44 results about "Data regression" patented technology

The invention discloses an Internet of Vehicles traffic management method based on Beidou positioning, and belongs to the field of traffic management, and the method comprises the steps: obtaining Beidou positioning data and sensor network local sensing data, and extracting global coordinate information and local traffic state information; performing time synchronization processing on the Beidou positioning data and the sampling data of the sensor network by adopting a timestamp alignment method, and eliminating the difference in time scale; whether signal interruption or drifting exists in the Beidou positioning data is judged, and if signal abnormity exists, missing positioning data is estimated through an interpolation algorithm according to local sensing data of the sensor network; according to the fused data, constructing a dynamic distribution model of vehicles in a global road network, and identifying a relationship between traffic density and road network distribution; and a multi-source data regression model based on a convolutional neural network is adopted to predict the vehicle position during signal interruption or drifting, and continuous global positioning information is generated.

The invention discloses a gas turbine running state data regression prediction method based on a fusion network, a plurality of sensors are arranged at different positions of a gas turbine, and the prediction method comprises the following steps: obtaining multi-source heterogeneous real-time data of the gas turbine in a current control period, the multi-source heterogeneous real-time data comprises sensor time sequence data, high-frequency vibration data, environment working condition data and control instruction log data, and constructing a unified data representation matrix of the multi-source heterogeneous real-time data; and based on the unified data representation matrix, a thermal efficiency prediction value and an emission concentration prediction value of the gas turbine in the next control period are obtained through a dynamic weight fusion network. By constructing the uniform data representation matrix of the multi-source heterogeneous real-time data and combining the dynamic weight fusion network, the problems of insufficient spatial-temporal characteristic difference processing of the multi-source data and rigid dynamic weight distribution in the prior art are effectively solved.

The invention discloses a relational data regression value prediction method based on composite ensemble learning, and belongs to the technical field of data analysis and machine learning. Firstly, principal component analysis and uniform manifold approximate projection are utilized to carry out dimensionality reduction and feature selection on data, main features of the data are extracted, dimensionality of the data is reduced, and main information in the data is reserved. Then, two stacked integrated learning models are adopted to predict the data subjected to dimension reduction, the two models are trained and predicted by adopting different base learners, and the advantages of the base learners are fully utilized; meanwhile, the output weight of the base learner is optimized by adopting a gradient descent method, so that the prediction accuracy is improved. And finally, integrating the output results of the first two models by using the third mulberry stack ensemble learning model, and optimizing the output weight of the model by using the gradient descent method. The multi-level integrated learning strategy can effectively improve the prediction stability and improve the generalization ability of the model. The method has universality and expansibility and is applied to various data intensive fields.

The invention belongs to the technical field of compressor performance prediction, and particularly provides a method and system for determining a surge judgment critical value of a compressor based on multi-source data fusion. Comprising the steps that a compressor surge test system is built and operated, and a first surge critical value is calculated; obtaining multi-source historical data related to surge of the compressor, selecting characteristic variables to construct historical input vectors, and fitting the multiple regression model by using the historical input vectors to obtain a discrimination model; training the machine learning model to obtain a trained machine learning model; a real-time input vector is constructed based on data, collected in real time, of the to-be-monitored compressor, and then a second surge critical value and a third surge critical value are obtained; and fusing the first surge critical value, the second surge critical value and the third surge critical value to obtain a predicted surge critical value of the to-be-monitored compressor. According to the method, high-precision dynamic testing and historical data regression analysis are combined, and the actual value of the surge critical value is finally and quantitatively determined through multi-parameter coupling measurement and data mining.

Methods and systems are disclosed for solving a regression problem, for example a data regression and / or a differential equation problem, over a problem domain. The method comprises: receiving or determining, by a classical computer, a regression problem description and a set of kernel points in the problem domain; receiving or determining, by the classical computer, a trial function associated with the regression problem, the trial function being based on a quantum kernel and being parameterized by kernel coefficient(s); determining, using a quantum computer, for each of the kernel points, a kernel value of the quantum kernel and / or a kernel derivative value of a derivative of the quantum kernel; determining, by the classical computer, a set of optimal kernel coefficients based on the kernel value and / or kernel derivative value and determining, by the classical computer, a solution function based on the trial function and the set of optimal kernel coefficients.

