A New Method for Identifying the Model Error of Adjust ment System

Some theory problems affecting parameter estimation are discussed in this paper. Influence and transformation between errors of stochastic and functional models is pointed out as well. For choosing the best adjustment model, a formula, which is different from the literatures existing methods, for estimating and identifying the model error, is proposed. On the basis of the proposed formula, an effective approach of selecting the best model of adjustment system is given.

Neural network based method for compensating model error

Two traditional methods for compensating function model errors, the method of adding systematic parameters and the least-squares collection method, are introduced. A proposed method based on a BP neural network （called the H-BP algorithm） for compensating function model errors is put forward. The function model is assumed as y =f（x1, x2,… ,xn）, and the special structure of the H-BP algorithm is determined as （ n ＋ 1） ×p × 1, where （n ＋ 1） is the element number of the input layer, and the elements are xl, x2,…, xn and y＇ （ y＇ is the value calculated by the function model）; p is the element number of the hidden layer, and it is usually determined after many tests; 1 is the dement number of the output layer, and the element is △y = y0-y＇（y0 is the known value of the sample）. The calculation steps of the H-BP algorithm are introduced in detail. And then, the results of three methods for compensating function model errors from one engineering project are compared with each other. After being compensated, the accuracy of the traditional methods is about ± 19 mm, and the accuracy of the H-BP algorithm is ± 4. 3 mm. It shows that the proposed method based on a neural network is more effective than traditional methods for compensating function model errors.

An approach to estimating and extrapolating model error based on inverse problem methods:towards accurate numerical weather prediction

Model error is one of the key factors restricting the accuracy of numerical weather prediction （NWP）. Considering the continuous evolution of the atmosphere, the observed data （ignoring the measurement error） can be viewed as a series of solutions of an accurate model governing the actual atmosphere. Model error is represented as an unknown term in the accurate model, thus NWP can be considered as an inverse problem to uncover the unknown error term. The inverse problem models can absorb long periods of observed data to generate model error correction procedures. They thus resolve the deficiency and faultiness of the NWP schemes employing only the initial-time data. In this study we construct two inverse problem models to estimate and extrapolate the time-varying and spatial-varying model errors in both the historical and forecast periods by using recent observations and analogue phenomena of the atmosphere. Numerical experiment on Burgers＇ equation has illustrated the substantial forecast improvement using inverse problem algorithms. The proposed inverse problem methods of suppressing NWP errors will be useful in future high accuracy applications of NWP.

An Online Model Correction Method Based on an Inverse Problem:Part I—Model Error Estimation by Iteration

Errors inevitably exist in numerical weather prediction （NWP） due to imperfect numeric and physical parameterizations. To eliminate these errors, by considering NWP as an inverse problem, an unknown term in the prediction equations can be estimated inversely by using the past data, which are presumed to represent the imperfection of the NWP model （model error, denoted as ME）. In this first paper of a two-part series, an iteration method for obtaining the MEs in past intervals is presented, and the results from testing its convergence in idealized experiments are reported. Moreover, two batches of iteration tests were applied in the global forecast system of the Global and Regional Assimilation and Prediction System （GRAPES-GFS） for July-August 2009 and January-February 2010. The datasets associated with the initial conditions and sea surface temperature （SST） were both based on NCEP （National Centers for Environmental Prediction） FNL （final） data. The results showed that 6th h forecast errors were reduced to 10% of their original value after a 20-step iteration. Then, off-line forecast error corrections were estimated linearly based on the 2-month mean MEs and compared with forecast errors. The estimated error corrections agreed well with the forecast errors, but the linear growth rate of the estimation was steeper than the forecast error. The advantage of this iteration method is that the MEs can provide the foundation for online correction. A larger proportion of the forecast errors can be expected to be canceled out by properly introducing the model error correction into GRAPES-GFS.

Approach for wideband direction-of-arrival estimation in the presence of array model errors

The presence of array imperfection and mutual coupling in sensor arrays poses several challenges for development of effective algorithms for the direction-of-arrival （DOA） estimation problem in array processing. A correlation domain wideband DOA estimation algorithm without array calibration is proposed, to deal with these array model errors, using the arbitrary antenna array of omnidirectional elements. By using the matrix operators that have the memory and oblivion characteristics, this algorithm can separate the incident signals effectively. Compared with other typical wideband DOA estimation algorithms based on the subspace theory, this algorithm can get robust DOA estimation with regard to position error, gain-phase error, and mutual coupling, by utilizing a relaxation technique based on signal separation. The signal separation category and the robustness of this algorithm to the array model errors are analyzed and proved. The validity and robustness of this algorithm, in the presence of array model errors, are confirmed by theoretical analysis and simulation results.

