546 results about "Independent predictor" patented technology

The invention relates to the technical field of photovoltaic power prediction, and discloses a photovoltaic power prediction method and system, and the method comprises the steps: obtaining the historical operation data of each photovoltaic station, carrying out the abnormal value elimination, expanding the sample data through a generative adversarial network, constructing a first training sample set, and carrying out the variable dimension reduction, screening principal component factors influencing the photovoltaic power to construct a second training sample set; calculating similar days by using the second training sample set, and screening and sorting; establishing a deep learning framework fusing the long short-term memory network, the maximum temperature prediction model and a parameter optimization algorithm, and based on a preset photovoltaic power prediction precision evaluation index, performing model training by taking the number of days of similar days and the weight as independent variables to establish a power prediction model; based on preset reanalysis data and weather forecast data, photovoltaic power prediction is carried out by using the trained power prediction model, precision evaluation and dynamic optimization of model parameters are carried out, and the photovoltaic power prediction precision and adaptability to different scenes are improved.

The invention provides a renewable energy equipment fault intelligent diagnosis method based on deep learning. The method comprises the following steps: S1, collecting data of a photovoltaic system and the like, carrying out data preprocessing, and constructing a training sample set with a fault label; s2, constructing a fractional order physical information neural network in a weighted shift Grunwald-Letnikov format, and outputting an equipment fault probability prediction value; s3, establishing a segmented Poisson regression model, taking a time variable and an environment virtual variable as independent variables, performing dynamic trend modeling on the equipment fault probability predicted value, and outputting the fault probability after statistical correction; s4, fusing the equipment fault probability prediction value and the corrected fault probability to generate a final fault probability; and S5, optimizing prediction strategy parameters based on a reinforcement learning method of a relative reward regression mechanism. According to the invention, the identification capability of implicit fault features is improved, and the fault diagnosis accuracy is significantly improved.

The invention provides a BIM-based bridge construction progress monitoring method and system, and the method comprises the steps: obtaining an observation matrix based on BIM model data and sensor monitoring data of a target bridge; inputting the observation matrix into a preset Kalman filter to obtain a corresponding fusion matrix; carrying out Granger causal test by taking the overall progress deviation as a dependent variable and taking the stress ratio, the environmental index, the resource delay rate and the process progress deviation as independent variables to obtain a Granger causal test result; and constructing a corresponding DAG, and obtaining the reason causing the overall progress deviation based on the DAG. According to the scheme, BIM model data and sensor monitoring data are accurately fused, the DAG between the overall progress deviation and each piece of quantitative information is constructed, the quantitative information causing the overall progress deviation can be accurately obtained, and an effective basis is provided for subsequent decision making.

A dynamic prediction method for the residual gas content of a pre-extraction coal seam comprises the steps that on-site working conditions are considered, parameters such as gas content changes and related geological factors and extraction factors are collected, and a dynamic monitoring data set is constructed; screening main control factors based on grey relational degree, and determining the main control factors influencing the residual gas content after extraction; a main control factor of the residual gas content is used as an independent variable, a pre-pumping dynamic correction factor is combined, a dynamic prediction model is established, and actually measured data is adopted regularly to verify and optimize parameters of the dynamic prediction model; dividing a pre-extraction working face into uniform grid units, and calculating the residual gas content of each grid based on a dynamic prediction model; an extraction blind area is judged according to a quantitative standard and is visually presented; and formulating a differentiated drilling optimization scheme according to the blind area distribution. According to the method, the residual gas content prediction model dynamically responding to mining condition changes is constructed, extraction blind area prediction and drilling optimization are achieved, the coal seam gas extraction efficiency can be improved, and gas disasters are reduced.

