Parameter Search Method Research Articles

Calculation of Parameters of Equivalent Circuit of Permanent Magnet Synchronous Motor Using PSO with Grouped Agents

This paper presents a calculation method of parameters of high frequency equivalent circuits of inverter-driven permanent magnet synchronous motors (PMSM). The high frequency oscillatory current such as leakage current to the motor frame (common-mode current) and high frequency oscillatory line current (normal-mode current), are caused due to switching of inverter transistors. A parameter search method based on the particle swarm optimization algorithm is proposed. Agents in the algorithm are grouped so that convergence performance is improved. Some simulation results illustrate the effectiveness of the proposed algorithm.

The role of action potential shape and parameter constraints in optimization of compartment models

Automated parameter search methods are commonly used to optimize compartment model parameters. An important step in parameter fitting is selecting an objective function that represents key differences between model and experimental data. We construct an objective function that includes both time-aligned action potential shape error and errors in firing rate and firing regularity. We then implement a variant of simulated annealing that introduces a recentering algorithm to handle infeasible points outside the boundary constraints. We show how our objective function captures essential features of neuronal firing patterns, and why our boundary management technique is superior to previous approaches.

Constraining Compartmental Models Using Multiple Voltage Recordings and Genetic Algorithms

Compartmental models with many nonlinearly and nonhomogeneous distributions of voltage-gated conductances are routinely used to investigate the physiology of complex neurons. However, the number of loosely constrained parameters makes manually constructing the desired model a daunting if not impossible task. Recently, progress has been made using automated parameter search methods, such as genetic algorithms (GAs). However, these methods have been applied to somatically recorded action potentials using relatively simple target functions. Using a genetic minimization algorithm and a reduced compartmental model based on a previously published model of layer 5 neocortical pyramidal neurons we compared the efficacy of five cost functions (based on the waveform of the membrane potential, the interspike interval, trajectory density, and their combinations) to constrain the model. When the model was constrained using somatic recordings only, a combined cost function was found to be the most effective. This combined cost function was then applied to investigate the contribution of dendritic and axonal recordings to the ability of the GA to constrain the model. The more recording locations from the dendrite and the axon that were added to the data set the better was the genetic minimization algorithm able to constrain the compartmental model. Based on these simulations we propose an experimental scheme that, in combination with a genetic minimization algorithm, may be used to constrain compartmental models of neurons.

Estimation of Fried's Parameter from Long-Exposure Solar Images

We propose a criterion for extending the parameter search method (Krishnakumar and Venkatakrishnan, 1997) of estimating the point spread function to solar data. In the parameter search method, the number of pixels with negative intensity values in the restored object is used as an estimator for determining the unknown parameters of the point spread function. As a solar image has a high background, the restored object does not contain negative values, thereby making the method unsuitable for solar data. We propose to use the intrinsic contrast of solar features as a criterion for identifying the unknown parameter. We validate our method through simulations. This method can not be used for image restoration but can be used for monitoring daytime seeing.

906 複数のモードを考慮した構造物の振動制御

This paper suggests an active vibration control with consideration of multiple vibration modes a structure is examined by proposing an autonomous active vibration control system. In general, an active vibration control needs the vibration analysis of the controlled system and the design of the control system which considers the controlled object. In addition, it is necessary to consider problems such as secular distortions and unstable spillovers of the controlled system. And design techniques and some trial-and-error are required for the design of the control system. A solution of those various problems is confirmed by this method applying the autonomous active vibration control by using parameter search method.

A comparative survey of automated parameter-search methods for compartmental neural models.

One of the most difficult and time-consuming aspects of building compartmental models of single neurons is assigning values to free parameters to make models match experimental data. Automated parameter-search methods potentially represent a more rapid and less labor-intensive alternative to choosing parameters manually. Here we compare the performance of four different parameter-search methods on several single-neuron models. The methods compared are conjugate-gradient descent, genetic algorithms, simulated annealing, and stochastic search. Each method has been tested on five different neuronal models ranging from simple models with between 3 and 15 parameters to a realistic pyramidal cell model with 23 parameters. The results demonstrate that genetic algorithms and simulated annealing are generally the most effective methods. Simulated annealing was overwhelmingly the most effective method for simple models with small numbers of parameters, but the genetic algorithm method was equally effective for more complex models with larger numbers of parameters. The discussion considers possible explanations for these results and makes several specific recommendations for the use of parameter searches on neuronal models.

