122 results about "Data regression" patented technology

A method and system for checking health of information technology infrastructure based on log data, in one aspect, collect log data non-intrusively from a production system, said log data at least associated with transactions occurring in the production system and resource utilization of the production system, normalize said log data into a plurality of log data types, perform data regression analysis using said plurality of log data types to estimate resources consumed by each of said transactions and throughput of each of said transactions, and use a queuing model to predict performance of the information technology infrastructure under various workloads.

System and method for optimization of a design associated with a response function, using a hybrid neural net and support vector machine (NN / SVM) analysis to minimize or maximize an objective function, optionally subject to one or more constraints. As a first example, the NN / SVM analysis is applied iteratively to design of an aerodynamic component, such as an airfoil shape, where the objective function measures deviation from a target pressure distribution on the perimeter of the aerodynamic component. As a second example, the NN / SVM analysis is applied to data classification of a sequence of data points in a multidimensional space. The NN / SVM analysis is also applied to data regression.

The invention relates to a method for establishing damaged rock constitutive relation by utilizing residual strength. The method comprises the following steps of: carrying out a rock indoor triaxial compression experiment, measuring the axle load sigma1, the confining pressure sigma 3, the peak value intensity strain Epsilon 1c and the residual intensity sigma r of a rock sample, calculating the rock elastic modulus E and poisson ratio Mu, carrying out data regression according to triaxial compression test results, and calculating the uniaxial compressive strength sigma ci and an experience intensity parameter mi of a rock block; then establishing a three-dimensional damage statistic constitutive relation capable of reflecting a rock post-peak softening characteristic; and finally, solving constitutive relation parameters n and F0 according to four boundary conditions of a rock full-stress-strain curve geometric characteristic, and drawing a constitutive relation curve by utilizing Matlab software. The constitutive relation can fully reflect rock post-peak softening and residual intensity characteristics, and can describe the full-stress-strain relation of the rock damage process well, so that a theory formula well corresponds to the actual condition of a rock material; and the method can be realized simply, is widely applied to the theory analysis of the rock material, and has an actual value.

The invention discloses an aero-engine thrust estimation algorithm through an adaptive RBF neural network. According to the algorithm, an improved particle swarm optimization algorithm is utilized tooptimize centers, width, connection weights and other neural network parameters of all nodes of the radial basis function (RBF) neural network, and meanwhile the network scale is optimized, so that the neural network is compacter under the condition of meeting a precision requirement. The algorithm can be used for medium and small-scale data regression and can be applied to estimation of thrust and other parameters in terms of aero-engine. Through the algorithm, the adaptive RBF neural network is provided based on the particle swarm optimization algorithm; in the improved particle swarm optimization algorithm, according to different network hidden layer node numbers, locally optimal solutions in the same number as types of the hidden layer node numbers are set; and the algorithm provides anew thought for aero-engine thrust estimation and is easy to understand, simple in parameter adjustment, easy to realize, high in applicability and capable of realizing high-precision thrust estimation.

The embodiments of the invention disclose a power sale quantity prediction method and device based on X13 seasonal adjustment and factor regression. The method comprises the steps of preprocessing historic power sale quantity data, and decomposing the preprocessed power sale quantity sequence into a trend item, a season item and a random item by using an X13 seasonal adjustment algorithm; respectively performing prediction by adopting multiple prediction algorithms according to the influence factors and curve characteristics of sub sequences to ensure the prediction precision and robustness of the trend item; summing the prediction results of the sub sequences to reconstruct power sale quantity prediction results, and finally selecting a prediction result having the optimal performance from the multiple prediction results. Meanwhile, the embodiments of the invention further sufficiently consider the influence of some influencing factors on each decomposition item, so the precision of the prediction result obtained by the solution of the embodiment is higher.

The invention relates to a method for tracking multi-target vehicles by adopting the MCMC (Markov Chain Monte Carlo) algorithm, comprising the following steps: modeling the multi-target vehicle tracking process by adopting the MCMC algorithm, establishing various preselection states, traversing the preselection states by using the Metropolis-Hastings sampling based simulated annealing algorithm, and selecting the preselection state with the maximum connection probability as the optimal solution. In the method, a limit is set to the transfer equivalence for the first time, and the parameters in the data association are estimated by the experimental data regression fitting, thus the optimal vehicle motion trajectory with maximum posteriori probability significance is obtained, and the vehicles are tracked. The invention solves the problems of frequent shielding splitting of the multi-target vehicles and has the advantages of high multi-target vehicle tracking precision and good real-time performance.