The invention relates to a construction method of a multi-class lipid retention time general prediction model and a prediction system. The method comprises the following steps: by taking experimental retention time tRE of a lipid compound in a training set as a dependent variable and characteristic structure parameters of the lipid compound as an independent variable, carrying out quantitative processing on the independent variable and then carrying out regression modeling analysis, the characteristic structure parameters comprise the total carbon number c and the total carbon-carbon double bond number d of a fatty acyl chain, an ether chain, an alkenyl ether chain or / and a sphingosine skeleton, the types of skeletons contained in the lipid compound and the number of corresponding skeletons, and the types of characteristic groups and residues in the lipid compound and the number of corresponding residues; carrying out numerical quantization on parameters according to the number of skeletons, characteristic groups or residues of the corresponding types; the QSRR general prediction model of the retention time of the multi-class lipid compounds is obtained by adopting regression modeling, a more accurate MRM data acquisition window can be set for the lipid compounds, and the sensitivity, stability and coverage can be improved.

The invention provides a vegetation prediction model construction method, a vegetation prediction method, a vegetation prediction device and vegetation prediction equipment, and relates to the technical field of data processing. The vegetation prediction model construction method comprises the following steps: taking ecological environment data of a target area as an independent variable, taking vegetation spatial distribution data corresponding to the target area as a dependent variable, and training a first random forest model to obtain a vegetation spatial distribution prediction model; taking the output of the vegetation parameter spatial distribution prediction model and the period-by-period meteorological data as independent variables, taking the period-by-period vegetation feature change data as dependent variables, training a second random forest model, and obtaining a vegetation feature time change prediction model; and splicing the vegetation spatial distribution prediction model and the vegetation feature time change prediction model to obtain a vegetation prediction model. According to the technical scheme, the data utilization efficiency can be improved, the model training time is shortened, and meanwhile the adaptability of the vegetation prediction model and the accuracy of vegetation prediction are improved.

The invention discloses a large-particle pneumatic conveying pipeline optimal wind speed calculation method based on digital twinning and machine learning, and the method comprises the steps: S1, carrying out the digital twinning modeling of a gas fluid region and large-volume particles, including an overall pipeline fluid model, fluid region grid division, a large-particle model, and a large-particle transportation volume fraction algorithm; s2, a gas-solid coupling numerical simulation experiment is carried out to calculate simulation data, and the optimal conveying wind speed for conveying spherical particles with different volumes and masses is obtained; and S3, fitting simulation data by adopting multiple linear regression analysis in machine learning, establishing a pipeline optimal wind speed calculation model, and determining a high-order term order of an independent variable. According to the method, digital twinborn modeling is conducted on the pneumatic conveying pipeline and the large-size particles, the optimal conveying wind speed for conveying the spherical particles of different sizes and masses is obtained through a gas-solid coupling numerical simulation experiment, and it is avoided that data are obtained through complex, tedious and even difficult-to-execute actual experiments.

The invention relates to a sarcopenia prediction model construction method and system without physique measurement, and belongs to the field of medical informatization. The method comprises the steps of establishing a crowd data set, preprocessing sample data, dividing a training set and a test set, screening modeling features, constructing a plurality of machine learning classifiers, selecting an optimal model, evaluating and predicting performance and the like. The high-performance sarcopenia prediction model can be constructed by only using information such as age, gender, height, weight and the like which can be obtained through questions and answers or self-evaluation as independent variables without physical fitness measurement such as grip strength and physical fitness test. The system comprises a data acquisition module, a data analysis module and a result output module, can realize rapid, low-cost, remote and self-evaluation prediction of sarcopenia risk, is suitable for scenes of various medical institutions, communities, families and the like at all levels, and is beneficial to early screening, intervention and dynamic monitoring of sarcopenia.

The invention discloses a method for automatically solving the critical heat flux density of a reactor based on a statistical principle, and belongs to the field of safety analysis of nuclear reactors. The problems that an over-fitting problem exists, a data set cannot be verified clearly through a fitting data set, and the development process is complicated when a traditional method is used for predicting independent variable combination items of a relational expression artificially given by CHF relational expression development are solved. The method comprises the following steps: collecting reactor CHF experimental data points; eliminating possible repeated experimental points and cold rod critical points to form a development database; based on a layered random sampling method, extracting data from the development database, and determining a fitting data set and a verification data set; based on the fitting data set, selecting a curve form of a single independent variable or a combination item; the independent variables and the combination items with the collinearity problem are removed; importance ranking analysis of the variables and the combination items is obtained through regression coefficient standardization; forming a final CHF relational expression; and based on the verification data set, verifying the CHF relational expression.