The invention discloses a new energy storage and charging pile resource configuration and charging and discharging scheduling strategy system, and the system comprises a data processing module which collects and processes charging order data, customer vehicle data, external environment data, electricity price data and battery data, and removes noise and abnormal values through data cleaning; the model training module is used for configuring a training set and a verification set for the data cleaned by the data processing module, and training an LSTM model; the reinforcement learning module is used for setting a reward function and optimizing an LSTM model output strategy based on the reward function so that the system can be adaptively adjusted in different scenes, multi-dimensional information such as hardware end battery indexes, power grid loads, user end vehicle types, habits, time periods, environment end temperature and seasons is fused, the state and the output space of the charging pile are predicted by means of historical data regression, and the system can be used for realizing self-adaptive adjustment of the charging pile. And the resource scheduling utilization efficiency is obviously improved.

The invention relates to the technical field of remote monitoring, in particular to a carbon brush intelligent monitoring system which comprises a linkage sampling module, a differential pressure offset module, a curvature tracking module, a track node module and an anomaly identification module. According to the invention, through continuous data acquisition and change trend identification of the operation state of the carbon brush assembly, dynamic monitoring of the rotation speed change and the current jump trend is realized, a periodic data regression determination trigger moment is combined to construct an abnormal characteristic time sequence, and continuous tracking and directional discrimination of the change trend of the carbon brush voltage drop difference are realized. By comparing the abnormal nodes with the periodic jump positions, the abnormal positions and times of the brush blocks in the space and time intersection area are accurately identified, the dynamic identification of the abnormal wear trend and position in the running state of the carbon brush is realized, the accuracy, the timeliness and the relevance of the abnormal wear identification of the carbon brush are improved, and the service life of the carbon brush is prolonged. And fine monitoring and management of the equipment operation state are promoted.

A blood pressure prediction method includes performing data fragment division on the pulse wave; generating an input data tensor according to a total fragment number and the pulse wave data; performing multilayer convolution and pooling calculation on the input data tensor by using a blood pressure CNN to generate a feature data tensor; generating an input data matrix according to the feature data tensor; performing feature data regression calculation on the input data matrix by using a blood pressure ANN to generate a blood pressure regression data matrix; when a prediction mode identifier is a mean prediction identifier, performing mean diastolic pressure data calculation and mean systolic pressure data calculation to generate mean diastolic pressure data and mean systolic pressure data; or, when the prediction mode identifier is a dynamic prediction identifier, extracting diastolic and systolic pressure data to generate a predicted blood pressure data sequence.

The invention discloses an unbalanced data regression method and system, a storage medium and a terminal, and belongs to the technical field of data regression, and the method comprises the steps: constructing a grid division-based multi-granularity space generation model, and dividing similar samples in the same granularity space; oversampling is independently carried out in each granularity space, and an oversampling range is defined by combining the space importance degree and the dynamic neighborhood radius. According to the method, under the scene of large sample size, the data stream and imbalance problems are comprehensively considered, the processing efficiency is improved, the overall distribution of data is considered, the interference of noise samples is reduced, the continuity and internal correlation of the samples are reserved, and a new thought is provided for the field of imbalance data stream regression.

The present invention discloses a method, system, storage medium, and terminal for unbalanced data regression, belonging to the field of data regression technology. The method includes: constructing a multi-granularity space generation model based on grid division, dividing similar samples into the same granularity space; independently oversampling in each granularity space, wherein the oversampling range is defined by combining spatial importance and dynamic neighborhood radius. In scenarios with large sample sizes, the present invention comprehensively considers data flow and imbalance issues, improving processing efficiency while taking into account the overall distribution of data, reducing the interference of noise samples, and retaining the continuity and intrinsic correlation of samples, providing new ideas for the field of unbalanced data stream regression.

The present invention discloses a local interpretation method for a time series regression model based on the Lime algorithm, which belongs to the field of machine learning technology. It includes: obtaining a time series data set and a label in a business scenario, wherein the time series data in the time series data set is multi-feature dimension data; using the time series data set in the business scenario to train a time series data regression model, the trained time series data regression model can output a prediction result based on the input time series data, and generate an explanation for the prediction result; the generation of an explanation for the prediction result refers to generating sub-features that have a greater impact on the prediction result, and generating features that have the greatest impact on the prediction result; drawing a positive and negative correlation line graph of the impact of the most influential feature on the prediction result at each time interval. The present invention can realize the interpretation and analysis of time series data, obtain the time period and its features that have the greatest impact on the prediction result, and can be used to guide the adjustment of production parameters in the corresponding business scenario.