The invention discloses a comprehensive electric energy meter verification method and system based on an improved least square method. The method comprises the steps: generating a scatter diagram of original data, deleting an abnormal value, and obtaining sample data; carrying out Pearson correlation analysis and VIF inspection on independent variables in the sample data; determining a multi-colinearity existence range between the independent variables; checking the multiple collinearity by fitting a sample error average regression line and a median regression line; performing multivariate regression analysis according to an inspection result, and preliminarily determining a regression equation; checking the credibility of the regression equation, and determining a data regression model; correcting the data regression model through residual analysis; and normalizing the weight of each variable, calculating an influence weight of each variable on the error, and substituting the influence weight into the data regression model to carry out comprehensive verification on the electric energy meter. According to the invention, whether the metering error of the electric energy metering device exceeds a standard specified range can be effectively verified, and the reliability and stability of the electric energy metering device are ensured.

A method for building a fine grain tailing project property index estimation empirical formula based on linear regression includes the following steps: 1) collecting and settling fine grain tailing data physico-mechanical index sample data; 2) adopting the mu+ / -3sigma abnormal value rejection principle to conduct preliminary screening on the sample data; 3) conducting secondary screening on the sample data based on mathematical statistics; 4) building a linear regression mathematical model; 5) determining model parameters; 6) conducting model goodness inspection; 7) building the fine grain tailing project property index estimation empirical formula. The method has the advantages of providing forceful support for scientificity and reliability of a data regression analysis result through massive test data information, providing reliable parameter choice for fine grain tailing base stability evaluation and design in future, greatly saving project investment cost and soil engineering test investment, avoiding unnecessary project building cost, enabling a research result to have popularization and application value and being simple, practical, reliable in result, efficient in computing and the like.

The invention discloses a numerical control system reliability rapid Bayes evaluation system under degradation tests. According to the failure data small sample characteristic of a high-reliability long-service-life numerical control system, a Bayes evaluation scheme based on the double stress stepping accelerated degradation tests is adopted to achieve rapid secondary acceleration evaluation effects. A double stress acceleration model is established on the basis of a temperature-humidity double stress stepping accelerated service life degradation test scheme; according to the conditions that test data are blended data composed of complete data and censored data, a multiple blended data regression method is adopted to estimate acceleration model parameters; according to the conditions that numerical control system failure modes are not single, a competition failure reliability evaluation model is established; and according to the failure data small sample characteristic, a Bayes method is utilized to solve prior distribution by means of historical information. According to the numerical control system reliability rapid Bayes evaluation system, the model parameters are evaluated by combination with a stepping accelerated test competition failure model, evaluation precision is improved, numerical control system reliability rapid evaluation theories are enriched, and a theory basis is provided for verifying a numerical control system failure mechanism and improving numerical control system reliability.

The invention relates to the field of ship performance testing and provides a ship manoeuvrability hydrodynamic testing device and method. The ship manoeuvrability hydrodynamic testing device comprises a ship model, a data acquisition module and a data processing module. The data acquisition module is mounted on the ship model, and the data processing module is connected with the data acquisitionmodule. The data acquisition module comprises two ship hull force sensors, a three-component rudder force sensor and a paddle dynamometer, and the two ship hull force sensors, the three-component rudder force sensor and the paddle dynamometer are used for acquiring ship hull hydrodynamic data, rudder hydrodynamic data and paddle hydrodynamic data respectively. The data processing module receives the hydrodynamic data acquired by the data acquisition module and performs data regression analysis on the hydrodynamic data to obtain a hydrodynamic derivative. The ship manoeuvrability hydrodynamic testing device and the method have advantages that accuracy in ship manoeuvrability hydrodynamic testing is improved, testing time and testing cost are saved, and simple structure, high operability anda promising application prospect are realized.

The invention discloses a ring main unit cable core temperature soft measurement method based on a neighborhood preserving embedded regression algorithm. The ring main unit cable core temperature soft measurement method comprises that firstly, based on the local feature extraction strategy of the neighborhood preserving embedded algorithm, a regression optimization function which takes the internal temperature and the internal humidity of a ring main unit, the cable core current and the cable surface temperature as input, and takes the cable core temperature of a cable in the ring main unit as output is established, local features of input data and output data are reserved, and the maximum relationship between data is obtained; then based on lower-dimension latent variables of data, input and output features which construct data regression are obtained; and a cable core temperature soft measurement model is established. The ring main unit cable core temperature soft measurement method is advantaged in that by means of a data local feature extraction method, a traditional neighborhood preserving embedded algorithm is modified to be a regression model, and key variable information, of the ring main unit, which cannot be measured easily is obtained. According to the invention, the problem that the temperature of the cable core in the ring main unit cannot be measured easily is solved, and the accuracy and operability of on-line monitoring and fault locating of the ring main unit are improved.