The invention discloses an indoor temperature and humidity spatial data prediction method fusing spatial interpolation and a graph convolutional network, and relates to the field of smart city and environment monitoring, and the method comprises the steps: obtaining indoor multi-point temperature and humidity data actually measured by a sensor, and the spatial position information of a monitoring mark point corresponding to the indoor multi-point temperature and humidity data; the temperature and humidity data, the spatial position information and the time dimension features of the marked points are used as input of Kriging interpolation, and temperature and humidity linear predicted values of part of unmarked points are generated; constructing a multiple linear regression model taking the Kriging prediction result, the measured data of the mark points, the time dimension characteristics, the spatial characteristics and the external meteorological data as independent variables and the target temperature and humidity values as dependent variables, and performing multi-dimensional fitting on the Kriging interpolation result; and constructing a feature matrix by using a linear predicted value output by the regression model and other features, and taking the feature matrix and an adjacent matrix as input of a graph convolutional network to generate indoor temperature and humidity three-dimensional space distribution. According to the invention, high-precision indoor space data prediction can be realized.

The invention relates to the field of soft measurement, discloses a soft measurement method based on an optimization algorithm, and solves the industrial pain points of fast precision attenuation, poor mechanism adaptation and high maintenance cost of a traditional soft measurement model. Comprising the steps of obtaining historical production data and a configuration file, and processing and integrating the historical production data and the configuration file into a complemented data file; determining a candidate state variable set according to the complemented data file, and searching an optimal independent variable set through optimization algorithms such as a genetic algorithm; training a data driving model by using the optimal independent variable set to generate a soft measurement model; an unmeasurable input variable is used as a genetic algorithm optimization target, a fitness function is constructed through an average absolute error of a predicted value and an actual value, and an optimal solution is output through iterative evolution. By comprehensively applying various optimization algorithms, the precision and the reliability of the soft measurement model are effectively improved, powerful support is provided for real-time optimization of an industrial control system, and the production efficiency and the product quality can be improved.

The invention discloses a Beidou ground-based enhancement-oriented ionized layer weighted modeling method, which comprises the following steps of: proposing a weight function which takes an included angle between a connecting line from an interpolation point to a main reference station and each baseline as an independent variable according to the correlation of ionized layer delay of a to-be-interpolated point and a baseline and the characteristic related to the included angle between the to-be-interpolated point and the baseline; according to the correlation of ionosphere delay of a point to be interpolated and a baseline and the characteristic related to the distance ratio between the point to be interpolated and the baseline, proposing a weight function which takes the ratio of the distance from the interpolation point to a main reference station to the length of each baseline as an independent variable; the normalized ionosphere delay sample weight can be obtained by integrating the two weight functions, and the value of the point to be interpolated is obtained by using the weight according to the interpolation model of the BLUP framework. High interpolation precision can be achieved in each area in a large and medium-sized network with an active ionized layer, and the service quality of network RTK (Real-Time Kinematic) is improved.

The invention discloses a sewage energy-saving control method and system based on multi-dimensional variable data analysis, and relates to the technical field of data analysis, and the method comprises the steps: analyzing the incidence relation between variable data in different processing stages, the incidence relation of variables comprises the relation between independent variables and dependent variables and the interaction relation between the independent variables, and the relation between the independent variables and the dependent variables is the relation between the independent variables and the interaction relation between the independent variables; the complex coupling relation among multiple variables is fully and accurately described, and a reliable basis is provided for follow-up regulation and control. And determining the corresponding regulation time delay and variability on each regulation node, and considering the regulation hysteresis and the variable change condition on the sewage treatment flow, thereby setting a threshold value and generating an evaluation index for guiding the subsequent sewage regulation. The regulation and control strategy of each regulation and control node is controlled based on the incidence relation and the regulation and control time delay between the variable data, the adaptability and accuracy of sewage node analysis and control are improved, the real-time sewage treatment effect is ensured, and the sewage treatment efficiency is improved.

