376 results about "Non linear model" patented technology

A system and method for historical database training of non-linear models for use in electronic commerce. The non-linear model is trained with training sets from a stream of electronic commerce data. The system detects availability of new training data, and constructs a training set from the corresponding input data. Over time, many training sets are presented to the non-linear model. When multiple presentations are needed to effectively train the non-linear model, a buffer of training sets is filled and updated as new training data becomes available. Once the buffer is full, a new training set bumps the oldest training set from the buffer. The training sets are presented one or more times each time a new training set is constructed. An historical database may be used to construct training sets for the non-linear model. The non-linear model may be trained retrospectively by searching the historical database and constructing training sets.

An efficient coding scheme with higher audio bandwidth and / or better audio quality at lower bitrates, wherein the scheme eliminates long-term and short-term frequency domain correlation in a signal via frequency domain predictors. The coding scheme compresses information consisting of coded low frequency components as well as a parametric representation for the high frequency components based on a non-linear model. Additionally, by working on the frequency domain representations of the signal (such as the MDCT representation which is naturally available to a PAC encoder and decoder), low pass and high pass signal components are easily obtained by windowing the appropriate ranges of frequencies in the signal. Furthermore, the power functions of the signal are replaced by corresponding convolution functions of the same order.

The invention relates to a composite amplifier (500) based on a main amplifier (510) and an auxiliary amplifier (520), and compensation for non-linear amplifier behavior by means of respective non-linear models (570, 575) of parasitic. In order to provide proper excitation of the non-linear models. a filter network (560) based on a linear model of the output network of the composite amplifier is provided. The linear filter network (550) basically determines ideal output node voltages, which are used as input to the respective non-linear models for generating appropriate compensation signals. The compensation signals are finally merged into the input signals of the respective sub-amplifiers (510, 520), thus effectively compensating for the effects of the non-linear parasitics.

A system and method for preprocessing input data to a non-linear model for use in electronic commerce (e-commerce). The non-linear model may include a set of parameters that define the representation of an e-commerce system. The non-linear model may operate in training or run-time mode. A data preprocessor may be provided for preprocessing received data in accordance with predetermined preprocessing parameters and outputting preprocessed data. The data preprocessor may include an input buffer for receiving and storing the input data, where the input data may be on different time scales. A time merge device may select a predetermined time scale and reconcile the input data so that all of the input data are placed on the same time scale. An output device may output the reconciled data from the time merge device as preprocessed data. The preprocessed data may then be used as input data to the non-linear model.

A computerized system for mapping an atmospheric phenomenon in a geographic region. Multiple free-space electromagnetic communications links are previously distributed in the geographic region. The system includes an interface to monitoring mechanisms attached respectively to the free-space electromagnetic communications links. The monitoring mechanisms respectively monitor attenuation levels of the free-space electromagnetic communications links. A processor simultaneously processes the attenuation levels, and maps in the geographic region the atmospheric phenomenon. The simultaneous processing preferably applies a non-linear model which relates the attenuation levels the atmospheric phenomenon, and solves a tomographic problem based on the non-linear model and the attenuation levels. The tomographic problem is preferably solved by an interactive algorithm based on consecutive refinement and linear inversion at each iteration. Alternatively, an interpolation is performed based on respective inverse distance from the communications links. Preferably, the interpolation is further based on respective lengths of communications links. A data interface preferably provides to subscribers temporal information related to the atmospheric phenomenon within portions of the geographic region.

A system and method for on-line training of a non-linear model for use in electronic commerce. The non-linear model is trained with training sets from a stream of process data. The system detects availability of new training data, and constructs a training set from the corresponding input data. Over time, many training sets are presented to the non-linear model. When multiple presentations are needed to effectively train the non-linear model, a buffer of training sets is filled and updated as new training data becomes available. Once the buffer is full, a new training set bumps the oldest training set from the buffer. The training sets are presented one or more times each time a new training set is constructed. An historical database may be used to construct training sets for the non-linear model. The non-linear model may be trained retrospectively by searching the historical database and constructing training sets.

The invention discloses a robust control method of burden error of pitch concrete stirring device, which comprises first building an input / output artificial neutral network model of three-layer structure, while the input layer has one neuron, the middle layer has five neurons and the output layer has one neuron, processing dynamic prediction and learning. The invention builds the non-linear model of flux and fly ash amount to be compared with a linear model of fixed parameters via continuous learning and adjustment, to adapt the change of external parameters, thereby effectively controlling the burden error of pitch mixture stirring device, reaching +-2% burden error as maximum and improving the burden accuracy of pitch mixture stirring device.