The invention provides a mixed data regression model-based pm2.5 influence factor analysis method and system, and the method comprises the steps: building a mixed data regression model of which covariables are component data and numerical data and dependent variables are functional data: the mixed data regression model is a pm2.5 concentration curve of a city, is the functional data, is the proportion of first yield, second yield and third yield of the city, and is called the proportion of third yield for short; the data is component data, logarithm of per capita GDP, average temperature and numerical data, is a to-be-estimated component type coefficient changing along with time, is distributed to a component type covariable at any moment, is a to-be-estimated function type coefficient and is a function type residual error; obtaining robust M-estimation based on equidistant logarithmic ratio transformation, functional basis expansion and an iterative reweighted least square method; and according to the estimated values, analyzing the influence of the three-yield ratio, the per capita GDP and the average temperature of each city on the pm 2.5. The method can be used for analyzing the pm 2.5 influence factors.

The present application belongs to the technical field of compressor performance prediction, and specifically proposes a compressor surge discrimination critical value determination method and system based on multi-source data fusion. It includes building and running a compressor surge test system, calculating the first surge critical value; obtaining multi-source historical data related to compressor surge, selecting characteristic variables to construct a historical input vector, fitting a multiple regression model using the historical input vector to obtain a discrimination model; training a machine learning model to obtain a trained machine learning model; constructing a real-time input vector based on real-time data collected from the compressor to be monitored, and then obtaining the second and third surge critical values; and fusing the first, second and third surge critical values to obtain the predicted surge critical value of the compressor to be monitored. The present application combines high-precision dynamic testing and historical data regression analysis, and ultimately quantitatively determines the actual value of the surge critical value through multi-parameter coupling measurement and data mining.

The invention relates to the technical field of control engineering, and discloses a control system and method for high-temperature gas heating, and the system comprises a high-temperature gas heating physical entity, a sensing detection group, and an advanced control unit. The advanced control unit comprises the following modules which work cooperatively: a data acquisition and preprocessing module which is used for acquiring system output (a gas outlet temperature measurement value), control input and measurable disturbance data acquired by the sensing detection group, and constructing a data regression vector reflecting dynamic historical information of the system; through a dynamic characteristic matrix self-correction module, a dynamic characteristic matrix representing a system input and output mapping relation is constructed and updated in real time by using a recursive algorithm with a forgetting factor; the method effectively solves the problems of strong nonlinearity and time-varying characteristics caused by dramatic changes of gas physical properties and heating element characteristics along with temperature in a high-temperature gas heating process, and remarkably improves the adaptive capacity of the model in a full working condition range.

The invention discloses a small sample table type data regression prediction method and system based on a graph neural network and a medium, and the method comprises the steps: carrying out the preprocessing of obtained table type data, and obtaining a to-be-predicted data sample; performing feature extraction on a to-be-predicted data sample by using a trunk feature extraction module to obtain an initial embedding containing the feature information of the sample; respectively inputting the initial embedding into a residual branch and a gating branch in a feature refining module, and respectively obtaining a direction correction vector and correction intensity corresponding to the to-be-predicted data sample; weighting the direction correction vector according to the correction intensity, then performing residual connection with the initial embedding, and further performing normalization to obtain refined embedding; constructing a sample relation graph according to refined embedding, and further performing information spreading and aggregation by using a graph attention network to obtain enhanced embedding; and splicing the enhanced embedding and the initial embedding, inputting a prediction head, and outputting a regression prediction value. According to the method, the accuracy of small sample table type data regression prediction can be effectively improved.