Experiments with dynamic programming algorithms for nonseparable problems

In this paper we report on numerical experiments with dynamic programming algorithms designed for solving problems with nonseparable objective functions of the C-programming type. The experiments were conducted with the view to evaluate the efficiency of C-programming's exclusionary rules when they are incorporated in the conventional parametric Lagrangian search method. Two classical nonseparable problems are considered: (1) the variance minimization problem and (2) the nonlinear fractional knapsack problem. The results indicate that the incorporation of C-programming's exclusionary rules in the Lagrangian procedure significantly improves the efficiency of the line search procedure.

Modelling Bovine Tuberculosis in Badgers

We use deterministic models to study the spread of bovine tuberculosis (Mycobacterium bovis) infection within badger Meles meles subpopulations. We obtain analytical results for the models and adopt an easy-to-use approach―the Logical Parameter Search (LPS) Method―to generate appropriate parameter values for the model equations. We compare our computer-generated model parameter values with those obtained from the literature and present numerical simulations for the model system, some of which mimic the observed cyclic trend in disease incidence and prevalence as suggested, for example, by Cheeseman et al. (1989)

Hough space transformations for discrimination and distortion estimation

A new use of the Hough transform space defined for straight lines is described. The Hough space is used directly with new efficient distortion parameter transformations and template matching. This technique allows multiclass discrimination, intra-class distortion invariant recognition, and multiple distortion parameter estimation. A new hierarchical distortion parameter search method and spatial quantization in Hough space make realization of this technique very attractive. Performance of our algorithm on aircraft imagery and in the presence of noise is provided.

Lunar seismology: The internal structure of the Moon

Tables A1–A4 and Figures A1–A3 are available with entire articleon microfiche. Order from American Geophysical Union, 2000 Florida Avenue, N.W., Washington, D.C. 20009. Document J81‐003;$1.00. Payment must accompany order.The arrival times of direct P and S waves, measured on seismograms recorded from natural lunar seismic events, have been analyzed using linearized inversion and parameter search methods to simultaneously determine event locations, origin times, and structural parameters (seismic velocities). Polarization‐filtered record section plots are correlated with theoretical travel time curves to identify later phases and obtain additional structural and velocity information. Shear wave amplitudes plotted as a function of distance provide data on the existence and magnitude of seismic velocity gradients in the interior. These studies are used to delineate the structure of the moon below the crust to a depth of about 1100 km. This structure has been divided into an upper and a lower mantle. The upper mantle, from 60‐ to 400‐km depth, has average seismic wave velocities of Vp = 7.70 ±0.15 km/s and Vs = 4.45 ± 0.05 km/s with corresponding Q values (taken from the literature) of Qp ∼ 5000 and Qs ∼ 3000. The shear wave velocity decreases with a negative gradient of at most −6 × 10−4 km/s/km, implying a Vs variation of from 4.57 km/s at 60‐km depth to 4.37 km/s at 400‐km depth; this decrease can be accounted for by increasing temperature, and so no major compositional gradient is required in the upper mantle. A small negative P wave velocity gradient may also be present. Between 400‐ and 480‐km depth there is a transition zone with a sharply decreasing shear wave velocity and possibly an accompanying small decrease in Vp. The dominant shear velocity decrease may occur at a 480‐km interface and may represent a compositional change, although the effects of increased temperature cannot be totally ruled out. The lower mantle extends from 480‐km depth to at least 1100‐km depth; few seismic waves recorded by the lunar network penetrate below 1100 km. The average velocities are Vp = 7.6 ± 0.6 km/s and Vs = 4.2 ± 0.1 km/s. Q values (taken from the literature) are Qp ∼ 1500 and Qs ∼ 1000. Below 1100 km there is some indication that the attenuation may increase still further for shear waves, with Qs dropping to a few hundreds or less. This seismic model of the moon is well constrained, with uncertainties on the above values given explicitly by the analysis methods, and so it serves as a strong control on the present‐day lunar compositional and thermal structure.