Global Optimal Parameters Research Articles

Table structure understanding and its performance evaluation

This paper presents a table structure understanding algorithm designed using optimization methods. The algorithm is probability based, where the probabilities are estimated from geometric measurements made on the various entities in a large training set. The methodology includes a global parameter optimization scheme, a novel automatic table ground truth generation system and a table structure understanding performance evaluation protocol. With a document data set having 518 table and 10,934 cell entities, it performed at the 96.76% accuracy rate on the cell level and 98.32% accuracy rate on the table level.

A Numerical Solution of the Inverse Problem in Classical Celestial Mechanics, with Application to Mercury's Motion

It is attempted to obtain the masses of the celestial bodies, the initial conditions of their motion, and the constant of gravitation, by a global parameter optimization. First, a numerical solution of the N-bodies problem for mass points is described and its high accuracy verified. The osculating elements are also accurately computed. This solution is implemented in the Gauss iterative algorithm for solving nonlinear least-squares problems. This algorithm is summarized, and its efficiency for the inverse problem in celestial mechanics is checked on a 3-bodies problem. Then it is used to assess the accuracy to which a Newtonian calculation may reproduce the DE403 ephemeris that involves general-relativistic corrections. The parameter optimization allows one to reduce the norm and angular differences between the Newtonian calculation and DE403 by a factor 10 (Mercury, Pluto) to 100 (Venus). The maximum angular difference of the heliocentric positions of Mercury is ca. 220″ per century before the optimization, and ca. 20″ after it. The latter is still far above the observational accuracy. On the other hand, Mercury's longitude of the perihelion is not affected by the optimization: it keeps the linear advance of 43″ per century.

Global geometry optimization of small silicon clusters with empirical potentials and at the DFT level

We have performed global parameter optimization of selected empirical potentials for silicon, resulting in improved performance for small to medium-sized silicon clusters, as judged by a comparison of globally optimized cluster structures to the structures accepted in the literature for the size range up to n = 10. Using global cluster structure optimizations with the resulting optimized model potential and ensuing local optimizations at the DFT level, we could find improved proposals for global minimum structures in the size region n = 10–16. This study confirms the applicability of our general global cluster optimization strategy for still larger silicon clusters.

Detailed Kinetics of Fischer−Tropsch Synthesis on an Industrial Fe−Mn Catalyst

The detailed kinetics of the Fischer−Tropsch synthesis over an industrial Fe−Mn catalyst was studied in a continuous integral fixed-bed reactor under the conditions relevant to industrial operations [temperature, 540−600 K; pressure, 1.0−3.0 MPa; H2/CO feed ratio, 1.0−3.0; space velocity, (1.6−4.2) × 10-3 Nm3 kg of catalyst-1 s-1]. Reaction rate equations were derived on the basis of the Langmuir−Hinshelwood−Hougen−Watson type models for the Fischer−Tropsch reactions and the water-gas-shift reaction. Kinetic model candidates were evaluated by the global optimization of kinetic parameters, which were realized by first minimization of multiresponse objective functions with a genetic algorithm approach and second optimization with the conventional Levenberg−Marquardt method. It was found that an alkylidene mechanism based model could produce a good fit of the experimental data. This model shows that the desorption of the products and the insertion of methylene into the metal−alkylidene bond are the rate-dete...

Transducer design with global parametric optimization

This paper presents a plan for optimizing transducer and motor designs. Designing transducers or motors always requires analysis of several parameters such as length, temperature, efficiency, weight, etc. To obtain the optimum results for given specifications, the method suggested here globally analyzes all the free parameters in one step, with an optimization algorithm based on Matlab software that includes the optimization toolbox. A general module makes it possible then to get the optimum design according to the constraints of the problem. In the future, the authors will develop a module for each type of transducer and motor as well as a module for the driver.