The invention provides a hardware-in-the-loop simulation method and device based on a data regression model, a medium and a product. The method comprises the steps that simulation data of a hardware-in-the-loop simulation system of a wind turbine generator is collected to serve as a training set of the data regression model; inputting the input data in the simulation data at different moments into the data regression model, taking the wind speed and wind direction data in the simulation data as the output data of the data regression model, and completing the training of the data regression model; inputting the data, collected in real time, of the wind turbine generator into the data regression model, and outputting reconstructed wind speed and reconstructed wind direction data; taking the reconstructed wind speed and the reconstructed wind direction data as the input of a hardware-in-the-loop simulation system of the wind turbine generator, and transmitting the output wind speed and wind direction data and the simulation operation state of the wind turbine generator to a control system of the wind turbine generator; the hardware-in-the-loop simulation system realizes a simulation process. The problem that the scene cannot be reproduced when wind speed measurement is adopted for simulation is solved.

This invention discloses a method for optimizing process parameters to reduce residual stress in the cylinder head of a large-bore diesel engine, comprising the following steps: establishing a J-C constitutive model of the cylinder head material; performing multi-process machining simulation analysis of the cylinder head using the "birth and death element" technique and the J-C constitutive model to obtain simulated residual stress values; obtaining experimental residual stress values ​​using the blind hole method, comparing the experimental values ​​with the simulated values, and controlling the error value within the allowable error range; establishing a nonlinear mapping relationship between cutting parameters and response targets based on a data regression prediction model using a BP neural network to obtain a prediction model; optimizing the cutting parameters of milling, turning, and boring processes of the cylinder head using the prediction model to predict the minimum residual stress value and the optimal cutting force; using the optimal cutting force as a boundary condition and performing machining process simulation analysis to obtain the optimized reduction range of residual stress values. This invention significantly reduces the residual stress in the cylinder head.

The invention belongs to the technical field of automobile sheet hot galvanizing, and particularly relates to a high-end automobile sheet hot galvanizing process surface vibration mark defect forecasting method, which comprises the following steps of S1, collecting main equipment and process parameters of an air knife of a hot galvanizing unit; s2, flow field characteristic parameters of air knife distribution on the surface of the strip steel are calculated; s3, collecting historical production data; s4, calculating a vibration coefficient of the historical production data; s5, statistically calculating a critical vibration coefficient of the historical production data; s6, formulating a vibration ripple generation discrimination model; and S7, the vibration discrimination model is used for predicting the vibration ripple defect of the to-be-produced coil. According to the overall structure provided by the embodiment of the invention, a fitting model of the surface of the strip steel affected by the air knife flow field is developed through a large amount of field data regression and theoretical analysis, and on the basis, the occurrence probability of the vibration marks is represented by the vibration coefficient; and a limit vibration coefficient and a corresponding vibration ripple prediction and judgment model are given in combination with regression of historical data.

The present invention relates to a method and system for assessing the contribution of functional channels to pilot workload. The method comprises: step S1: collecting multidimensional physiological data of a test pilot in an aircraft cockpit while completing a task, and processing the data to obtain workload; step S2: classifying the pilot's aircraft task into different subtask functional categories based on attention shift information during the pilot's interaction with the aircraft, thereby obtaining a multidimensional functional channel execution sequence; and step S3: dynamically adjusting the length of the pilot's cognitive time domain segments for different subtask functions to correct the workload, constructing a data regression model to perform data regression on the multidimensional functional channel execution sequence and the corrected workload, and inversely calculating the weight ratio of the functional channels to the pilot's workload to obtain the contribution ratio of the functional channels to the pilot's workload. Compared with existing technologies, the present invention has the advantages of more comprehensive and accurate workload assessment.

The invention provides a small sample table data regression classification method based on causal reasoning, and belongs to the technical field of artificial intelligence, and the method specifically comprises the steps: generating a plurality of synthetic data sets based on a structured causal model; performing post-processing of nonlinear distortion, quantization processing and missing value simulation on the synthesized data set; constructing a CTFN model with a double-axis attention mechanism based on a Transform network architecture, and alternately executing self-attention calculation in a feature dimension and a sample dimension; carrying out single-step inference type training on the CTFN model by utilizing a context learning mechanism; and applying the trained CTFN model to regression and classification tasks of small sample table data, and outputting a prediction result of a target variable in single forward reasoning. Through the processing scheme of the invention, the small sample generalization ability of the model under highly heterogeneous data is improved.

