Particle Swarm Optimization and Pure Adaptive Search in Finish Goods’ Inventory Management

Abstract

Inventory management deals with a tradeoff between the benefits of keeping stocks of goods that allows fulfillment of the customer’s demand, and the cost of carrying inventory. Inventory control techniques are very important components and the most organizations can substantially reduce their costs associated with the flow of materials. This paper presents new inventory management model based on particle swarm optimization and pure adaptive search global optimization algorithm in production-inventory system. The proposed model is focusing on planned level of demand for finished goods, production and raw materials cost, production capacity as the norm, change of the production cost and inventory capital cost, all of which are typical factors in automobile manufacture industry. The model determines different factors such as the minimizing inventory quantity, minimizing inventory value, and minimizing production cost based on demand for production items. The model is tested with original real-world dataset obtained from the automotive company Lear from US and its factory in Novi Sad, Serbia.     

A magneto-optical setup for studying the time evolution of nanoscale domain-wall displacements under pulsed magnetization

A magneto-optical setup for studying the time evolution of nanoscale domain-wall displacements of domain walls in magnetic films with a regular domain structure is described. Methods for spatial filtering of the output optical flux using the Fourier image of the domain structure and the impulse transient response in the real-time mode are applied in the setup. This allowed an increase in the sensitivity to domain-wall displacements of up to 5 nm and registration of their evolution with a temporal resolution of 1 ns.     

Columnar shaped microresonators for mass detection and gas analysis

In this paper miniaturized silicon columns mounted perpendicular to a surface are presented as microresonators for the detection of an exceedingly small mass. In a first sensor design the column resides on a rigid silicon substrate and is forced to oscillate in the first flexural eigenmode. Mass sensitivity was investigated loading a single latex sphere on the top surface of the column and detecting the frequency shift of the resonant frequency. Oscillation was monitored using the optical beam deflection method, focusing a laser beam at the same surface. However, optical readout was substantially hampered by scattering of the laser beam at the loaded particles and diffraction at the column surface. In a second optimized design a silicon column was mounted on a flexible silicon nitride membrane. In this way the optical readout was applied from the flip side of the membrane completely separating the load area and the optical beam path. A proof of principal is given for the optimized sensor design investigating the influence of pure gases on the extrinsic damping behavior of the membrane columnar oscillator.     

Features of surface modes in metamaterial layers

In this paper, we present a study of the waveguiding properties of metamaterial layers. We pay particular attention to peculiar features that have not yet been given sufficient consideration in previous publications, with an emphasis on the physical interpretation of these results. We attempt to follow the transition of waves guided by a solitary “metamaterial–vacuum” boundary (the so-called true surface wave—TSW) to modes guided by the metamaterial layer. We considered the way TSW of different types (forward, backward and degenerated) transform to the layer modes and examined the features arising in the dispersion curves. For example, “pulling” of the dispersion curve into the frequency region where TSWs do not exist separately, presence of the bend for all modes produced by backward TSWs, existence of the waves with complex conjugate propagation constants, etc.     

Nonlinear dynamical factor analysis for state change detection

Changes in a dynamical process are often detected by monitoring selected indicators directly obtained from the process observations, such as the mean values or variances. Standard change detection algorithms such as the Shewhart control charts or the cumulative sum (CUSUM) algorithm are often based on such first- and second-order statistics. Much better results can be obtained if the dynamical process is properly modeled, for example by a nonlinear state-space model, and then the accuracy of the model is monitored over time. The success of the latter approach depends largely on the quality of the model. In practical applications like industrial processes, the state variables, dynamics, and observation mapping are rarely known accurately. Learning from data must be used; however, methods for the simultaneous estimation of the state and the unknown nonlinear mappings are very limited. We use a novel method of learning a nonlinear state-space model, the nonlinear dynamical factor analysis (NDFA) algorithm. It takes a set of multivariate observations over time and fits blindly a generative dynamical latent variable model, resembling nonlinear independent component analysis. We compare the performance of the model in process change detection to various traditional methods. It is shown that NDFA outperforms the classical methods by a wide margin in a variety of cases where the underlying process dynamics changes.     

Distribution of iron,zinc and manganese in milling streams of common Serbian wheat cultivars: Preliminary survey

This study investigates distribution of three micronutrients in milling streams of three common wheat cultivars to estimate nutritive value of white and brown flours. Among the analysed wheat cultivars, the majority of flour streams of NS Todorka showed a higher content of soluble micronutrients than the other two investigated wheat cultivars. It is necessary to improve bioavailability of Fe and Zn in flours since these micronutrients were mostly in insoluble form. In contrast, considering average Serbian portion sizes for bread, bread made from flour Type ‘110’ could supply almost 75% of the recommended daily intake for Mn.     

Simulation of hydrogen diffusion affected by stress-strain heterogeneity in polycrystalline stainless steel

The effect of microstructure-induced strain-stress heterogeneity on the evolution of hydrogen distribution was studied on 316 L stainless steel. A crystal plasticity-transient hydrogen diffusion finite element analysis was developed. To reach the stress-strain heterogeneity, synthetic microstructures were computed under uniaxial loading. The results show that hydrostatic stress gradients is the main cause of the hydrogen redistribution and the strain rate has a crucial effect on the redistribution degree. Statistical analysis of H-segregation reveals clustering of H-enriched zones at the mesoscopic scale.     

The RT-32 radio telescope pointing system

Quasar-KVO RT-32 radio telescopes of the radio interferometric complex of the Russian Academy of Sciences are equipped with unique 32-m-diameter antennas. The telescopes provide radio astronomy observations in the centimeter wavelength range within a network of radio telescopes, as well as in the single dish mode. The antenna system structure has a significant size and weight, thus determining the engineering difficulties in controlling the movement of the RT-32. The electrical drive of the antenna system must provide two different operating modes: on the one hand, a rapid change in the antenna angular position, and on the other, precise tracking of a cosmic signal source. The high operational load and requirements for the service of radio telescopes as parts of radio interferometry networks impose stringent reliability requirements on the electrical drive and control system. The pointing system consists of subsystems that include dc drives, power equipment for controlling these drives, position sensors based on rotary transformers, and other items. All subsystems are integrated using switching devices of control signals and coordinate-transforming devices. The system is operated by a working monitoring and control station, which is based on an industrial computer and specially developed software.     

Beyond feedforward models trained by backpropagation: a practical training tool for a more efficient universal approximator.

Cellular simultaneous recurrent neural network (SRN) has been shown to be a function approximator more powerful than the multilayer perceptron (MLP). This means that the complexity of MLP would be prohibitively large for some problems while SRN could realize the desired mapping with acceptable computational constraints. The speed of training of complex recurrent networks is crucial to their successful application. This work improves the previous results by training the network with extended Kalman filter (EKF). We implemented a generic cellular SRN (CSRN) and applied it for solving two challenging problems: 2-D maze navigation and a subset of the connectedness problem. The speed of convergence has been improved by several orders of magnitude in comparison with the earlier results in the case of maze navigation, and superior generalization has been demonstrated in the case of connectedness. The implications of this improvements are discussed.