Application of model order reduction with Cauer ladder networks to industrial inductors

Purpose The Cauer ladder network (CLN) model order reduction (MOR) method is applied to an industrial inductor. This paper aims to to anaylse the influence of different meshes on the CLN method and their parameters. Design/methodology/approach The industrial inductor is simulated with the CLN method for different meshes. Meshes considering skin effect are compared with equidistant meshes. The inductor is also simulated with the eddy current finite element method (ECFEM) for frequencies 1 kHz to 1 MHz. The solution of the CLN method is compared with the ECFEM solutions for the current density in the conductor and the total impedance. Findings The increase of resistance resulting from the skin effect can be modelled with the CLN method, using a uniform mesh with elements much larger than the skin depth. Meshes taking account of the skin depth are only needed if the electromagnetic fields have to be reconstructed. Additionally, the convergence of the impedance is used to define a stopping criterion without the need for a benchmark solution. Originality/value The work shows that the CLN method can generate a network, which is capable of mimicking the increase of resistance usually accompanied by the skin effect without using a mesh that takes the skin depth into account. In addition, the proposed stopping criterion makes it possible to use the CLN method as an a priori MOR technique.

Phase Role in the Non-Uniformity of Main-Line Couplings in Asymmetric Extracted-Pole Inline Filters

The use of classical symmetrical polynomial definition to synthesize fully canonical inline filters with an asymmetrical distribution of the transmission zeros along the topology leads to the occurrence of uneven admittance inverter in the main-line. This form introduces some limitations to transform such topology into a ladder network. Despite circuital transformation can be used to accommodate both technology and topology, it is usual that extra reactive elements are necessary to implement phase shifts required to achieve the complete synthesis. This article introduces a novel method able to determine the required phase correction that has to be applied to the characteristic polynomials in order to equalize all the admittance inverters in the main path to the same value. It has been demonstrated that a suitable pair of phase values can be accurately estimated using a developed hyperbolic model which can be obtained from the transmission and reflection scattering parameters. To experimentally validate the proposed method, a Ladder-type filter with asymmetrical polynomial definition has been synthesized, fabricated, and measured, demonstrating the effectiveness of the developed solution.

Deep Learning Software Defect Prediction Methods for Cloud Environments Research

This paper provides an in-depth study and analysis of software defect prediction methods in a cloud environment and uses a deep learning approach to justify software prediction. A cost penalty term is added to the supervised part of the deep ladder network; that is, the misclassification cost of different classes is added to the model. A cost-sensitive deep ladder network-based software defect prediction model is proposed, which effectively mitigates the negative impact of the class imbalance problem on defect prediction. To address the problem of lack or insufficiency of historical data from the same project, a flow learning-based geodesic cross-project software defect prediction method is proposed. Drawing on data information from other projects, a migration learning approach was used to embed the source and target datasets into a Gaussian manifold. The kernel encapsulates the incremental changes between the differences and commonalities between the two domains. To this point, the subspace is the space of two distributional approximations formed by the source and target data transformations, with traditional in-project software defect classifiers used to predict labels. It is found that real-time defect prediction is more practical because it has a smaller amount of code to review; only individual changes need to be reviewed rather than entire files or packages while making it easier for developers to assign fixes to defects. More importantly, this paper combines deep belief network techniques with real-time defect prediction at a fine-grained level and TCA techniques to deal with data imbalance and proposes an improved deep belief network approach for real-time defect prediction, while trying to change the machine learning classifier underlying DBN for different experimental studies, and the results not only validate the effectiveness of using TCA techniques to solve the data imbalance problem but also show that the defect prediction model learned by the improved method in this paper has better prediction performance.