The embodiment of the invention provides a mudstone interlayer prediction method and device, electronic equipment and a storage medium. The method comprises the steps of performing high-resolution frequency extraction processing on pre-stack sub-angle gather data; pre-stack seismic elastic inversion of well logging constraint is carried out by using high longitudinal and transverse resolutions of well logging and earthquakes; determining independent and sensitive reservoir elastic parameters of the reservoir predicted by the mudstone interlayer; the selected reservoir elastic parameters serve as independent variables, a mudstone factor recognition function of the mudstone interlayer is constructed, and a calculation model of the mudstone factor recognition function of the mudstone interlayer is established; and according to the established mudstone factor identification function model and the selected reservoir elastic parameter inversion data as variable parameters, a division standard is established, and mudstone interlayer prediction is completed. According to the method, the earthquake identification capability of the mudstone interlayer can be improved, the accuracy of turbidite sandstone earthquake prediction can be improved, and the method has great application values for analyzing the connectivity of oil reservoir sand bodies, optimizing an oil reservoir development scheme and improving the oil reservoir recovery efficiency.

A facility for estimating a value relating to a occurrence is described. The facility receives a first occurrence that specifies a first value for each of a plurality of independent variables that include a distinguished independent variable designated to be monotonically linked to a dependent variable. The facility subjects the first independent variable values specified by the received occurrence to a statistical model to obtain a first value of the dependent variable. The facility receives a second occurrence that specifies a second value for each of the plurality of independent variables, the second value of the distinguished independent variable varying from the first value of the distinguished independent variable in a first direction. The facility subjects the second independent variable values specified by the received occurrence to the statistical model to obtain a second value of the dependent variable, the second value of the dependent variable being guaranteed not to vary from the first value of the dependent variable in a second direction that is opposite the first direction.

A computer-implemented method to generate training data with increased fairness. The method includes identifying a set of training data comprising at least one independent variable and dependent variable. The method includes analyzing the set of training data to identify correlations between the independent and the dependent variables. The method further includes identifying at least one correlation between the one or more independent variable and the dependent variable. The method includes calculating a fairness score for each first independent variable against the dependent variable. The method further includes creating, based on the analyzing, a fairness profile for the set of training data. The method also includes generating, by a generative adversarial network (GAN) and based on the set of training data and the fairness profile, a set of synthetic training, where the GAN is configured to increases the fairness score for each variable with a disparate effect score above a threshold.

The present application relates to a lung-cancer screening model construction apparatus, a lung-cancer screening apparatus, and a device and a medium. The lung-cancer screening model construction apparatus comprises: a sample acquisition and data collection module, which is used for acquiring benign pulmonary nodule patient samples, and collecting baseline parameters of the benign pulmonary nodule patient samples, wherein the baseline parameters comprise radial growths of pulmonary nodules and serum thymidine kinase 1; a model training module, which is used for performing model training by using the baseline parameters as independent variables and using, as dependent variables, risks of lung cancer newly diagnosed within three years after baseline visits; and a model evaluation and verification module, which is used for performing prediction efficiency evaluation and internal verification on the trained model, so as to obtain a risk prediction model for newly diagnosed lung cancer of benign pulmonary nodule patients within three years. The present application can be used for predicting the risk of progression from benign pulmonary nodule to lung cancer in a large-sample Chinese population.