A method and system for process control using a model predictive controller. The control system can have one or more control devices operably coupled to a processing system for controlling a process of the processing system; a modeling tool to provide a non-linear model based at least in part on the process and to provide a plurality of linearized models based at least in part on the non-linear model, where the plurality of linearized models are linearized at different linearization rates; and a controller operably coupled to the modeling tool. The controller can select one of the plurality of linearized models based on a comparison of the plurality of linearized models with a reference model. The controller can send one or more control signals to at least one of the one or more control devices. The one or more control signals can be determined using the selected one of the plurality of linearized models.

The invention discloses a vertical take-off and landing airplane robust fault-tolerant control system and method based on cascaded observers, and belongs to the technical field of aircraft control. A vertical take-off and landing airplane dynamical model taking external disturbance and actuator faults into consideration is built, the dynamical model is converted into a non-linear model in a standard form, the disturbance, the faults and state variables are estimated in real time by using the disturbance observer and the fault diagnosis observer which is in cascade connection with the disturbance observer, the non-linear model is decoupled into two subsystems, the control rate of each subsystem is determined, and the fault-tolerant control rate is determined according to the control rates and fault estimation values of the two subsystems. Under the action of a fault-tolerant controller, the system taking effects of the external disturbance and the faults into consideration compensates for the faults and the disturbance in real time, influence of the faults and the disturbance is reduced rapidly, and robustness of the control system is improved.

The invention discloses a building method for the universal comprehensive models of single-rotor helicopters and turboshaft engines. The method includes the building of the non-linear models of a helicopter and an engine, the static trimming of the single-rotor helicopter and the turboshaft engine, the dynamic calculation of the single-rotor helicopter and the turboshaft engine and the actual operation solution of the single-rotor helicopter and the turboshaft engine, wherein the static trimming adopts the Levenberg-Marquarat algorithm, the dynamic calculation adopts the fourth-order Ronge-Kutta method to solve the motion equation system of the helicopter and the single pass algorithm to solve the dynamic balance equation of the turboshaft engine, and the actual operation state solution adopts a modified linear quadratic regulator algorithm to carry out helicopter solution and a cascade PID (Proportion-Integration-Differentiation) plus torque feedforward plus collective pitch compensation and torsional vibration filter method to carry out engine solution and load filtration. The method can meet the test requirements of the universal models and control methods of the single-rotor helicopters and the turboshaft engines, and plays an active role in helping to shorten the development cycle and reduce the test risk and the cost.

A system and method maximize profits by determining the optimal combination of sale price for a product and bids for advertising. The system and method address the non-linear relationship between product pricing and advertising through an advertising bidding system. In particular, a seller sells a product at a plurality of sale prices and buys the at least one advertisement by submitting a plurality of bid prices to the at least one advertising venue, resulting in a combination of bid prices for the at least one advertisement at each of the plurality of sale prices. Data from advertising with each of the at least one advertisement is collected. Then, a non-linear model for pricing and bidding is determined from the collected data, and an optimal price for the product and an optimal advertising bid for each of the at least one advertisement is determined from the non-linear model.

The invention discloses a nondestructive detection method of swelled kiwi fruits (kiwi fruits obtained by applying a swelling agent) based on a hyperspectral imaging technology. The method comprises the following steps: scanning a batch of intact swelled kiwi fruit samples and non-swelled kiwi fruit samples with a hyperspectral imaging system as a tool to obtain hyperspectral image information of the samples; carrying out black and white correction and masking treatment of hyperspectral images, and extracting the average spectrum of a whole fruit area; preprocessing the average spectrum; dividing to form a correction set and a prediction set; reducing dimensions of spectral data to extract characteristic wavelengths or characteristic variables reflecting the swelled kiwi fruits; and establishing a linear or non-linear model for the nondestructive identification of the swelled kiwi fruits, and examining the model. The correct recognition rate of the method to the swelled kiwi fruits is greater than 99%. The method used for detecting whether the swelling agent is applied to the kiwi fruits or not based on the hyperspectral imaging technology has the advantages of nondestructive property, high efficiency, high accuracy and reliable result.

The invention discloses a pre-distortion model device, a signal pre-distortion treatment apparatus, a signal pre-distortion processing system, and a method for pre-distortion treatment of signal by utilizing the pre-distortion model device. Only one pre-distortion parameter should be exported by the pre-distortion model that the invention realized, and can cascade the complex multiplier, and can calibrate memory distortion and transient distortion of signal amplification equipment simultaneously. By using the invention, nonlinear model of the signal amplification equipment can not be extracted, which directly extract the pre-distortion model, thereby reducing the complexity of digital pre-distortion treatment, and can reduce the error.