The invention provides a method for detecting the activity of a catalyst of a selective catalytic reduction (SCR) denitration system, and a system thereof. The method comprises the following steps: sampling the catalyst of the SCR denitration system to obtain a catalyst sample, and detecting the activity of the catalyst sample to obtain the apparent activity of the catalyst; performing an on-site performance test on the SCR denitration system to obtain on-site performance test data; and substituting the apparent activity, which serves as the relative proportion of the activity of each layer of catalyst, of the catalyst into a reactor performance calculation model of the SCR denitration system and performing on-site performance test data regression to obtain the actual activity of the catalyst. Through the method, the activity of the catalyst on the operation site of the SCR denitration system can be measured accurately, and the operation control accuracy of the SCR denitration system is improved.

The invention discloses an anti-collision early-warning method based on pedestrians and riders in front of a road, and belongs to the field of driving auxiliary systems. The anti-collision early-warning method comprises three aspects of environment perception, information interpretation and target state judgment. According to the anti-collision early-warning method, a collected training set is transmitted to a YOLO-R network to be trained, target detection and classification are conducted, and multi-target tracking is achieved through Kalman filtering; an image is subjected to inverse perspective transformation, thus an IPM image is obtained, a relation curve between original image pixel coordinates and IPM image pixel coordinates is fitted through data regression modeling, and the distance is estimated through a linear relation between the IPM image pixel coordinates and world coordinates; and according to the vehicle speed and the braking distance, an early-warning activation area isdetermined, then a fuzzy early-warning algorithm is adopted for targets in the activation area so as to judge whether the targets in the activation area are in danger or not, if the danger exists, drivers are reminded in time, accordingly, accidents are effectively reduced, and the safety of the pedestrians and the riders is protected.

A pattern recognition method induces ensembles of decision rules from data regression problems. Instead of direct prediction of a continuous output variable, the method discretizes the variable by k-means clustering and solves the resultant classification problem. Predictions on new examples are made by averaging the mean values of classes with votes that are close in number to the most likely class.

A method for representing single-crystal Ni-base alloy creep resistance comprises the steps of building a creep curve model suitable for single-crystal Ni-base alloy creep characteristics; determining model parameters to obtain a creep curve equation; performing creep data regression to obtain a fitting creep curve for describing a creep process; obtaining a creep rate equation, obtaining the minimum calculation creep rate and calculating a creep rate curve; combining specific requirements for representing the single-crystal Ni-base alloy creep resistance.Various single-crystal Ni-base alloy creep curves under constant temperature and constant stress except the extremely-high temperature or extremely-high stress and creep rate distribution can be accurately expressed; the corresponding relations of various creep curve equation items and creep parameters and alloy creep curves is revealed; the problem that some alloy cannot represent creep characteristics easily through the creep activated energy and stress indexes is solved; and the method is of great importance in further knowing the creep characteristics and rules of single-crystal Ni-base alloy.

The invention discloses a purchasing power prediction method and a purchasing power prediction device, and belongs to the field of data regression prediction. The purchasing power prediction method comprises the following steps: acquiring a purchasing power prediction model, wherein the purchasing power prediction model is a model trained according to the purchasing factor of a sample user in a first historical time period, and the purchasing power prediction model includes the corresponding relation between the predicted purchasing power and the purchasing factor; acquiring the historical purchasing information of a target user in a second historical time period, wherein the historical purchasing information includes the target purchasing factor of the target user in the second historical time period; and predicting the target purchasing power of the target user in a target time period in the future according to the historical purchasing information and the purchasing power prediction model. The problem in the prior art that the accuracy of purchasing power prediction is low is solved. The accuracy of predicted purchasing power is improved.