GLOBAL CONTROLLER OPTIMIZATION USING HOROWITZ BOUNDS

A procedure for global optimization of PID type controller parameters for SISO plants with model uncertainty is presented. Robustness to the uncertainties is guaranteed by the use of Horowitz bounds, which are used as constraints when low frequency performance is optimized. The basic idea of both the optimization and the parameter tuning is to formulate separate criteria for low, mid and high frequency closed loop properties. The trade-off between stability margins, high frequency robustness and low frequency performance is then elucidated and, hence, the final choice of parameters is facilitated. The optimization problems are non-convex and ill-conditioned and we use a combination of new global and standard local optimization algorithms available in the TOMLAB optimization environment to solve the problem. The method does not rely on a good initial guess and converges fast and robustly. It is applied to a controller structure comparison for a plant with an uncertain mechanical resonance. For a given control activity and stability margin as well as identical tuning parameters it is shown that a PID controller achieves slightly improved low frequency performance compared to an ℋ∞ controller based on loop-shaping. The reason for this somewhat surprising result is the roll-off in the ℋ∞ controller, which adds additional high frequency robustness compared to the PID controller. Computationally, a factor of 10–20 has been gained compared to an earlier, less general, version of the procedure.

Real-Time Offset Transitioning Algorithm for Coordinating Traffic Signals

A traffic signal offset transitioning algorithm is introduced that can be viewed as an integrated optimization approach designed to work with traditional coordinated-actuated systems. The proposed approach assumes a fixed cycle length (selected by either time of day or some traffic-responsive technique). The splits are determined by each local controller subject to maximum and minimum constraints traditionally imposed by coordinated-actuated signal systems. End-of-green offsets at each intersection are continually adjusted by the proposed algorithm with the objective of providing smooth progression of a platoon through an intersection using the volume and occupancy profile of advance detectors. The algorithm was implemented in a hardware-in-the-loop simulation and evaluated with a set of National Transportation Communications for Intelligent Transportation Systems Protocol (NTCIP) National Electrical Manufacturers Association controllers. The offsets were manipulated with NTCIP messages. The unique aspect of this algorithm is the cycle-based procedure used to tabulate volume and occupancy profiles, which can then be used to adjust the offsets in a traditional coordinated-actuated signal system. This automatic tuning process is analogous to an engineer or technician standing beside the cabinet and tuning the offset so that the coordinated phase turns green at the appropriate time to facilitate smooth progression of the upstream platoon. Because only the offsets are tuned, it is unlikely that this algorithm will be able to achieve the global optimum parameters or rapid adaptation possible with model-based approaches; nevertheless, this algorithm has the advantage of working within the framework of traditional coordinated-actuated signal systems that are familiar to system operators.

Qualitative-evolutionary design of greenhouse environment control agents

A novel technique that fuses qualitative reasoning and evolutionary computation in the design of control agents for the control of the environment of a greenhouse is proposed in this paper. Linguistic rules relate the attributes of the greenhouse performance in response to perturbations in the desired set points to the suitability of the agent. De-fuzzification of the fuzzy suitability membership function yields a quantitative measure of the fitness of the agent to satisfy the system performance specifications. The fitness measure is used in a genetic algorithm that performs a stochastic search for the global optimum parameters of the agent.

A semi-distributed parallel-type linear reservoir rainfall-runoff model and its application in Taiwan

The main purpose of this paper is to introduce a semi-distributed parallel surface rainfall-runoff conceptual model. In this paper, a general solution of the instantaneous unit hydrograph (IUH) has been derived successfully for N linearly connected reservoirs, each having a different storage constant. The solution is a function of geomorphologic parameters, meteorologic factors and roughness coefficients. The model also takes into account the hydrologic response which is influenced by outflow downstream of a reservoir. For calibration, the shuffled complex evolution (SCE) algorithm is used to search for the global optimal parameters of the model. Because of the parallel structure, the mean roughness parameter of the channel becomes a “conceptual” parameter without a real physical meaning. To evaluate the adaptability of the model adopted, three watersheds around the city of Taipei in Taiwan were chosen to test the effectiveness of the model. The study provides an appropriate rainfall-runoff model for planning flood mitigation in Taiwan. Copyright © 1999 John Wiley & Sons, Ltd.