A low cost signature test for RF and analog circuits. A model is provided to predict one or more performance parameters characterizing a first electronic circuit produced by a manufacturing process subject to process variation from the output of one or more second electronic circuits produced by the same process in response to a selected test stimulus, and iteratively varying the test stimulus to minimize the error between the predicted performance parameters and corresponding measured values for the performance parameters, for determining an optimized test stimulus. A non-linear model is preferably constructed for relating signature test results employing the optimized test stimulus in manufacturing testing to circuit performance parameters.

Non-linear model with disturbance rejection. A method for training a non linear model for predicting an output parameter of a system is disclosed that operates in an environment having associated therewith slow varying and unmeasurable disturbances. An input layer is provided having a plurality of inputs and an output layer is provided having at least one output for providing the output parameter. A data set of historical data taken over a time line at periodic intervals is generated for use in training the model. The model is operable to map the input layer through a stored representation to the output layer. Training of the model involves training the stored representation on the historical data set to provide rejection of the disturbances in the stored representation.

A system and computer-implemented method for determining properties of a subsurface region of interest from seismic data is disclosed. An embodiment of the method performs full waveform inversion by non-linear model update to compute a velocity model. The method includes obtaining actual seismic data representative of the subsurface region and an initial earth property model for the subsurface region, performing forward modeling using the initial earth property model to create modeled seismic data with similar acquisition specifications as the actual seismic data, calculating a residual between the actual seismic data and the modeled seismic data in a time or transform domain, and inverting the residual to generate a model produced by non-linear model update components. The system includes a data source, user interface, and processor configured to execute computer modules that implement the method.

A distortion model for a predistortion system uses tap output normalization to normalize the variance of data signals generated from different basis functions in a set of basis functions to a predetermined value. The distortion model is used by a distortion modeling circuit to calculate the weighting coefficients for a digital predistorter.

The servo control system of magnetically suspended control moment gyroscope frame with precise friction compensation includes mainly a magnetically suspended control moment gyroscope frame, a frame servo moment motor, a frame servo system controller and an adaptive friction compensator, which includes a frame servo system non-linear model identification unit and a frame servo system nominal model unit. The adaptive friction compensator is introduced for real-time acquiring the current signal and angular rate signal of the frame servo system and precise non-linear model identification, so as to compensate the friction of the frame servo system, regulate adaptively the friction moment compensation coefficient, raise the angular rate output precision and response speed of the frame servo system and improve the moment output precision and response speed of the whole magnetically suspended control moment gyroscope system.

The invention discloses an anti-disturbance guidance method for precision landing of a planet, relates to a guidance method for the precision landing of the planet and belongs to the technical field of deep-space detection. A non-linear model predication control method capable of directly treating non-linear and non-convex constraints is introduced in a plant descending process; a precision landing track optimization problem of the planet is only calculated on a limited-dimension rolling time domain and the calculation amount and the solving difficulty are reduced; online generation of a precision landing guidance rule of the planet and an optimal track is realized; meanwhile, external disturbance in the descending process is considered, real-time estimation is carried out on the external disturbance by adopting an expansion state observer and correcting control quantity, so as to realize disturbance compensation and inhibition and improve the safety of a landing task. The anti-disturbance guidance method has the following two advantages that (1) the calculation amount and solving difficulty of the precision landing track optimization problem of the planet can be reduced and the online generation of the optimal track is realized; (2) influences, caused by the external disturbance, on a system are reduced and the safety of the landing task is improved.

A method and apparatus for controlling a non-linear mill. A linear controller is provided having a linear gain k that is operable to receive inputs representing measured variables of the plant and predict on an output of the linear controller predicted control values for manipulatible variables that control the plant. A non-linear model of the plant is provided for storing a representation of the plant over a trained region of the operating input space and having a steady-state gain K associated therewith. The gain k of the linear model is adjusted with the gain K of the non-linear model in accordance with a predetermined relationship as the measured variables change the operating region of the input space at which the linear controller is predicting the values for the manipulatible variables. The predicted manipulatible variables are then output after the step of adjusting the gain k.

A method and system for process control using a model predictive controller. The control system can have one or more control devices operably coupled to a processing system for controlling a process of the processing system; a modeling tool to provide a non-linear model based at least in part on the process and to provide a plurality of linearized models based at least in part on the non-linear model, where the plurality of linearized models are linearized at different linearization rates; and a controller operably coupled to the modeling tool. The controller can select one of the plurality of linearized models based on a comparison of the plurality of linearized models with a reference model. The controller can send one or more control signals to at least one of the one or more control devices. The one or more control signals can be determined using the selected one of the plurality of linearized models.