The embodiment of the invention provides a feature screening method and device, a storage medium and electronic equipment, and the method comprises the steps: obtaining an initial feature set and a target dependent variable; taking each abnormal value threshold in the plurality of abnormal value thresholds as a current abnormal value threshold, and executing the following feature screening operation to obtain a plurality of feature sets: according to a correlation coefficient between each initial independent variable and a target dependent variable and the current abnormal value threshold, performing linear regression processing on each initial independent variable in sequence to obtain a plurality of feature sets; and obtaining a saliency probability value of each initial independent variable under the current abnormal value threshold, so as to screen the initial independent variables from the initial feature set, obtain a feature set corresponding to the current abnormal value threshold, and select a target feature set. The problem that the reliability of a prediction model is low when data containing abnormal values are used for training is solved, and the effect of improving the prediction efficiency of the model is achieved.

The invention provides a disease burden prediction and prevention and control decision-making method and system based on machine learning, and relates to the technical field of disease prediction and public health decision-making, and the method comprises the steps: obtaining epidemiological data of tuberculosis and related diseases from an authoritative database, and carrying out the preprocessing of the epidemiological data; respectively training an XGBoost model, an RF (Radio Frequency) model and a Prophet model; training a random forest meta-model by adopting a Stacking fusion strategy, and constructing a hybrid prediction model; calculating RMSE, MAE, MAPE and Rindex evaluation model performance; based on the obtained hybrid prediction model; based on the variable importance analysis result and the prediction result, the influence of the key independent variable on the tuberculosis burden dependent variable is quantified, the effect of the independent variable change on the dependent variable is simulated, and a tuberculosis prevention and control intervention strategy suggestion is generated. By fusing multi-source data and a hybrid modeling technology, tuberculosis prediction precision is remarkably improved, confidence interval quantization and prevention and control strategies are linked for the first time, and data-driven decision support is provided for global tuberculosis prevention and control.

The invention belongs to the crossing field of urban design and multidisciplinary, and particularly relates to a mountain historical block ownership perception evaluation method based on multi-modal data, public spaces of mountain historical blocks are divided into multiple types, an experimental path is selected from each public space, and multiple experimental parking points are selected from each experimental path; the evaluation index system comprises physiological data, psychological data and spatial data, and the spatial data comprises feature data and acoustic environment data; for each experiment path, obtaining feature data, recruiting a plurality of experimenters, collecting acoustic environment data and physiological data of each experimenter in real time, and collecting psychological data from the experimenters in real time in a questionnaire mode at an experiment stop point; independent variables are defined as spatial data, dependent variables are defined as psychological data and physiological data, based on experimental data, the relation between the independent variables and the dependent variables is learned through a random forest model, and bar charts of the influence degrees of the independent variables and the dependent variables are output; and the evaluation result is more accurate and persuasive.

The invention discloses a sintering batching optimization scheme, and relates to the technical field of sintering ore batching control, and the technical key points are as follows: a random forest proxy model is used to predict the performance of finished ore, and a nonlinear mapping relationship between an independent variable and a dependent variable is used as a target function; carrying out Pareto optimal solution set search through non-dominated sorting and crowding distance calculation by adopting an NSGA-II (Non-dominated Sorting Genetic Algorithm-II) algorithm; calculating the weight of each target index by using an entropy weight method based on a Pareto solution set; weight is determined from a Pareto solution set generated by an NSGA-II algorithm, an optimal batching scheme closest to an ideal solution and far away from a negative ideal solution is screened out through a TOPSIS method, and multi-target comprehensive weighing is achieved; according to the method, sintering batching optimization based on finished ore performance feedback is realized, dynamic optimization of a batching plan can be effectively guided, the sinter quality control level is improved, the production cost is reduced, and technical support is provided for efficient and stable operation of a blast furnace ironmaking process.

The invention relates to the technical field of electric power, in particular to an electricity price fluctuation quantification method, device, equipment and medium based on regression analysis, and the method comprises the steps: taking electric power price fluctuation in peak load of electric power as a dependent variable, and taking new energy power generation, thermal power generation, electric power demand, generating capacity and capacity electricity price as independent variables; establishing a multiple linear regression model; verifying whether the independent variable and the dependent variable have a linear relationship or not, preliminarily carrying out multiple linear regression analysis, finding an optimal regression coefficient by minimizing a residual sum of squares, and carrying out multiple linear problem judgment; a ridge regression analysis method is adopted, a regularization item is added into a loss function to limit the size of a regression coefficient, overfitting of the model is reduced, and a ridge function is confirmed through cross validation; performing overall significance test and regression coefficient significance test on the multiple linear regression model, and analyzing the fitting degree; and performing quantitative analysis on electricity price fluctuation by using the fitted multiple linear regression model.

