2, third feature F3, fourth feature F4, fifth feature F5 and sixth feature F6; step S2, taking the first to sixth features in step S1 as an input vector X=[F1, F2, F3, F4, F5, F6], establishing a hypothesis function formula of a multiple linear regression model: Y=θ0+θ1F1+θ2F2+…+θ6F6, and preliminarily determining a parameter vector φ=[θ0, θ1, …, θ6] by using a gradient descent algorithm; step S3, randomly generating N initialization particles according to the parameter vector φ, updating each initialization particle by using Gaussian white noise, and updating the parameter vector and the predicted value of the multiple linear regression model after importance sampling and resampling. The application extracts multiple features from different angles to construct a more accurate multiple linear regression model, and updates the model parameters online, thereby improving the efficiency and accuracy of online estimation.