31 results about "Generalized additive model" patented technology

A method and an apparatus for improved duration modeling of phonemes in a speech synthesis system are provided. According to one aspect, text is received into a processor of a speech synthesis system. The received text is processed using a sum-of-products phoneme duration model that is used in either the formant method or the concatenative method of speech generation. The phoneme duration model, which is used along with a phoneme pitch model, is produced by developing a non-exponential functional transformation form for use with a generalized additive model. The non-exponential functional transformation form comprises a root sinusoidal transformation that is controlled in response to a minimum phoneme duration and a maximum phoneme duration. The minimum and maximum phoneme durations are observed in training data. The received text is processed by specifying at least one of a number of contextual factors for the generalized additive model. An inverse of the non-exponential functional transformation is applied to duration observations, or training data. Coefficients are generated for use with the generalized additive model. The generalized additive model comprising the coefficients is applied to at least one phoneme of the received text resulting in the generation of at least one phoneme having a duration. An acoustic sequence is generated comprising speech signals that are representative of the received text.

Systems and methods for analyzing power plant data include accessing data for one or more monitored parameters, generating a continuous time-series profile model (e.g., by generalized additive model and / or principal curves techniques) of selected parameters during instances of at least one given type of power plant operation, and conducting shape analysis by comparing the continuous time-series profile model of selected monitored parameters to an entitlement curve representing ideal performance for the at least one given type of power plant operation. Data associated with the profile modeling and shape analysis techniques may be provided as electronic output to a user. Additional steps or features may involve determining a capability index for selected monitored parameters, clustering together identified groups of profile models having similar profiles, monitoring and / or optimizing parameters to identify and improve power plant performance.

A diagnostic system in which, at every point in time, a forecast is made of the future response times of each EPP (End-to-end Probe Platform) probe in a battery of probes. Thresholds are established in terms of the distribution of future EPP values. The theory of Generalized Additive Models is used to build a predictive model based on a combination of a) data normally generated by network nodes, b) results of a battery of probes and c) profile curves reflecting expected response times (i.e. based on recent history) corresponding to this battery for various times of day, days of week, month of year, etc. The model is pre-computed, and does not have to be dynamically adjusted. The model produces, at regular intervals, forecasts for outcomes of various EPP probes for various horizons of interest; also, it produces thresholds for the respective forecasts based on a number of factors, including acceptable rate of false alarms, forecast variance and EPP values that are expected based on the recorded history. The system is capable of maintaining a pre-specified low rate of false alarms that could otherwise cause a substantial disturbance in network operation.

A diagnostic system in which, at every point in time, a forecast is made of the future response times of each EPP (End-to-end Probe Platform) probe in a battery of probes. Thresholds are established in terms of the distribution of future EPP values. The theory of Generalized Additive Models is used to build a predictive model based on a combination of a) data normally generated by network nodes, b) results of a battery of probes and c) profile curves reflecting expected response times (i.e. based on recent history) corresponding to this battery for various times of day, days of week, month of year, etc. The model is pre-computed, and does not have to be dynamically adjusted. The model produces, at regular intervals, forecasts for outcomes of various EPP probes for various horizons of interest; also, it produces thresholds for the respective forecasts based on a number of factors, including acceptable rate of false alarms, forecast variance and EPP values that are expected based on the recorded history. The system is capable of maintaining a pre-specified low rate of false alarms that could otherwise cause a substantial disturbance in network operation.

The present invention provides a microalloyed steel mechanical property prediction method based on globally additive model, including the following steps: determining some influencing factors of the microalloyed steel mechanical property prediction model; calculating the components and contents of carbonitride precipitation in a microalloyed steel rolling process; expressing the microalloyed steel mechanical property prediction model as an additive form of several submodels according to generalized additive model; estimating the microalloyed steel mechanical property prediction model; and verifying reliability of the submodels. The microalloyed steel property prediction models obtained in the foregoing solution have advantages such as high prediction precision and a wide adaptation range, and may be used for design of new products and steel grade component optimization, so as to reduce the quantity of physical tests, shorten the product research and development cycle, and reduce costs.