The invention relates to an anaerobic ammonia oxidizing bacteria enrichment optimization control method based on ecological niche regulation and control, which comprises the following steps: S1, selecting key ecological niche indexes of a constructed wetland as investigation factors, performing difference setting on the investigation factors, forming experimental systems of different ecological niche, controlling and monitoring the constructed wetland, and collecting data; s2, taking the relative abundance and / or absolute abundance of the anaerobic ammonia oxidizing bacteria collected in the step S1 as dependent variables, taking key ecological niche indexes as independent variables, and performing fitting so as to predict a model; and S3, performing prediction through the prediction model, comparing a prediction result with a control target, and performing optimization control on the anaerobic ammonia oxidizing bacteria enrichment of the constructed wetland according to a comparison result. According to the method, the prediction model with relatively high prediction accuracy can be obtained, so that the enrichment of anaerobic ammonia oxidizing bacteria can be ensured, the sewage treatment efficiency can be improved and a relatively good total nitrogen removal effect can be obtained by adjusting ecological niche indexes in the constructed wetland by virtue of the prediction model.

The invention discloses a distribution transformer load loss prediction method and system based on pilot-scale test monitoring data. The method comprises the following steps: collecting pilot-scale test data of a distribution transformer load loss related variable; preprocessing the collected pilot plant test data, and performing correlation analysis on the preprocessed variable data to obtain an independent variable in a distribution transformer finished product load loss prediction function; performing multiple regression analysis on the selected independent variables to obtain a regression model, correcting the regression model to obtain a distribution transformer finished product load loss prediction function, and verifying the load loss prediction function; and performing load loss prediction on the distribution transformer finished product based on the verified load loss prediction function. According to the method, the predicted value of the load loss can be quickly obtained by analyzing pilot test data, the time and the cost of a traditional experiment are reduced, and the calculated load loss prediction precision can be effectively improved by analyzing variables and verifying functions.

The invention discloses a fracture pressure prediction and internal friction angle joint inversion method for a heterogeneous stratum. The invention belongs to the technical field of petroleum engineering and artificial intelligence cross application, well depth is used as an independent variable, the minimum / maximum horizontal stress, pore pressure and cohesion are combined to construct well depth mixed input, only multi-scale periodic basis function transformation is carried out on the well depth, and a linear well depth channel is reserved; establishing a rupture pressure depth mapping model, introducing an augmented Lagrange physical constraint containing dual variables and a dynamic threshold, adaptively adjusting a physical weight according to a physical residual error and a threshold difference value, realizing dynamic balance of a rupture criterion and data fitting, and searching and optimizing a model topology by using a multi-target network structure; under a limited sample, a fracture pressure profile and a physically reasonable internal friction angle can be obtained at the same time, and the prediction precision and the physical consistency are improved.

The present invention discloses a method and device for determining the effect of activity communication, a storage medium, and a terminal, which relate to the field of media communication technology, and the main purpose is to solve the problem of low quantification accuracy of the effect of activity communication. It mainly includes obtaining the index parameters of the multi-dimensional activity communication index of the sample activity; determining a regression model with the activity communication index as the independent variable and the activity effect index as the dependent variable, and determining a neural network model with the activity communication index as the input and the activity effect index as the output, using the first weight coefficient of the regression model and the second weight coefficient of the neural network model to perform a fusion calculation to determine the target weight coefficient, and based on the target weight coefficient, the activity communication index and the activity effect index, generate an activity effect index formula; using the activity effect index formula, calculate the index parameters of the multi-dimensional activity communication index of the target activity to determine the target activity effect index value.

