Hybrid method based on neural networks and Monte Carlo simulation in view of a tradeoff between accuracy and computational time

2020 ◽  
pp. 1-10
Author(s):  
Yacin Jerbi ◽  
Samira Chaabene
Keyword(s):  
Neural Networks ◽  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Hybrid Method ◽  
Computational Time

Ship Damage Stability Approval Document Generation by a Amonte Carlo Method

2021 ◽  
Author(s):  
Stefan Krüger ◽  
Katja Aschenberg
Keyword(s):  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Information Needs ◽  
Computational Time ◽  
Strong Impact ◽  
Design Phase ◽  
Efficient Design ◽  
Stability Problems ◽  
Initial Design ◽  
Damage Stability

Abstract The revised SOLAS 2020 damage stability regulations have a strong impact on possible future ship designs. To cope with these requirements, damage stability investigations must become a central part of the initial design phase, and many internal subdivision concepts need to be investigated. Unfortunately, if damage stability calculations are performed in the classical way, they are very time consuming with respect to modelling and computational time. This fact has impeded the consequent subdivision optimization in the past. Therefore, a simulation procedure for damage stability problems was developed which treats damage stability as a stochastic process which was modeled by a Monte Carlo simulation. If statistical damage distributions are once known, the Monte Carlo simulation delivers a population of damages which can be automatically related to certain damage cases. These damage cases can then be investigated with respect to their survivability. Applying this principle to damage stability problems reduces the computational effort drastically where at the same time no more manual modelling is required. This development does especially support the initial design phase of the compartmentation and leads to a safer and more efficient design. If this very efficient simulation principle shall now also be used after the initial design phase for the generation of approval documents, additional information needs to be generated by the simulation method which is not directly obtained during the simulation: This includes detailed individual probabilities in all three directions and the integration of all damage cases into predefined damage zones. This results in fact in a kind of reverse engineering of the manual damage stability process to automatically obtain this required information. It can be demonstrated that the time to obtain the final documents for the damage stability approval can be drastically reduced by implementing this principle.

The Use of Quantitative Methods in Investment Decisions

2019 ◽  
pp. 256-275 ◽  
Author(s):  
Serkan Eti
Keyword(s):  
Neural Networks ◽  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Artificial Neural Networks ◽  
Decision Tree ◽  
Quantitative Methods ◽  
Simulation Analysis ◽  
Investment Decision ◽  
Tree Algorithms ◽  
Artificial Neural

Quantitative methods are mainly preferred in the literature. The main purpose of this chapter is to evaluate the usage of quantitative methods in the subject of the investment decision. Within this framework, the studies related to the investment decision in which quantitative methods are taken into consideration. As for the quantitative methods, probit, logit, decision tree algorithms, artificial neural networks methods, Monte Carlo simulation, and MARS approaches are taken into consideration. The findings show that MARS methodology provides a more accurate results in comparison with other techniques. In addition to this situation, it is also concluded that probit and logit methodologies were less preferred in comparison with decision tree algorithms, artificial neural networks methods, and Monte Carlo simulation analysis, especially in the last studies. Therefore, it is recommended that a new evaluation for investment analysis can be performed with MARS method because it is understood that this approach provides better results.

Supply Chain Inventory Coordination under Uncertain Demand via Combining Monte Carlo Simulation and Fitness Inheritance PSO

2015 ◽  
Vol 6 (1) ◽  
pp. 1-22
Author(s):  
Heting Cao ◽  
Xingquan Zuo
Keyword(s):  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Distribution Function ◽  
Supply Chain ◽  
Simulation Model ◽  
Uncertain Demand ◽  
Computational Time ◽  
Arbitrary Distribution ◽  
Demand Distribution ◽  
Inventory Coordination

Supply chain coordination consists of multiple aspects, among which inventory coordination is the most widely used in practice. Inventory coordination is challenging due to the uncertainty of customers' demand. Existing researches typically assume that the demand is either a deterministic constant or a stochastic variable following a known distribution function. However, the former cannot reflect the practical costumers' demand, and the later make the model inaccurate when the demand distribution is ambiguous or highly variable. In this paper, the authors propose a Monte Carlo simulation model of such problem, which can mimic the inventory changing procedure of a supply chain with uncertain demand following an arbitrary distribution function. Then, a PSO is combined with the simulation model to achieve a coordination decision scheme to minimize the total inventory cost. Experiments show that their approach is able to produce a high quality solution within a short computational time and outperforms comparative approaches.

Sign in / Sign up

Export Citation Format

Share Document