Abstract:A model is developed to predict rates of undesirable reactions in the low‐moisture, high‐temperature finishing stage of nylon 66 production. The model contains 56 unknown parameters and initial conditions, which are ranked based on their influence on model predictions, correlation with other parameters and uncertainty in their initial guesses. A mean‐square‐error criterion is used to determine that 43 of 56 parameters should be estimated. The proposed model, which describes the effect of melt‐phase water conce… Show more
“…McLean et al used Wu's r CC criterion to simultaneously rank and select parameters for estimation in nonlinear dynamic models (Algorithm ) . In situations involving a noninvertible FIM, the r CC criterion can readily be used to rank or select parameters . The only assumption required is that the value of the weighted least‐squares objective function (i.e., Eq.…”
Section: Background Informationmentioning
confidence: 99%
“…The two approaches for parameter ranking and selection (i.e., [using orthogonalization and fixing of parameters to achieve an invertible FIM] and [using a pseudoinverse]) were investigated. In Tables and the results are compared against the typical strategy that uses orthogonalization and r CC (without considering W ) so that the importance of using W when ranking and selecting parameters can be elucidated. Note that when the approach is used, only the first five columns corresponding to the first five parameters are included in W ext because the sixth and seventh parameters are automatically removed from the estimation problem during the preliminary orthogonalization/ranking step (i.e., they are held fixed at their initial guesses).…”
Section: Linear Regression Examplementioning
confidence: 99%
“…Several parameter subset‐selection techniques have been developed and used to decide which parameters in an EM should be estimated and which parameters should be fixed at their nominal values to create a corresponding SM . Orthogonalization‐based techniques (Algorithm ), which use the Fisher information matrix (FIM) to rank model parameters from most estimable to least estimable are popular for parameter ranking and selection, and have been widely used to estimate parameters in complex fundamental models . The FIM summarizes information about the effects of model parameters on predictions and accounts for uncertainties in measurements that are used for parameter estimation.…”
Section: Introductionmentioning
confidence: 99%
“…A singular FIM occurs in many high‐dimensional parameter estimation problems and also in system identification with over‐parameterized models . In many practical chemical engineering models, it is impossible to rank all of the model parameters using the orthogonalization‐based technique in Algorithm , due to a noninvertible FIM In this situation, the algorithm stops when numerical problems are encountered and some of the least influential parameters remain unranked. A singular FIM also leads to a significant complication in the analysis of parameter estimation problems and for the theory of the Cramér–Rao lower bound, which is a popular means to bound the variances of unbiased estimators …”
Two approaches are developed to rank and select model parameters for estimation in complex models when data are limited, the Fisher information matrix (FIM) is noninvertible, and accurate predictions are desired at key operating conditions. These approaches are evaluated using synthetic data sets in a linear regression example to examine the influence of several factors including: the quality of initial parameter guesses, uncertainty ranges for initial parameter values, noise variances, and the operating region of interest. It is shown that using a reduced FIM with full rank leads to more reliable model predictions for a variety of cases than the alternative approach using the pseudoinverse of the FIM. The proposed reduced-FIM methodology also provides better predictions than related techniques that do not consider the operating region where reliable predictions are required. The methodology is illustrated using a nonlinear differential equation model of a polymer film casting process.Case 1: b 0 5 1.1*b, s b0 5 0.1*b 0 ; Case 2: b 0 5 1.5*b, s b0 5 b 0 ; Case 3: b 0 5 10*b, s b0 5 0.1*b 0 . r 1 CCW corresponds to the first approach and r 2 CCW corresponds to the second approach for parameter ranking and selection when FIM is not invertible.Case 1: b 0 5 1.1*b, s b0 5 0.1*b 0 ; Case 2: b 0 5 1.5*b, s b0 5 b 0 ; Case 3: b 0 5 10*b, s b0 5 0.1*b 0 . r 1 CCW corresponds to the first approach and r 2 CCW corresponds to the second approach for parameter ranking and selection when FIM is not invertible. SM0 contains three parameters b 1 , b 2 , and b 3 and was selected using the r CC approach. SM1 contains two parameters b 1 and b 2 and was selected using the r 1 CCW approach. SM2 contains two parameters b 5 and b 6 and was selected using the r 2 CCW approach. The EM contains all seven parameters.