The invention provides a tower pole icing disaster risk prediction method based on the safety margin. With the method, finite element numerical value simulation and data regression analysis are combined; a finite element method is adopted to build a 1:1 overhead transmission line tower pole coupling model; the main stressed factor of a line under various static conditions is comprehensively considered; simulation calculation is carried out to various operation working conditions of an icing line model; and on the basis, a simulation result is subjected to regression analysis and data fitting by the concept of the safety margin to obtain the safety margin curve of a power transmission tower line system so as to obtain the safety level of the power transmission tower line system. The method disclosed by the invention belongs to the safety evaluation field of the power transmission tower line system, the icing and tower falling risk of the power transmission tower line system can be solved, the safety margin range of the tower line system is given in real time so as to determine the line icing risk level at present, a basis is given for formulating an icing-proof transformation scheme, the formulation principle of the icing-proof transformation scheme is determined, and the operation reliability of the power transmission tower line system is improved.

A method for calibration of digital image capture devices is presented. This simplified method provides a calibration based on the human visual perception of the colors input into the device using simple test targets, measurement devices, and software with a minimum of labor and expertise. This analysis may be performed using the data analysis tools of a conventional electronic spreadsheet. The method normalizes the test target data to both black and white, and converts the normalized data into the color space of the capture device through white point adaptation. The raw captured image data is also normalized to both black and white, and is regressed with the converted, normalized target data to determine the expected measurement values. These values are used to compensate the device output to achieve a high level of color fidelity. To ensure that the level of fidelity is acceptable, the CIE color difference equations are used.

The invention relates to a Parkinson's disease auxiliary diagnosis system based on a multi-modal magnetic resonance brain image. The auxiliary diagnosis system comprises an input module, a feature screening module, a feature selection module and a diagnosis module. The input module comprises inputs of a T1WI image, a DTI image and a QSM image. The feature screening module is used for carrying outimage preprocessing on the three groups of images, extracting feature data of a region of interest, connecting the feature data of different modes in series to form a series feature matrix X, and connecting MOCA and sample labels in series to form a corresponding matrix Y. The feature selection module is used for extracting features with high representativeness, and the diagnosis module is used for carrying out learning, data regression and classification on the features and finally obtaining a diagnosis result, so that more accurate auxiliary diagnosis is provided for doctors.

The invention relates to a method for calibrating particle size analysis data by using a laser method and a sieve analysis method, and belongs to the technical field of particle size analysis. The method comprises the following steps of: calculating sieve-analysis-method mid-values and laser-method mid-values of various narrow particle size distribution range samples which are selected from representative samples by using a sieve analysis method, and performing data regression analysis on the obtained sieve-analysis-method mid-values and the obtained laser-method mid-values so as to obtain a calibration formula; and finally, calibrating laser-method particle size analysis results of the samples into sieve-analysis-method particle size analysis results by using the calibrating formula, or calibrating the sieve-analysis-method particle size analysis results into the laser-method particle size analysis results by using the calibrating formula. According to obvious differences between data analyzed by the laser method and data analyzed by other methods, the calibrating formula is established on the basis of contrastive analysis of a large quantity of experimental data, and laser-method and sieve-analysis-method particle size analysis data are reasonably calibrated during data comparison and data docking in principle. The method can be used for calibrating data acquired by other particle size analysis methods.

The invention relates to a method for predicting the life of an SCR denitration catalyst in a power plant. The method comprises the following steps: step 1, on the basis of analyzing the three most important factors in deactivation of the SCR denitration catalyst, establishing a deactivation model of the SCR denitration catalyst, wherein the three factors include chemical poisoning, porous channelblockage and fly ash abrasion; step 2, calculating an activity value and an activity decreasing trend in an operation process of the SCR denitration catalyst; step 3, according to acquired activity data, combining the deactivation model to obtain a relational expression between the catalyst activity and the operation time through a previous actual data regression fitting deactivation model; and step 4, acquiring the activity value of the catalyst in the future according to the expression, and comparing the activity value with preset activity values that cannot meet denitration tasks to judgethe remaining available time of the catalyst. According to the method disclosed by the invention, the activity values of the SCR denitration catalyst at different times can be accurately predicted byusing operation data acquired on site in combination with the deactivation model, and the life of the catalyst can be judged.

The invention discloses an analysis method for steam turbine valve flow characteristics on the basis of historical data regression analysis. The historical operation data of a unit is taken as the input value of multiple regression analysis to calculate each parameter value in a multiple regression expression; in the obtained multiple regression expression, parameter values, including main steam pressure, the post pressure of each regulation valve and the like, under a corresponding single valve experiment working condition are input; the pressure change situation of a regulating stage chamberin a process that a target regulation valve is completely closed from completed opened is calculated; and a valve flow characteristic calculation method can be applied to obtain the flow characteristic curve of the target regulation valve. Compared with the prior art, the invention adopts the analysis method for steam turbine valve flow characteristics on the basis of the historical data regression analysis to research the practical valve flow characteristics of the steam turbine, a historical data driven calculation method for steam turbine valve flow characteristics is realized, and manpower and material resources can be saved.