Recursive weighted least squares estimation algorithm based on minimum model error principle

Kalman filter is commonly used in data filtering and parameters estimation of nonlinear system,such as projectile's trajectory estimation and control.While there is a drawback that the prior error covariance matrix and filter parameters are difficult to be determined,which may result in filtering divergence.As to the problem that the accuracy of state estimation for nonlinear ballistic model strongly depends on its mathematical model,we improve the weighted least squares method(WLSM)with minimum model error principle.Invariant embedding method is adopted to solve the cost function including the model error.With the knowledge of measurement data and measurement error covariance matrix,we use gradient descent algorithm to determine the weighting matrix of model error.The uncertainty and linearization error of model are recursively estimated by the proposed method,thus achieving an online filtering estimation of the observations.Simulation results indicate that the proposed recursive estimation algorithm is insensitive to initial conditions and of good robustness.

An Online Model Correction Method Based on an Inverse Problem:PartⅡ——Systematic Model Error Correction

An online systematic error correction is presented and examined as a technique to improve the accuracy of real-time numerical weather prediction, based on the dataset of model errors （MEs） in past intervals. Given the analyses, the ME in each interval （6 h） between two analyses can be iteratively obtained by introducing an unknown tendency term into the prediction equation, shown in Part I of this two-paper series. In this part, after analyzing the 5-year （2001-2005） GRAPES- GFS （Global Forecast System of the Global and Regional Assimilation and Prediction System） error patterns and evolution, a systematic model error correction is given based on the least-squares approach by firstly using the past MEs. To test the correction, we applied the approach in GRAPES-GFS for July 2009 and January 2010. The datasets associated with the initial condition and SST used in this study were based on NCEP （National Centers for Environmental Prediction） FNL （final） data. The results indicated that the Northern Hemispheric systematically underestimated equator-to-pole geopotential gradient and westerly wind of GRAPES-GFS were largely enhanced, and the biases of temperature and wind in the tropics were strongly reduced. Therefore, the correction results in a more skillful forecast with lower mean bias and root-mean-square error and higher anomaly correlation coefficient.

Application of Backward Nonlinear Local Lyapunov Exponent Method to Assessing the Relative Impacts of Initial Condition and Model Errors on Local Backward Predictability

Initial condition and model errors both contribute to the loss of atmospheric predictability.However,it remains debatable which type of error has the larger impact on the prediction lead time of specific states.In this study,we perform a theoretical study to investigate the relative effects of initial condition and model errors on local prediction lead time of given states in the Lorenz model.Using the backward nonlinear local Lyapunov exponent method,the prediction lead time,also called local backward predictability limit(LBPL),of given states induced by the two types of errors can be quantitatively estimated.Results show that the structure of the Lorenz attractor leads to a layered distribution of LBPLs of states.On an individual circular orbit,the LBPLs are roughly the same,whereas they are different on different orbits.The spatial distributions of LBPLs show that the relative effects of initial condition and model errors on local backward predictability depend on the locations of given states on the dynamical trajectory and the error magnitudes.When the error magnitude is fixed,the differences between the LBPLs vary with the locations of given states.The larger differences are mainly located on the inner trajectories of regimes.When the error magnitudes are different,the dissimilarities in LBPLs are diverse for the same given state.

The Combined Effect of Initial Error and Model Error on ENSO Prediction Uncertainty Generated by the Zebiak-Cane Model

Initial errors and model errors are the source of prediction errors. In this study, the authors compute the conditional nonlinear optimal perturbation （CNOP）-type initial errors and nonlinear forcing singular vector （NFSV）- type tendency errors of the Zebiak-Cane model with respect to El Nifio events and analyze their combined effect on the prediction errors for E1 Nino events. The CNOP- type initial error （NFSV-type tendency error） represents the initial errors （model errors） that have the largest effect on prediction uncertainties for E1 Nifio events under the perfect model （perfect initial conditions） scenario. How- ever, when the CNOP-type initial errors and the NFSV- type tendency errors are simultaneously considered in the model, the prediction errors caused by them are not am- plified as the authors expected. Specifically, the predic- tion errors caused by the combined mode of CNOP-type initial errors and NFSV-type tendency errors are a little larger than those caused by the NFSV-type tendency er- rors. This fact emphasizes a need to investigate the opti- mal combined mode of initial errors and tendency errors that cause the largest prediction error for E1 Nifio events.

FORMING DYNAMIC EQUATIONS OF ELASTIC LINKAGE AND INVESTIGATION OF MODEL ERROR

Dynamic equations of elastic linkage are formed by means of Kane equation and finitemethod. Eight main influential factors on the dynamic response of elastic linkage are considered inthese equations. Model error caused by the eight factors are investigated. Some useful conclusionsabout model error are derived from theoretical analysis and the numerical calculation of twenty-sixexamples.