A semi‐distributed parallel‐type linear reservoir rainfall‐runoff model and its application in Taiwan

The main purpose of this paper is to introduce a semi-distributed parallel surface rainfall-runoff conceptual model. In this paper, a general solution of the instantaneous unit hydrograph (IUH) has been derived successfully for N linearly connected reservoirs, each having a different storage constant. The solution is a function of geomorphologic parameters, meteorologic factors and roughness coefficients. The model also takes into account the hydrologic response which is influenced by outflow downstream of a reservoir. For calibration, the shuffled complex evolution (SCE) algorithm is used to search for the global optimal parameters of the model. Because of the parallel structure, the mean roughness parameter of the channel becomes a “conceptual” parameter without a real physical meaning. To evaluate the adaptability of the model adopted, three watersheds around the city of Taipei in Taiwan were chosen to test the effectiveness of the model. The study provides an appropriate rainfall-runoff model for planning flood mitigation in Taiwan. Copyright © 1999 John Wiley & Sons, Ltd.

The potential energy surfaces of the six lowest singlet states of HOCl: global optimization of parameters for an extended anti-Morse function and a diatomics-in-molecules (DIM) model

The potential energy function of the electronic ground state and the five lowest excited singlet states of HOCl are described by analytical functions. The parameters contained in these functions were optimized through a global fit to abinitio data from Nanbu and Iwata (J. Phys. Chem., 1992, 96, 2103), employing a genetic algorithm. An improved parametrization of the functional form originally proposed by Nanbu and Iwata was found, which uses 12 parameters for each surface. A fit of equal quality was obtained with functions derived with the diatomics-in-molecules (DIM) approach. These describe all 6 singlet states simultaneously with a common parameter set of only 20 parameters. Within this model the extension for including the O–H distance as an additional coordinate is straightforward.

A genetic algorithm for multivariate isotonic regression

Isotonic regression problems under partial order restrictions are usually formulated as a minimizing, problem with a multimodal objective function subject to partial order restrictions. Some numerical methods to this kind of problems have been reported when some additional conditions are imposed. However, the classic methods (for example, pool adjacent violators algorithm) are usually not suited to multivariate isotonic regression problems. In order to obtain the global optimal parameters of multivariate isotonic regression, this paper presents a genetic algorithm and illustrates its effectiveness by some numerical examples.

Flexible molecular models for molecular dynamics study of near and supercritical water

Four flexible water models based on the harmonic and a modified anharmonic form of Toukan and Rahman intramolecular potential models grafted onto 3-site and 4-site intermolecular models were proposed, and the influence of their flexibility and intermolecular potential on the phase behavior were examined. A global optimization of the potential parameters was performed according to the physical properties and IR spectra of water at the ambient state. The critical properties of the four models were estimated by using direct molecular dynamics (MD) simulation of liquid–vapor coexisting phases and fitting the coexisting densities to the law of rectilinear diameter and the scaling law for density with a scaling exponent of β c=0.325. Based on a similar intermolecular model (the simple point charge model, SPC) we found that the anharmonic model gave a critical point ( T c=370.3±3.6°C; ρ c=0.32±0.006 g/cm 3) which agreed well with the experimental properties ( T c=374.0°C; ρ c=0.322 g/cm 3), while the harmonic model gave low values ( T c=337.3±8.5°C; ρ c=0.290±0.007 g/cm 3). On the other hand, for anharmonic flexible water models adopting the experimental value, the model grafted onto a 3-site TIPS model (the transferable intermolecular potential functions, TIP3P) did not provide satisfactory reproduction of the critical point ( T c=320.1±5.2°C; ρ c=0.288±0.005 g/cm 3), while the model grafted onto a 4-site TIPS model (modified TIP4P model with M-site on the center of mass, referred to as cm4P) provided good representation of critical properties ( T c=368.4±6.2°C; ρ c=0.307±0.005 g/cm 3). The flexible models with anharmonic form of flexibility grafted the SPC or cm4P model are suitable model for study of water and aqueous solutions in state points ranging from liquid to near- and supercritical region.