The application discloses a kind of based on dimensionless array combination's heavy oil steam flooding main control factor analysis and recovery factor prediction method.The method includes: obtaining fluid property, interface property and core physical property parameter;Carry out heavy oil steam flooding experiment under different working conditions, obtain recovery factor data;Capillary number, bond number and gravity number are calculated, the relative action of viscous force, capillary force and gravity is characterized;Dimensionless array combination parameter including bond number, capillary number and viscosity ratio before and after oil displacement is constructed;Quantitative prediction model of recovery factor and dimensionless array combination parameter is established by determining correction coefficient through experimental data regression;The model is applied to the recovery factor prediction and process parameter optimization of steam drive under different preheating temperature, steam dryness, injection speed and injection dip angle and other working conditions.The application realizes the unified quantitative characterization of heavy oil steam flooding multi-acting force, establishes recovery factor quantitative prediction method, and provides theoretical basis for steam drive development plan optimization.

The present application relates to a method, device and storage medium for predicting the production of a horizontal well with volume fracturing in an unconventional reservoir. The method includes the steps of: obtaining the horizontal well parameters of the horizontal well to be predicted; obtaining the production of the vertical wells around the horizontal well to be predicted; constructing a prediction formula for the post-fracturing production of the horizontal well to be predicted; determining the undetermined coefficients of the prediction formula for the post-fracturing production of the horizontal well; and calculating the production of the horizontal well to be predicted. The present application uses the regression analysis method of actual field data to calculate the production of the horizontal well after volume fracturing, takes into account the influence of the degree of oil and gas enrichment and the degree of fracturing transformation on the production increase effect, and at the same time, by introducing the production of the surrounding vertical wells, reduces the difference in the production increase effect in different blocks and different parts. Practice has proved that this calculation method has a certain degree of accuracy; it has strong operability and has been well applied in the exploration and development practice of low-grade reservoirs.

The embodiment of the invention provides a data processing method and a related device, which are used for performing panel regression analysis on a plurality of high-dimensional data by adopting a tree search mode so as to realize clustering of the high-dimensional data and improve the accuracy of clustering of the high-dimensional data. The method comprises the following steps: acquiring a data set, wherein the data set comprises a plurality of pieces of high-dimensional data; panel regression analysis is carried out on the multiple pieces of high-dimensional data, and a panel data regression model of the multiple pieces of high-dimensional data is obtained; performing significance test on a plurality of first intercept in the panel data regression model to obtain a plurality of first significance test results, the first intercept representing differences among the plurality of high-dimensional data; and according to the plurality of first significance test results, the plurality of high-dimensional data are divided into a plurality of data categories, and one data category comprises at least one piece of high-dimensional data.

The present invention is a plating assistance system, a plating assistance device, and a storage medium, which easily determine an implementation condition for improving in-plane uniformity obtained by a plating process. The plating assistance system includes: a simulator (362) that predicts an in-plane uniformity value of a plated film formed on a substrate in accordance with a hypothetical condition related to an electrolytic plating process of the substrate; a numerical analysis data storage section (370) that stores, with respect to a plurality of hypothetical conditions, numerical analysis data that corresponds each hypothetical condition to the in-plane uniformity value; a regression analysis section (250) that calculates, by regression analysis based on the numerical analysis data, a model that takes the in-plane uniformity value as a target variable and takes variables of the hypothetical conditions as explanatory variables; and an implementation condition exploration section (252) that explores, using the calculated model, an implementation condition that is a recommended value of the hypothetical condition related to the in-plane uniformity of the plated film formed in the electrolytic plating process of a plating target substrate.

The invention belongs to the field of power grid abnormity identification, and relates to an abnormal data identification method and system for time-day-and-frequency-band characteristic engineering, and the method comprises the steps: carrying out the processing of initial multi-source data, and obtaining multi-dimensional time series data; decomposing and reconstructing the multi-dimensional time sequence data to obtain frequency band multi-dimensional time sequence data of a plurality of frequency bands; for each piece of frequency band multi-dimensional time sequence data, establishing a frequency band data regression model and screening a core feature set; based on a core feature set of a to-be-evaluated time period, outputting a monitoring data predicted value of the to-be-evaluated time period through the frequency band data regression model, and based on the monitoring data predicted value and target evaluation data, determining abnormal monitoring data; therefore, whether the power grid monitoring data is abnormal or not can be accurately identified, and technical support is provided for fine technical management and control of power grid operation, equipment troubleshooting and scheduling strategy optimization.