The invention relates to a big data-based time-space confusion exposure degree assessment system and method. The system comprises a time-space data mining block, a multi-source heterogeneous data fusion module, a final variable selection module, a time-space generalized additive model building module, a re-sampling model module, a variation function time-space modeling module and a concentration estimation module; massive time-space data is mined; a relationship between multiple influence factors and pollutant concentration is established by adopting an accumulative nonlinear method; and through residual variation function fitting, spatial autocorrelation is considered, so that the prediction precision and effect are greatly improved.

The invention relates to a construction method of the quick forecasting operation system of fishery fishing condition. The construction method comprises the following steps: adopting a satellite remote sensing system, a geographic information system and a fishery environment feature library to establish a multi-granularity fishery industry fishing condition analysis database; adopting a generalized additive model to carry out quantitative analysis on relationship between the fishery and a plurality of environment factors and fishing factors, definitizing the center fishery space-time change rule and the influence factors thereof of a long time sequence, and mastering the center fishery space-time change rule and a formation mechanism; and according to the fishery space-time change rule and the formation mechanism, constructing a fishery forecasting model based on a key marine environment factor driving mechanism, establishing various fishery forecasting models to realize the comprehensive forecast of a fishery, and proposing a fishery forecasting accuracy improvement scheme. Since a control is used for carrying out visual representation to main environment factors, a series of standard business-oriented flows is constructed to determine the analysis, the charting and the quick release of a fishery fishing condition forecast map, and reference and guide are provided for fishery quest for direct scene fishery industry production.

The invention discloses a hot rolling band steel deformation resistance prediction method based on a generalized additive model. The hot rolling band steel deformation resistance prediction method comprises the following steps of 1, variable pre-analysis: determining the form of a connection function and a model; determining a model attributive variable and an independent variable; 2, model setting: determining the basic form of the model; selecting a connection function according to the distribution of the attributive variable; determining the function form of each independent variable; 3, model estimation: estimating the connection function and the smooth function to obtain different models; selecting an optimum module according to the GCV (gross calorific value); 4, model result and evaluation: obtaining the estimation value of the model parameter part and the non-parameter part; performing evaluation and analysis on the fitting result of the smooth function; 5, model correction: combining the statistics analysis method and rolling theory, verifying the model result at different angles; performing model correction when being required. The method has the advantages that the deformation resistance of the hot rolling band steel under different rolling work conditions can be predicted; the basis is provided for the high-precision rolling force calculation and thickness control in the band steel rolling process.

The invention discloses an individualized living yield prediction model and a construction method. The construction method comprises the following steps: a, collecting living yield outcome, individualized ovarian function information, individualized biological information and individualized ovarian excretion promoting information of each sample patient; b, extracting parameter variables accordingto the accumulated live birth outcome and the individualized information; c, constructing a semi-parameter generalized additive model gm2 of the parameter variable; d, performing verification and sensitivity analysis on the model gm2, and training and evaluating the stability of the model gm2; and e, drawing a living yield prediction model curve graph. The method has the advantages that guidance can be provided for doctors and patients receiving ART treatment, the possibility of obtaining at least one live delivery infant can be predicted according to the age of the patients and the egg takingnumber, and the patients can be informed whether the live delivery possibility of the patients is remarkably increased or not after the patients receive extra stimulation periods.

The invention discloses a hand-foot-and-mouth disease epidemic situation prediction method based on case, weather and aetiology monitoring data. The prediction method includes: analyzing the relationship of the hand-foot-and-mouth disease cases with weather and aetiology factors and the hysteresis effect, and screening the indexes included in a model; on the basis of multi-source data of the hand-foot-and-mouth disease case, weather and aetiology, constructing a hand-foot-and-mouth disease model through a time series generalized additive model; dividing the multi-source data into a training data set and a verifying data set, thus evaluating the fitting situation and prediction result of the hand-foot-and-mouth disease epidemic situation prediction model. By combining the case, weather, aetiology and population data, a prediction model is constructed by using the time series generalized additive model. The data then is separated into different data sets to train and verify the fitting situation and prediction result of the model. Therefore, foresight prediction and risk warning on the trend of the hand-foot-and-mouth disease are conducted in real time. The method is more reliable inprediction result and is higher in timeliness and practicability.