Multidisciplinary optimization in aircraft design using analytic technology models

An approach to multidisciplinary optimization is presented which combines the Global Sensitivity Equation method, parametric optimization, and analytic technology models. The result is a powerful yet simple procedure for identifying key design issues. It can be used both to investigate technology integration issues very early in the design cycle, and to establish the information flow framework between disciplines for use in multidisciplinary optimization projects using much more computational intense representations of each technology. To illustrate the approach, an examination of the optimization of a short takeoff heavy transport aircraft is presented for numerous combinations of performance and technology constraints.

Generalization ability of optimal cluster separation networks

Optimal separation of two clusters of normalized vectors can be performed in a neural network with adjustable threshold and weights, which is trained to maximum stability. Generalization from arbitrarily selected training clusters to a given bipartitioning of input space is studied. The network's threshold becomes a global optimization (and order) parameter, This causes the generalization ability to increase rapidly with the distance of the cluster separation plane from the origin. Separation is shown to be stochastic for small and deterministic for large training cluster sizes.

A method for modeling the manufacturability of IC designs

A methodology for modeling the manufacturability of MOS circuits has been developed. The fabrication line is described using a small set of measurable process parameters, whose variation explains the range of circuit performance seen during production. These same parameters form the basis of a statistical MOSFET model which combines physical measurements, global optimization, and regression modeling of key fitting parameters to accurately predict transistor characteristics over a wide range of process variation. The fabrication line description in conjunction with the MOSFET model was used to develop a manufacturing application, specifically, a performance prediction model which uses the process parameters as measured on the manufacturing floor to predict the performance of fabricated integrated circuits before packaging and final test. The MOSFET model and the performance prediction model are integrated, and data taken from the manufacturing line can be used to verify the models, to identify process shifts, and suggest design improvements for further manufacturability enhancements. The method was successfully applied to an industrial 1.5-/spl mu/m CMOS process, and models were developed and tested for a 1-Mbit EPROM.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Systematic parameter estimation of watershed acidification model

AbstractA joint strategy for parameter estimation which can systematically identify the important model parameters is presented. the strategy includes a regionalized sensitivity analysis (RSA) and an automatic parameter calibration technique (APCT). the RSA is based on a large number of Monte‐Carlo simulations to identify the sensitive parameters and to establish a range of appropriate values for each sensitive parameter. the APCT adjusts the values of the sensitive parameters based on changes in the residual variances between the predicted and observed values. the strategy is applied to the watershed acidification model—ILWAS. the strategy succeeds in identifying the sensitive parameter and increases the likelihood of obtaining a global optimal parameter set. Improvements in the model prediction of the streamflow and chemistry are obtained.

Variation-free iterative method for global optimal control

Abstract A variation-free method for global optimal control is proposed. The method is based on a global parameter optimization algorithm and does not employ variational principles or variation-based techniques.

Comparison of two unit cold standby reliability models with three types of repair facilities : K. Murari and Vibha Goyal. Microelectron. Reliab.24 (1), 35 (1984)

This paper considers a multi-repairmen problem comprising of M operating machines with W warm standbys (spares). Both operating and warm standby machines are subject to failures. With a coverage probability c, a failed unit is immediately detected and attended by one of R repairmen if available. If the failed unit is not detected with probability 1−c, the system enters an unsafe state and must be cleared by a reboot action. The repairmen are also subject to failures which result in service (repair) interruptions. The failed repairman resumes service after a random period of time. In addition, the repair rate depends on number of failed machines. The entire system is modeled as a finite-state Markov chain and its steady state distribution is obtained by a recursive matrix approach. The major performance measures are evaluated based on this distribution. Under a cost structure, we propose to use the Quasi-Newton method and probabilistic global search Lausanne method to search for the global optimal system parameters. Numerical examples are presented to demonstrate the effectiveness of our approach in solving a highly complex manufacturing system subject to multiple uncertainties.