The invention provides a dynamic prediction algorithm for monitoring late-onset infection of a premature infant and an upgrading system. The dynamic prediction algorithm for monitoring late-onset infection of the premature infant and the upgrading system comprise the following steps: a, collecting clinical data of the premature infant, including birth weight, gestational age, 1-minute and 5-minute Apgar scores, right hand perfusion index, lower limb perfusion index and other related clinical information, and b, determining the early-onset infection of the premature infant through medical history collection and physical sign analysis. The method comprises the following steps: collecting data of 11 classification independent variables: prenatal antibiotic use conditions (existence and absence); according to the dynamic prediction algorithm for monitoring late-onset infection of the premature infant and the upgrading system, the infection risk index is effectively calculated through high-risk factors analyzed by the Lasso regression model in combination with clinical basic data of the premature infant, early warning of infection of the premature infant is provided for medical staff, and the accuracy and timeliness of infection prediction are remarkably improved. Besides, the system can automatically remind medical staff to intervene the high-risk child patient through red warning, so that the death rate caused by delayed discovery and delayed treatment is reduced, and the clinical intervention effect is improved.

The invention discloses a drainage pipe network liquid level and risk prediction method, relates to the technical field of drainage pipe networks, and solves the problems that in the prior art, a hydrology and hydrodynamics mechanism model needs a large amount of basic data and is complex in source, and if the data precision and accuracy are not enough, the deviation between a calculation result of the model and an actual result is large. The invention relates to a method for judging rainy days and screening rainfall periods according to rainfall data. A least square method is adopted to establish a multiple linear regression relation among the independent variable accumulative rainfall capacity, the rainfall duration and the dependent variable pipe network net rising liquid level, algorithm training is completed, and therefore the drainage pipe network liquid level is predicted; the critical rainfall and the rainfall are calculated through the set drainage pipe network liquid level threshold value, the rainfall is compared with the critical rainfall, and whether the drainage pipe network liquid level has risks or not can be judged; according to the method, drainage pipe network facilities and related topographic data do not need to be considered, and the over-fitting phenomenon possibly caused by complicated parameter training is avoided.

The invention discloses a robust classification method for images with noise labels based on continuous learning time-varying regularization, and the method comprises the steps: enabling the learning of the noise labels to be regarded as a continuous learning process which comprises a plurality of learning tasks, and enabling network parameters to be decomposed into a clean part and a noise part; the dynamic regularization technology is adopted to control the updating process of clean parameters and the updating process of noise parameters, and the method comprises the specific steps that conventional training is conducted on a data set with noise labels through cross entropy loss, the accuracy of a model in each epoch is tested through a verification data set, and a verification curve A is obtained; respectively carrying out normalization and reverse normalization processing on the verification curve to obtain an incremental curve A1 and a decline curve A2, and solving an incremental function alpha (t) and a decline function beta (t) corresponding to the curves by taking epochs as independent variables; and applying different dynamic regularization constraints to the clean parameter and the noise parameter by taking alpha (t) and beta (t) as dynamic coefficients. According to the robust classification method for the images with the noise labels, the combination of a continuous learning method and a noise label method is promoted, the setting of parameters is avoided, and the interference of the noise labels on the model is inhibited.

Certain aspects involve training tree-based machine-learning models for computing predicted responses and generating explanatory data for the models. For example, independent variables having relationships with a response variable are identified. Each independent variable corresponds to an action or observation for an entity. The response variable has outcome values associated with the entity. Splitting rules are used to generate the tree-based model, which includes decision trees for determining relationships between independent variables and a predicted response associated with the response variable. The tree-based model is iteratively adjusted to enforce monotonicity with respect to representative response values of the terminal nodes. For instance, one or more decision trees are adjusted such that one or more representative response values are modified and a monotonic relationship exists between each independent variable and the response variable. The adjusted model is used to output explanatory data indicating relationships between independent variable changes and response variable changes.