“…McLean et al used Wu's r CC criterion to simultaneously rank and select parameters for estimation in nonlinear dynamic models (Algorithm ) . In situations involving a noninvertible FIM, the r CC criterion can readily be used to rank or select parameters . The only assumption required is that the value of the weighted least‐squares objective function (i.e., Eq.…”
Section: Background Informationmentioning
confidence: 99%
“…The two approaches for parameter ranking and selection (i.e., [using orthogonalization and fixing of parameters to achieve an invertible FIM] and [using a pseudoinverse]) were investigated. In Tables and the results are compared against the typical strategy that uses orthogonalization and r CC (without considering W ) so that the importance of using W when ranking and selecting parameters can be elucidated. Note that when the approach is used, only the first five columns corresponding to the first five parameters are included in W ext because the sixth and seventh parameters are automatically removed from the estimation problem during the preliminary orthogonalization/ranking step (i.e., they are held fixed at their initial guesses).…”
Section: Linear Regression Examplementioning
confidence: 99%
“…Several parameter subset‐selection techniques have been developed and used to decide which parameters in an EM should be estimated and which parameters should be fixed at their nominal values to create a corresponding SM . Orthogonalization‐based techniques (Algorithm ), which use the Fisher information matrix (FIM) to rank model parameters from most estimable to least estimable are popular for parameter ranking and selection, and have been widely used to estimate parameters in complex fundamental models . The FIM summarizes information about the effects of model parameters on predictions and accounts for uncertainties in measurements that are used for parameter estimation.…”
Section: Introductionmentioning
confidence: 99%
“…A singular FIM occurs in many high‐dimensional parameter estimation problems and also in system identification with over‐parameterized models . In many practical chemical engineering models, it is impossible to rank all of the model parameters using the orthogonalization‐based technique in Algorithm , due to a noninvertible FIM In this situation, the algorithm stops when numerical problems are encountered and some of the least influential parameters remain unranked. A singular FIM also leads to a significant complication in the analysis of parameter estimation problems and for the theory of the Cramér–Rao lower bound, which is a popular means to bound the variances of unbiased estimators …”
Two approaches are developed to rank and select model parameters for estimation in complex models when data are limited, the Fisher information matrix (FIM) is noninvertible, and accurate predictions are desired at key operating conditions. These approaches are evaluated using synthetic data sets in a linear regression example to examine the influence of several factors including: the quality of initial parameter guesses, uncertainty ranges for initial parameter values, noise variances, and the operating region of interest. It is shown that using a reduced FIM with full rank leads to more reliable model predictions for a variety of cases than the alternative approach using the pseudoinverse of the FIM. The proposed reduced-FIM methodology also provides better predictions than related techniques that do not consider the operating region where reliable predictions are required. The methodology is illustrated using a nonlinear differential equation model of a polymer film casting process.Case 1: b 0 5 1.1*b, s b0 5 0.1*b 0 ; Case 2: b 0 5 1.5*b, s b0 5 b 0 ; Case 3: b 0 5 10*b, s b0 5 0.1*b 0 . r 1 CCW corresponds to the first approach and r 2 CCW corresponds to the second approach for parameter ranking and selection when FIM is not invertible.Case 1: b 0 5 1.1*b, s b0 5 0.1*b 0 ; Case 2: b 0 5 1.5*b, s b0 5 b 0 ; Case 3: b 0 5 10*b, s b0 5 0.1*b 0 . r 1 CCW corresponds to the first approach and r 2 CCW corresponds to the second approach for parameter ranking and selection when FIM is not invertible. SM0 contains three parameters b 1 , b 2 , and b 3 and was selected using the r CC approach. SM1 contains two parameters b 1 and b 2 and was selected using the r 1 CCW approach. SM2 contains two parameters b 5 and b 6 and was selected using the r 2 CCW approach. The EM contains all seven parameters.
“…When developing fundamental models of polymerization reactors, it is important to determine whether all of the kinetic parameters in the model should be estimated, or whether only a subset of the parameters should be estimated from the available data . Estimating too many parameters using limited data leads to large uncertainty ranges for the parameters and can produce worse predictions than when fewer parameters are estimated.…”
Section: Parameter Estimation and Simulation Resultsmentioning
A mathematical model is developed for estimating kinetic parameters that influence the production of arborescent polyisobutylene via carbocationic copolymerization of inimer (IM) and isobutylene. Six different propagation rate constants arise due to the two types of vinyl groups and three types of carbocations. These six parameters are estimated using parallel simulation systems in PREDICI that track (1) functional groups, (2) internal and dangling segments in the polymer, and (3) concentrations of IM and polymer molecules. Parameter estimates obtained using the proposed model result in a better fit to literature data than was obtained using a previous model that neglected two types of propagations reactions. Predictions from the proposed model are consistent with Monte Carlo simulations for molecular weight distribution of the internal and dangling segments.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.