The invention provides a method for estimating a road adhesion coefficient according to an on-line modified automobile state parameter. The method comprises the following steps: carrying out on-line identification on related parameters; transmitting a modified parameter to a state estimator; feeding back an estimation value obtained through the state estimator to a parameter modification part; then estimating the cornering force of a tire according to the estimation value, the identified parameter values and signals obtained through all sensors; transmitting the estimation result to a road adhesion coefficient estimator; finally, estimating the road adhesion coefficient; feeding back the result to a cornering force estimating module, thereby enabling the estimation result to be more accurate. The method adopts a parameter real-time modification function during the state estimation process, and feeds back the estimation value for re-modification, so that the problem of high filter divergence possibility during the state estimation process due to error caused by linearization of a non-linear model is solved, on-line renewal and modification of model parameters are realized, self-adaption of the estimation effect can be realized, and on-line estimation accuracy can be improved.

A system and method for predicting operation of a plant or process receive an input value from the plant or process. An integrity of a non-linear model corresponding to a local input space of the input value may be determined. The non-linear model may include an empirical representation of the plant or process. If the integrity is above a first threshold, non-linear model may be used to provide a first output value. However, if the integrity is below the first threshold, a linearized first principles model may be used to provide a second output value. The linearized first principles model may include an analytic representation of the plant or process. Additionally, the analytic representation of the plant or process may be independent of the empirical representation of the plant or process. The first output value and / or the second output value may be usable to manage the plant or process.

The invention relates to a processing method of lithography machine alignment signals, comprising the following steps: setting a non-linear model for processing the alignment signals of the lithography machine silicon slice; carrying out alignment scanning on the lithography machine silicon slice to obtain position sampling data and light intensity sampling data; carrying out real-time accumulative processing on the position sampling data and the light intensity sampling data to obtain accumulation coefficient; confirming non-linear parameter initial value of the non-linear model according to a linear model containing dark current, sine coefficient and cosine coefficient; obtaining the linear parameters of the non-linear model according to the accumulation coefficient and the non-linear parameter; obtaining iteration increment of the non-linear parameter according to the accumulation coefficient and the linear parameters of the non-linear model; updating the non-linear parameters of the non-linear model according to the iteration increment; judging whether the iteration increment meets accuracy requirement so as to determine the final linear parameters and the final non-linear parameters of the non-linear model.

A receiver is configured to receive a sample of an inter-symbol correlated (ISC) signal, the sample corresponding to a time instant when phase and / or amplitude of the ISC signal is a result of correlation among a plurality of symbols of a transmitted symbol sequence. The receiver may linearize the sample of the ISC signal. The receiver may calculate a residual signal value based on the linearized sample of the ISC signal. The receiver may generate an estimate of one or more of said plurality of symbols based on the residual signal value. The linearization may comprise applying an estimate of an inverse of a non-linear model. The non-linear model may be a model of nonlinearity experienced by the ISC signal in a transmitter from which the ISC signal originated, in a channel through which the ISC signal passed en route to the receiver, and / or in a front-end of the receiver.

The invention discloses a self-learning inverse model control method of an electronic throttle. The control method comprises two parts: 1) taking a radial basic function neural network as a recognizer to create the non-linear model of the electronic throttle; 2) taking another RBF neural network to serve as an inverse model controller to obtain an appropriate control amount. The two RBF neural networks adopt a self-learning strategy which comprises an off-line learning step and an on-line learning step. In off-line learning, the two RBF neural networks adjust and optimize network weighting parameters according to on-site sample data; after the off-line learning is completed, an on-line learning algorithm is adopted to adjust and optimize the network weighting parameters of the two RBF neural networks, and then the control performance can meet the requirements. The invention can overcome the difficulties of nonlinearity, time varying and the like in electronic throttle control and improve the control effect and performance.

The invention discloses a predistortion model device, a predistortion processing device for signals, a predistortion processing system for signals and a method for performing predistortion processing to the signals by using the predistortion model device. The predistortion model realized through the invention can only output a predistortion parameter, can use a complex multiplier for cascade connection and also can perform calibration to the transient distortion and the memory distortion of signal amplification device at the same time. By using the invention, the non-linear model of the signal amplification device is not required to be extracted, and the predistortion model is directly extracted, therefore, the complexity of digital predistortion processing can be reduced, and the error is also reduced.

A method and apparatus for controlling a non-linear mill. A linear controller is provided having a linear gain k that is operable to receive inputs representing measured variables of the plant and predict on an output of the linear controller predicted control values for manipulatible variables that control the plant. A non-linear model of the plant is provided for storing a representation of the plant over a trained region of the operating input space and having a steady-state gain K associated therewith. The gain k of the linear model is adjusted with the gain K of the non-linear model in accordance with a predetermined relationship as the measured variables change the operating region of the input space at which the linear controller is predicting the values for the manipulatible variables. The predicted manipulatible variables are then output after the step of adjusting the gain k.