The invention provides an in-pipeline corrosion growth prediction method and device based on a generalized additive model. The method comprises the steps: screening independent variables, obtained inadvance, of an in-pipeline corrosion growth prediction model through a Lasso compression estimation algorithm; based on a generalized linear additive model, establishing an in-pipeline corrosion growth prediction model according to the screened independent variables and the in-pipeline corrosion growth rate; and predicting the in-pipeline corrosion growth rate according to the in-pipeline corrosion growth prediction model. According to the method, modeling can be carried out on the basis of existing corrosion data in the pipeline, and the future corrosion depth is accurately estimated. Therefore, the internal detection period can be determined, a maintenance plan can be made, and safe pipeline operation and cost saving are facilitated.

The invention discloses a tuna habitat quantitative evaluation method. The method comprises the following steps: (1) acquiring a fixed influence factor of a target sea area; dispersing the target seaarea into a plurality of space-time units, and obtaining the unit fishing effort fishing yield, the random effect factor and the marine environment variable of each space-time unit; (2) adopting a generalized hybrid effect model with an error term in Gaussian distribution, and standardizing the fishing yield of unit fishing effort according to a fixed influence factor and a random effect factor; (S3) obtaining a habitat model by adopting a generalized additive model of logarithmic normal error distribution according to the marine environment variable of each space-time unit and the standard unit fishing effort fishing yield; and (S4) calculating a comprehensive habitat index according to the habitat model. According to the method, the two models are combined and applied, so that the environment can be associated with the actual abundance instead of the nominal CPUE, and the acquired habitat characteristic deviation is smaller.

The invention discloses a layered water temperature-based fishery resource abundance prediction method and application thereof, and the method comprises the steps: building a plurality of CPUE prediction models through employing a generalized additive model for the production statistical data of fishes C in a sea area B in a time period A, the longitude and latitude data of the sea area B in the time period A and the sea surface temperatures corresponding to different depths; analyzing each CPUE prediction model based on a red pool information criterion, and selecting the model with the minimum AIC value as an optimal prediction model; inputting the sea surface temperature and latitude and longitude corresponding to the water layer depth of the sea area B fish C in the to-be-predicted timeperiod into the final prediction model to complete prediction of the CPUE of the sea area B fish C in the to-be-predicted time period. The fishery resource abundance is predicted on the basis of thelayered water temperature, the model is established on the basis of the GAM model, the prediction precision is greatly improved, reliable guidance can be provided for development of ocean fishery resources, and the method has great application prospects.

The invention discloses a lake suitable ecological water level determination method based on a paleo-marsh method, and belongs to the technical field of environmental protection. The method comprises the following steps: firstly, collecting a sediment columnar sample in a lake area with less man-made interference, and performing equidistant slicing and layering on the sediment columnar sample for pretreatment; secondly, determining the age of each sample layer by using 210Pb and 137Cs dating technologies; then, identifying and calculating the variety and quantity of sporopollen of each sample layer to represent aquatic plant information of an ecological system in a time sequence, collecting and measuring corresponding hydro meteorology and physical and chemical indexes, and constructing an environmental factor data set; and finally, establishing a generalized additive model according to the key environmental factors suggested by the classification regression tree model, quantifying the response relationship between the environmental factors and the sporopollen concentration, and determining a suitable lake ecological water level. According to the method, the problem that stable long-time sequence biological data is difficult to obtain in an existing design is effectively solved, and a quantitative method for exploring the response relation between water level fluctuation and local ecological indexes is provided.

Systems and methods for use a multi-model block capacity forecasting approach are provided to predict when a distributed storage system will reach a fullness threshold. According to one embodiment, given a time series telemetry dataset collected from multiple distributed storage systems, a forecasting algorithm trains multiple time series forecasting models (e.g., Simple linear regression (SLR), Autoregressive Integrated Moving Average (ARIMA), Generalized additive model (GAM), and / or others) for each of the distributed storage systems. The best performing time series forecasting model is then independently selected for each of the distributed storage systems based on a respective performance metric (e.g., root mean squared error) associated with the time series forecasting models. Forecasted data points for each distributed storage system and the corresponding future time frames in which one or more predetermined or configurable block capacity fullness thresholds are predicted to be crossed may be determined based on the selected time series forecasting models.