A hybrid deep learning and data assimilation method for model error estimation

Forecast errors of numerical weather prediction consist of model errors and the growth of initial condition errors,while the initial condition is often optimized based on short-term forecasts.Thus it is difficult to untangle the initial condition error and model error,but it is essential to infer model errors not just for prediction but also for data assimilation(DA).A hybrid deep learning(DL)and DA method is proposed here,aiming to correct model errors.It uses a convolutional neural network(CNN)to extract characteristics of initial conditions and forecast errors,and then provides estimations for model errors.The CNN-based model error estimation method can consider the model error resulted from inaccurate model parameters,or simultaneously consider the model error and initial condition error.Based on the Lorenz05 model,offline and online experiments demonstrate that the CNN-based model error estimation method can effectively correct model errors resulted from inaccurate model parameters,including the forcing F,coupling coefficient c,and relative scale b.For both online and offline model error estimations,simultaneously considering model errors and initial condition errors are beneficial to infer the model errors,compared to considering model errors only.Moreover,using the observations to verify the forecasts has advantages over using the analyses,to estimate the model errors.Using observations can also achieve a faster convergence of model error estimation with online DA than using analyses.

Error modeling of 3-RSR parallel robot based on D-H transformation matrix

By selecting any one limb of 3-RSR parallel robot as a research object, the paper establishes a position and orienta- tion relationship matrix between the moving platform and the base by means of Denavit-Hartenberg （D-H） transformation matrix. The error mapping model is derived from original error to the error of the platform by using matrix differential method. This model contains all geometric original errors of the robot. The nonlinear implicit function relation between po- sition and orientation error of the platform and the original geometric errors is simplified as a linear explicit function rela- tion. The results provide a basis for further studying error analysis and error compensation.

BDS satellite clock offset prediction based on a semiparametric adjustment model considering model errors

In view of the influence of model errors in conventional BeiDou prediction models for clock offsets,a semiparametric adjustment model for BeiDou Navigation Satellite System(BDS)clock offset prediction that considers model errors is proposed in this paper.First,the model errors of the conventional BeiDou clock offset prediction model are analyzed.Additionally,the relationship among the polynomial model,polynomial model with additional periodic term correction,and its periodic correction terms is explored in detail.Second,considering the model errors,combined with the physical relationship between phase,frequency,frequency drift,and its period in the clock sequence,the conventional clock offset prediction model is improved.Using kernel estimation and comprehensive least squares,the corresponding parameter solutions of the prediction model and the estimation of its model error are derived,and the dynamic error correction of the clock sequence model is realized.Finally,the BDS satellite precision clock data provided by the IGS Center of Wuhan University with a sampling interval of 5 min are used to compare the proposed prediction method with commonly used methods.Experimental results show that the proposed prediction method can better correct the model errors of BDS satellite clock offsets,and it can effectively overcome the inaccuracies of clock offset correction.The average forecast accuracies of the BeiDou satellites at 6,12,and 24 h are 27.13%,37.71%,and 45.08%higher than those of the conventional BeiDou clock offset forecast models;the average model improvement rates are 16.92%,20.96%,and 28.48%,respectively.In addition,the proposed method enhances the existing BDS satellite prediction method for clock offsets to a certain extent.

Time-Dependent Nonlinear Forcing Singular Vector-Type Tendency Error of the Zebiak-Cane Model

Based on the Zebiak-Cane model, the timedependent nonlinear forcing singular vector （NFSV）-type tendency errors with components of 4 and 12 （denoted by NFSV-4 and NFSV-12） are calculated for predetermined El Nifio events and compared with the constant NFSV （denoted by NFSV-1） from their patterns and resultant prediction errors. Specifically, NFSV-1 has a zonal dipolar sea surface temperature anomaly （SSTA） pattern with negative anomalies in the equatorial eastern Pacific and positive anomalies in the equatorial central-western Pa- cific. Although the first few components in NFSV-4 and NFSV-12 present patterns similar to NFSV-1, they tend to extend their dipoles farther westward; meanwhile, the positive anomalies gradually cover much smaller regions with the lag times. In addition, the authors calculate the predic- tion errors caused by the three kinds of NFSVs, and the results indicate that the prediction error induced by NFSV-12 is the largest, followed by the NFSV-4. However, when compared with the prediction errors caused by random tendency errors, the NFSVs generate significantly larger prediction errors. It is therefore shown that the spatial structure of tendency errors is important for producing large prediction errors. Furthermore, in exploring the tendency errors that cause the largest prediction error for E1 Nifio events, the timedependent NFSV should be evaluated.