The invention relates to a structural health monitoring data exception identification method based on a space-time diagram convolutional network, and solves the problem that an existing exception identification method based on building structure health monitoring data is difficult to distinguish sensor faults and structural variation. The identification method comprises: carrying out space-time correlation modeling on structure monitoring data by utilizing a space-time diagram convolutional network capable of learning an adjacent matrix, hierarchically applying information of adjacent nodes ofeach order to data regression, and designing a corresponding network structure and an objective function penalty term; and a monitoring system is used to establish initial measured data as a trainingset, a network is trained and an adjacent matrix is acquired, subsequent measured data is input into the network and then a model residual error and a diagnosis index are calculated, and the diagnosis index and a key adjacent edge are combined to determine whether data abnormity originates from a sensor fault or structural variation. Data modes of sensor abnormity and structure abnormity can be effectively distinguished, the fault sensor can be accurately identified, and the method is suitable for management and maintenance of various structure health monitoring systems.

The invention provides a curve overtaking risk analysis method for a semitrailer. The curve overtaking risk analysis method for the semitrailer comprises the following steps: simplifying vehicle lane changing track to be a reverse convex curve formed by an easement curve, obtaining a semitrailer linear road segment overtaking track model, calculating elements, such as curve length, and obtaining a vehicle simulation path; according to a lateral load drift rate (LTR), judging safe conditions of the vehicle, simulating by using a ultrahigh transverse slope, a running speed, a vehicle turning radius and a road circular curve radius, and respectively performing orthogonal experiment to the four elements; relating a side turn risk indicator with four elements of the ultrahigh transverse slope, the running speed, the vehicle turning radius and the road circular curve radius; and performing regression analysis by using data regression analysis software, and obtaining a forecasting model of vehicle curve overtaking risk. The curve overtaking risk analysis method for the semitrailer is capable of, through building the forecasting model of the vehicle curve overtaking risk and analyzing the risk of the semitrailer curve overtaking, ensuring the traffic safety of the semitrailer in the curve running.

The invention relates to a conversation connection and establishment method for roaming subscribers. After establishing an RP connection and completing a PPP link control and a PPP authentication, a visitor location PDSN obtains an IP address of a home location PDSN; a renegotiation procedure and the PPP authentication of the PPP link control are started through a PP connection and an access request authentication is processed on an access terminal; after passing the authentication, an IP address is distributed to the access terminal through a PPP IP address control procedure and the connection between the access terminal and the home location PDSN is established. The invention further relates to a conversation connection and establishment system for the roaming subscribers and a conversation connection release method for the roaming subscribers and a system thereof, wherein, the visitor location PDSN uses the PP connection to complete the process of connection establishment between the access terminal and the home location PDSN, thus using the PP connection to provide a transmission channel of business data so as to support the data regression for the roaming subscribers.

The invention discloses a regional highway main channel traffic demand forecasting method based on multiple-factor regression. The method comprises the following steps: determining influencing factors of regional highway main channel traffic demand forecasting; determining at least one selectable path which is same with the starting point and end point of a main channel path; constructing a multiple-factor regression model according to the determined influencing factors and the selectable path; and forecasting the traffic demand y1 of the main channel by adopting the established regression model. The method improves an existing regional highway main channel traffic demand forecasting method which carries out regression forecasting only on a single path without considering the influence of other adjacent paths, the trends of which are roughly same with the trend of the path; the traffic demand forecasting of some path in the channel is carried out by utilizing the multiple-factor united regression; and the method improves existing demand forecasting precision and can provide more accurate demand forecasting result for the follow-up jobs in engineering practice.

Embodiments of the present specification provide a data selection method and apparatus, wherein the method may include training a machine learning model based on input variables and tags in a trainingsample; The training sample also includes non-modular variables; Input the input variables of the test sample into the machine learning model to obtain the predicted value; The test sample also includes a label; According to the label and the predicted value of the test sample, the corresponding residual error of the test sample is obtained. Sending the residuals to at least two second data parties, respectively, so that each second data party regresses the residuals using the possessed second data, respectively, and obtains a regression evaluation index; The regression evaluation indicatorsreturned by the at least two second data parties are received to select a portion of the second data party by comparing the regression evaluation indicators of the at least two second data parties.