The invention discloses a heating load prediction method. The method comprises the steps of obtaining a historical data set influencing a heating load and processing the historical data set to obtain a training data set; building a heating load prediction model based on the generalized additive model; training the constructed heating load prediction model by adopting the training data set and obtaining an air temperature coefficient; and calculating a final heating load prediction value according to the air temperature coefficient. According to the heating load prediction method provided by the invention, through innovatively establishing the prediction model and the prediction means, the prediction of the heating load is realized, and the method is high in reliability, high in accuracy and wide in application range.

The invention discloses a treatment method based on lake nutritive salt classification, which is based on background factors, total nitrogen concentration, total phosphorus concentration and chlorophyll a concentration of different lakes in summer, uses a simple linear regression model to analyze classification necessity, and performs unified nutritive salt upper limit treatment on lakes which donot need to be classified; for on lakes which need to be classified, the method sequentially identifies the influence of each lake background factor on the relationship between the chlorophyll a concentration and the total nitrogen concentration and the influence of each lake background factor on the relationship between the chlorophyll a concentration and the total phosphorus concentration through a 95% quantile regression model and a generalized additive model, classifies based on the identified significant background factors, and formulates various lake type nutritive salt treatment schemesthrough a simple linear regression model. The method provided by the invention has sufficient ecological basis, the classification result is objective, the management efficiency can be improved, andthe economic cost is saved.

The invention discloses a method and a system for quantifying an atmospheric ozone pollution source based on a domestic hyper-spectral satellite, which are used for analyzing influence factors of atmospheric ozone pollution based on a generalized additive model, so that the influence process of different factors on ozone concentration change can be known, and the relative contribution of meteorological conditions and control measures can be quantified.

The invention relates to the field of fishery, and discloses a construction method for a sthenoteuthis oualaniensis cutin jaw growth model based on marine environment factors. The method comprises the following steps of 1) performing principal components to select parameters that can represent the cutin jaw morphology best; 2) calculating the growth rate of the sthenoteuthis oualaniensis during the first sexual maturity, wherein GR = L / d; 3) calculating the operation sea area SST of each month within a certain time span; 4) carrying out canonical correlation analysis on the operation sea area SST of each month, the maximum correlation month of the sthenoteuthis oualaniensis in the corresponding year and the corresponding sea area position; 5) selecting the SST and GR in the month with the maximum correlation to carry out generalized additive model analysis, wherein after log conversion is carried out on the sthenoteuthis oualaniensis cutin jaw growth rate, the sthenoteuthis oualaniensis cutin jaw growth rate accords with normal distribution, and a GAM result is optimized by means of AIC; and 6) obtaining a relational expression between the SST and the GR to establish the model. The model constructed through the method can be used for analyzing and mastering the cutin jaw growth rule of the sthenoteuthis oualaniensis under the influence of different marine environments, and reference is provided for follow-up sustainable development of the sthenoteuthis oualaniensis.

The invention discloses a correlation analysis method for key parameters and environmental factors of an icing growth model of a power transmission line. The method defines a collision coefficient anda freezing coefficient of conductor icing as the key parameters of the icing growth model, and comprises steps of selecting terminal data, and determining an independent variable and a dependent variable; environment factors (wind speed, wind direction, environment temperature and air liquid water content) being taken as independent variables, and a collision coefficient and a freezing coefficient being taken as dependent variables; respectively establishing generalized additive models GAM of the collision coefficient, the freezing coefficient and the environmental factors; solving a generalized additive model of the collision coefficient and the freezing coefficient, measuring fitting superiority of the statistical model through a red pool information criterion, and selecting an optimalgeneralized additive model; and analyzing the correlation between the key parameters of the icing growth model and the environmental factors according to the result of the GAM model and the statistical parameters of the GAM model. According to the method, the mathematical relationship between the icing growth key parameters and a plurality of environmental factors can be determined, and the nonlinear effect of the environmental factors is reflected.

The invention discloses a bank financial product recommendation method based on a generalized additive model combined with matrix decomposition, relates to the technical field of data processing in abank financial product management system, and solves the technical problem that an existing bank financial product recommendation method has no interpretability or is poor in precision, and the bank financial product recommendation method comprises the steps of 1, performing feature extraction processing on data; dividing the original input into user information, product information and behavior information; 2, respectively establishing generalized additive models for the user information and the product information; 3, performing matrix decomposition fitting on the user behavior information for the residual part; 4, repeating the steps 2 to 3 by adopting an iterative method until the model converges; and step 5, obtaining a final prediction result by combining the generalized additive model and the matrix decomposition model. On the premise of ensuring high recommendation precision, various model explanations of a linear part and a matrix decomposition part can be given. The problem of maximum cold start of an existing recommendation system is solved. Therefore, the recommendation credibility of the user is improved, and the recommendation quality is improved.