To provide an ion channel analysis device capable of obtaining a mathematical parameter for analyzing a state of an ion channel.SOLUTION: An ion channel analysis device 1 includes: a detection unit 20 that detects an ion current I flowing through an ion channel 10; and an analysis unit 30 that acquires first data 40 related to the ion current I detected by the detection unit 20. There is a regression model M that uses, as teacher data, the known first data 40 and a known mathematical parameter 70 corresponding to the known first data 40. The analysis unit 30 inputs the first data 40 related to the ion current I detected by the detection unit 20 to the regression model M, thereby outputting the mathematical parameter 70 corresponding to the input first data 40.SELECTED DRAWING: Figure 1

The invention relates to the technical field of electric power measurement, in particular to a data center PUE real-time monitoring method and system, and the method comprises the following steps: obtaining current, voltage, power, environment and business load data, setting a window to calculate total energy consumption and IT equipment energy consumption, dividing the total energy consumption and IT equipment energy consumption to obtain a PUE value, storing the total energy consumption according to a time sequence, constructing a historical sequence by the PUE and a business load, and calculating the PUE value according to the historical sequence. And inputting the LSTM network to predict a future energy consumption curve. In the invention, current and voltage values, input and output power of a power supply module and operation power of a core component are acquired in real time, and calculation mapping of total energy consumption of a data center and energy consumption of IT equipment is established, so that the problem of one-sided PUE results caused by limitation of traditional measurement points is overcome, and the power supply use efficiency is obtained; energy consumption values and service load data are associated to construct a historical energy consumption sequence, and a future energy consumption expected curve is output by using time series data regression prediction capability, so that the defect that only post statistics can be performed and the trend cannot be pre-judged in the prior art is overcome, and decision support is provided for active energy efficiency optimization.

The specification provides a fracturing wellhead pressure prediction method and device based on multivariate data regression. Multivariate fracturing construction data of a target well in a target area is acquired; a crack initiation point and a pump stop point are determined according to a first-order derivative of a displacement and a first-order derivative of a tubing pressure in the multivariate fracturing construction data, and effective operation data from the crack initiation to the pump stop are determined by using the crack initiation point and the pump stop point; the effective operation data is mapped to a preset sampling step length by using a preset linear interpolation rule to determine a fixed-length feature sequence meeting a preset dimension; a final multivariate feature tensor is determined according to the fixed-length feature sequence by using a preset feature derivation calculation rule; and a full-sequence wellhead pressure prediction value of the target well is determined by using a preset pressure prediction model and the final multivariate feature tensor. Thus, full-sequence wellhead pressure prediction of the target well is realized, effectively solving the problems of insufficient prediction accuracy and poor sequence integrity in the prior art.

The application relates to the technical field of intelligent diagnosis, in particular to a power equipment installation and debugging fault intelligent diagnosis and process optimization method. The application realizes rapid positioning and accurate diagnosis of faults by comprehensively analyzing damage conditions of power equipment parts caused by mechanical, electrical and environmental factors in the installation and debugging process, quantifies mechanical deformation, current loss, temperature and humidity, magnetic field and other multi-physical field abnormalities, and constructs a cross-domain comprehensive damage evaluation model. Through dynamic threshold adjustment and historical data regression, the adaptability and early warning capability of diagnosis are improved. By adopting descending order sorting to preferentially check high-risk positions and combining average value comparison to publish targeted process optimization suggestions, the fault positioning efficiency is improved, and closed-loop management from fault repair to preventive process improvement is realized.

Forecasting resource allocation is disclosed. An example method includes receiving operating data from a resource; applying periodicity tests to the received operating data using a plurality of sketches of time series of prior operating data, the periodicity tests generating periodic components; applying regression models to the received operating data, the regression models collectively generating a trend component, each regression model being applied over a different time scale of a plurality of time scales; computing a trend model using the periodic components and a trend component; determining a random process describing the historical evolution of the trend model; and calculating and providing a mean prediction, an upper bound, and a lower bound for resource utilization at a future time using the trend model and a predicted distribution.