A Credit Card Fraud Model Prediction Method Based on Penalty Factor Optimization AWTadaboost

With the popularity of online payment, how to perform creditcard fraud detection more accurately has also become a hot issue. And withthe emergence of the adaptive boosting algorithm (Adaboost), credit cardfraud detection has started to use this method in large numbers, but thetraditional Adaboost is prone to overfitting in the presence of noisy samples.Therefore, in order to alleviate this phenomenon, this paper proposes a newidea: using the number of consecutive sample misclassifications to determinethe noisy samples, while constructing a penalty factor to reconstruct thesample weight assignment. Firstly, the theoretical analysis shows that thetraditional Adaboost method is overfitting in a noisy training set, which leadsto the degradation of classification accuracy. To this end, the penalty factorconstructed by the number of consecutive misclassifications of samples isused to reconstruct the sample weight assignment to prevent the classifierfrom over-focusing on noisy samples, and its reasonableness is demonstrated.Then, by comparing the penalty strength of the three different penalty factorsproposed in this paper, a more reasonable penalty factor is selected.Meanwhile, in order to make the constructed model more in line with theactual requirements on training time consumption, the Adaboost algorithmwith adaptive weight trimming (AWTAdaboost) is used in this paper, so thepenalty factor-based AWTAdaboost (PF_AWTAdaboost) is finally obtained.Finally, PF_AWTAdaboost is experimentally validated against other traditionalmachine learning algorithms on credit card fraud datasets and otherdatasets. The results show that the PF_AWTAdaboost method has betterperformance, including detection accuracy, model recall and robustness, thanother methods on the credit card fraud dataset. And the PF_AWTAdaboostmethod also shows excellent generalization performance on other datasets.From the experimental results, it is shown that the PF_AWTAdaboost algorithmhas better classification performance.

Elastoplastic analysis of solid structures using penalty-based couple stress finite element method within framework of Cosserat continuum

To obviate the complexities of the straight forward couple stress finite element method,the penalty-based couple stress finite element method(named PcouFEM)within the framework of the Cosserat continuum is utilized to obtain the approximate solution by relaxing the C1 continuity.To examine the performance of the PcouFEM,three well known numerical examples are investigated.For the analysis on stress concentration around the circular hole of the plane strain specimen,it was found that as long as the penalty factor G_(c) is not less than 5 times the shear modulus of the classical continuum G(i.e.,G_(c)≥5G),the stress concentration factors calculated by the PcouFEM with the reduced integration scheme agree well with the analytical solutions.For the strain localization analysis in the uniaxial compression test,it was observed that by applying the PcouFEM,the pathologically mesh-dependent problem associated with the conventional FEM can be alleviated or even removed,and based on numerical simulations,it is recommended to define 5G≤G_(c)≤10G from the perspective of numerical accuracy.For the soil slope subjected to an eccentric load through the rigid strip footing,it was found that the mesh-dependent problem of the shear band simulation can be largely alleviated by applying the PcouFEM.

In Tube Condensation:Changing the Pressure Drop into a Temperature Difference for a Wire-on-Tube Heat Exchanger

A theoretical study based on the Penalty factor(PF)method by Cavallini et al.is conducted to show that the pressure drop occurring in a wire-on-tube heat exchanger can be converted into a temperature difference for two types of refrigerants R-134a and R-600a typically used for charging refrigerators and freezers.The following conditions are considered:stratified or stratified-wavyflow condensation occurring inside the smooth tube of a wire-on-tube condenser with diameter 3.25,4.83,and 6.299 mm,condensation temperatures 35℃,45℃,and 54.4℃ and cover refrigerant massflow rate spanning the interval from 1 to 7 kg/hr.The results show that the PF variation is not linear with vapor quality and attains a maximum when the vapor quality is 0.2 and 0.18 for the R-134a and R-600a refrigerants,respectively.The PF increases with the refrigerant massflow rate if the inner diameter and saturation temperature constant,and it decreases on increasing the inner diameter to 6.299 mm for constant refrigerant massflow rate and saturation temperature.The PF for R-600a is higher than that for R-134a due to the lower saturation pressure in thefirst case.Furthermore,a stratifiedflow produces higher PF in comparison to the annularflow due to the effect of the surface tension.

LARGE DEFORMATION OF INCOMPRFSSIBLE RUBBER CYLINDER UNDER PLANE STRAIN

The large deformation of incompressible rubber cylinder under inner pressure is ana- lyzed by a kind of new rubber materials strain energy function.The theory formulation for the dis- placement and stress is presented.The penalty finite element formulation is established in order to ana- lyze nonlinear rubber materials,and the results of finite element method agree well with theoretical ones.A new method for controlling the calculating stability and convergence rates is put forward.The selection of the appropriate penalty factor and its influence on calculated results are discussed.

Fully Distributed Risk-based Robust Reserve Scheduling for Bulk Hybrid AC-DC Systems

To enhance the cost-effectiveness of bulk hybrid AC-DC power systems and promote wind consumption,this paper proposes a two-stage risk-based robust reserve scheduling(RRRS)model.Different from traditional robust optimization,the proposed model applies an adjustable uncertainty set rather than a fixed one.Thereby,the operational risk is optimized together with the dispatch schedules,with a reasonable admissible region of wind power obtained correspondingly.In addition,both the operational base point and adjustment capacity of tielines are optimized in the RRRS model,which enables reserve sharing among the connected areas to handle the significant wind uncertainties.Based on the alternating direction method of multipliers(ADMM),a fully distributed framework is presented to solve the RRRS model in a distributed way.A dynamic penalty factor adjustment strategy(DPA)is also developed and applied to enhance its convergence properties.Since only limited information needs to be exchanged during the solution process,the communication burden is reduced and regional information is protected.Case studies on the 2-area 12-bus system and 3-area 354-bus system illustrate the effectiveness of the proposed model and approach.

Simulation methodology development for rotating blade containment analysis

An experimental and numerical investigation on the aeroengine blade/case containment analysis is presented. Blade out containment capability analysis is an essential step in the new aeroengine design, but containment tests are time-consuming and incur significant costs; thus, developing a short-period and low-cost numerical method is warranted. Using explicit nonlinear dynamic finite element analysis software, the present study numerically investigated the high-speed impact process for simulated blade containment tests which were carried out on high-speed spin testing facility. A number of simulations were conducted using finite element models with different mesh sizes and different values of both the contact penalty factor and the friction coefficient. Detailed comparisons between the experimental and numerical results reveal that the mesh size and the friction coefficient have a considerable impact on the results produced. It is shown that a finer mesh will predict lower containment capability of the case, which is closer to the test data. A larger value of the friction coefficient also predicts lower containment capability. However, the contact penalty factor has little effect on the simulation results if it is large enough to avoid false penetration.