The invention belongs to the field of ecological protection and provides a method for controlling and measuring zizania latifolia biomass. The specific control method comprises the following steps: regulating a water depth or water level of a zizania latifolia growth environment during tillering of the zizania latifolia, namely controlling the water depth to be at least 1m in March to April every year, preferably controlling the water depth to be at least 1.5m in early March; and performing biomass cutting assisted control when the tiller number of the zizania latifolia starts to decrease and above-ground biomass starts to decline, wherein the relationship between the zizania latifolia biomass and the water depth or the water level accords with a GAM (Generalized Additive Model). According to the method disclosed by the invention, the biomass of the zizania latifolia is controlled through the water level of the growth environment in March to April every year according to growth habits of the zizania latifolia, and the problems that the traditional physical method is huge in cleaning workload and the ecological environment is damaged by herbicides are solved.

The embodiment of the invention discloses a wind power prediction method and device, equipment and a storage medium. The method comprises the steps of collecting sample data, and performing feature extraction on the sample data to obtain a model input feature data set; performing clustering analysis on the model input feature data set by adopting a k-means clustering algorithm, and adding a data category obtained by clustering into the model input feature data set as a new feature; training a preset generalized additive model according to the model input feature data set to obtain a wind power prediction model; and inputting to-be-tested data into the wind power prediction model to obtain a data category of the to-be-tested data and a prediction result of the wind power. According to the technical scheme provided by the embodiment of the invention, the influence of each variable on the wind power is explained through the generalized additive model, and the precision of wind power prediction is effectively improved in combination with clustering analysis.

The invention provides a method for predicting the mechanical properties of microalloy steel based on a global additive model, comprising the following steps: determining the influencing factors of the mechanical property prediction model; calculating the composition and content of carbonitride precipitation in the rolling process of the microalloy steel; Based on additive model theory, the mechanical property prediction model of microalloyed steel is expressed as an additive form of several sub-models; the mechanical property prediction model of microalloyed steel is estimated; and the reliability of each sub-model is verified. The microalloy steel product performance prediction model obtained by the above scheme has the advantages of high calculation accuracy and wide application range, and can be used for new product design and steel composition optimization to achieve the purpose of reducing the number of physical tests, shortening the product development cycle, and reducing costs.

The embodiment of the invention discloses a network fraud influence factor analysis method, equipment and a storage medium, and relates to the technical field of network fraud analysis. The method comprises the steps: acquiring a plurality of influence factors of network fraud; based on the statistical data of the plurality of influence factors and the statistical data of the number of network fraud cases, selecting an optimal independent variable combination meeting an akaike information criterion from the plurality of influence factors by adopting a stepwise regression method, wherein the optimal independent variable combination comprises at least one influence factor; and based on the statistical data of the at least one influence factor and the statistical data of the network fraud case quantity, constructing a generalized additive model which takes the at least one influence factor as an independent variable and takes the network fraud case quantity as a dependent variable; wherein the generalized additive model is used for analyzing the influence of each influence factor on the network fraud. According to the embodiment, the comprehensiveness of network fraud influence factor analysis can be expanded, and the accuracy of network fraud influence factor analysis is improved.

The invention provides a natural reserve evaluation method based on human footprint distribution and natural factors. The natural reserve evaluation method comprises the following steps: step 1, acquiring data of a natural reserve, wherein the data comprises human footprint index data and basic geographic data, the human footprint index data comprises population density data, land utilization typedata, domestic production total value data and road traffic accessibility data, the basic geographic data comprises natural environment data; step 2, obtaining a human footprint index according to the human footprint index data; step 3, constructing a generalized additive model according to the human footprint index and the natural environment data; and step 4, evaluating the natural reserve according to the human footprint index and an analysis result of the generalized additive model. According to the method, the natural reserve can be evaluated more accurately, and accurate guidance is provided for subsequent work.