Revisiting Total Model Errors and Model Validation

The paper contains a discussion of earlier work on Total Model Errors and Model Validation.It is maintained that the recent change of paradigm to kernel based system identification has also affected the basis for(and interest in)giving bounds for the total model error.

Non-gaussian Test Models for Prediction and State Estimation with Model Errors

Turbulent dynamical systems involve dynamics with both a large dimensional phase space and a large number of positive Lyapunov exponents. Such systems are ubiqui- tous in applications in contemporary science and engineering where the statistical ensemble prediction and the real time filtering/state estimation are needed despite the underlying complexity of the system. Statistically exactly solvable test models have a crucial role to provide firm mathematical underpinning or new algorithms for vastly more complex scien- tific phenomena. Here, a class of statistically exactly solvable non-Gaussian test models is introduced, where a generalized Feynman-Ka~ formulation reduces the exact behavior of conditional statistical moments to the solution to inhomogeneous Fokker-Planck equations modified by linear lower order coupling and source terms. This procedure is applied to a test model with hidden instabilities and is combined with information theory to address two important issues in the contemporary statistical prediction of turbulent dynamical systems： the coarse-grained ensemble prediction in a perfect model and the improving long range forecasting in imperfect models. The models discussed here should be use- ful for many other applications and algorithms for the real time prediction and the state estimation.

R-Factor Analysis of Data Based on Population Models Comprising R- and Q-Factors Leads to Biased Loading Estimates

Effects of performing an R-factor analysis of observed variables based on population models comprising R- and Q-factors were investigated. Although R-factor analysis of data based on a population model comprising R- and Q-factors is possible, this may lead to model error. Accordingly, loading estimates resulting from R-factor analysis of sample data drawn from a population based on a combination of R- and Q-factors will be biased. It was shown in a simulation study that a large amount of Q-factor variance induces an increase in the variation of R-factor loading estimates beyond the chance level. Tests of the multivariate kurtosis of observed variables are proposed as an indicator of possible Q-factor variance in observed variables as a prerequisite for R-factor analysis.

Error Modeling and Sensitivity Analysis of a Parallel Robot with SCARA(Selective Compliance Assembly Robot Arm) Motions

Parallel robots with SCARA（selective compliance assembly robot arm） motions are utilized widely in the field of high speed pick-and-place manipulation. Error modeling for these robots generally simplifies the parallelogram structures included by the robots as a link. As the established error model fails to reflect the error feature of the parallelogram structures, the effect of accuracy design and kinematic calibration based on the error model come to be undermined. An error modeling methodology is proposed to establish an error model of parallel robots with parallelogram structures. The error model can embody the geometric errors of all joints, including the joints of parallelogram structures. Thus it can contain more exhaustively the factors that reduce the accuracy of the robot. Based on the error model and some sensitivity indices defined in the sense of statistics, sensitivity analysis is carried out. Accordingly, some atlases are depicted to express each geometric error’s influence on the moving platform’s pose errors. From these atlases, the geometric errors that have greater impact on the accuracy of the moving platform are identified, and some sensitive areas where the pose errors of the moving platform are extremely sensitive to the geometric errors are also figured out. By taking into account the error factors which are generally neglected in all existing modeling methods, the proposed modeling method can thoroughly disclose the process of error transmission and enhance the efficacy of accuracy design and calibration.

ENSO ensemble prediction:Initial error perturbations vs. model error perturbations

Based on our developed ENSO (El Nio-Southern Oscillation) ensemble prediction system (EPS), the impacts of stochastic initial-error and model-error perturbations on ENSO ensemble predictions are examined and discussed by performing four sets of 14-a retrospective forecast experiments in both a deterministic and probabilistic sense. These forecast schemes are differentiated by whether they considered the initial or model stochastic perturbations. The comparison results suggest that the stochastic model-error perturbations, which are added into the modeled physical fields to mainly represent the uncertainties of the physical model, have significant, positive impacts on improving the ensemble prediction skills during the entire 12-month forecast process. However, the stochastic initial-error perturbations have relatively small impacts on the ensemble prediction system, and its impacts are mainly focusing on the first 3-month predictions.

Error model identification of inertial navigation platform based on errors-in-variables model

Because the real input acceleration cannot be obtained during the error model identification of inertial navigation platform, both the input and output data contain noises. In this case, the conventional regression model and the least squares （LS） method will result in bias. Based on the models of inertial navigation platform error and observation error, the errors-in-variables （EV） model and the total least squares （TLS） method axe proposed to identify the error model of the inertial navigation platform. The estimation precision is improved and the result is better than the conventional regression model based LS method. The simulation results illustrate the effectiveness of the proposed method.