Ayca Altay

Fuzzy cognitive mapping in factor elimination: A case study for innovative power and risks

Abstract Factor or criteria prioritization is essential for decision making and planning. In most areas in decision making, integrating the related literature yields an exuberance of criteria which leads a robust decision. Yet, an excess number of criteria may handicap decision making or evaluations in terms of computational time and complexity. In these circumstances, decreasing the number of factors in exchange for a negligible amount of knowledge can emancipate the decision maker yet does not affect the quality of the decision. This elimination can be conducted through qualitative methods such as interviews or quantitative methods. However, quantitative methods are more trustworthy since qualitative methods can be deceptive due to the perceptions of the interviewee. Furthermore, working with larger groups is more prone to neutrality in terms of group thinking. On the subject of innovative power and risks, the literature offers 48 criteria depending on the industry, size or demographics of related companies. Prioritizing and working with these criteria for their decision making applications becomes computationally expensive, especially when embedded in more complex algorithms. In this study, 48 criteria will be reduced using Fuzzy Cognitive Maps and it is believed to provide a sufficient number of criteria with a negligible loss of information and comparisons will be conducted.

Prediction of Aircraft Failure Times Using Artificial Neural Networks and Genetic Algorithms

The aviation industry lived through the global economic crisis by mergers and a review of cost cent factors. Maintaining a fleet is more delicate than ever because low price is the focus of demand, but providing safety is the greatest cost. These facts lead to dynamic maintenance of the aircraft in order to eliminate excessive maintenance costs while ensuring safety. Preemptive maintenance can be run only if the failure times of aircraft are predicted ahead of occurrence. This study aims to predict when the failure will happen by aircraft type and age. An application of 60 aircraft, which lived through 532 failures, is modeled with artificial neural networks and genetic algorithms, which are known to be the preferred forecasting tools. The model proposed gives a good forecast correlation rate between the target and actual failure schedules of aircraft.

Fuzzy Decision Trees

Decision trees are one of the most widely used classification techniques because of their easily understandable representation. In the literature, various methods have been developed to generate useful decision trees. ID3 and SLIQ algorithms are two of the important algorithms generating decision trees. Although they have been applied for various real life problems, they are inadequate to represent ambiguity and vagueness of human thinking and perception. In this study, fuzzy ID3 and fuzzy SLIQ algorithms, which generate fuzzy decision trees, are discussed as well as their enhanced versions. Their performances are also tested using simple training sets from the literature.

Churn detection and prediction in automotive supply industry

Companies have both large certified enterprises and small unauthorized service providers as their competitors in the automotive supply industry. As technology related industries undergo more intensive competition, churn detection and prediction become essential to be precautious about leaving customers. The literature for churn detection offers numerous statistical and intelligent methods. In this study, Artificial Neural Networks and Decision Trees are applied to detect the churn in and analyze the validity of these methods for the automotive supply industry. The problem involves both categorical and continuous numerical decision inputs which cannot simultaneously fed into Decision Trees. In this case, continuous inputs should be divided into binary categorical ones by splitting into various intervals which are called buckets. Particle Swarm Optimization algorithm is implemented for finding optimal buckets for the churn problem data. Results indicate that while both algorithms are promising, the bucket tuning for Decision Trees complicate the churn detection process.

An Intelligent Prediction of Self-produced Energy

The need for energy has been aggressively increasing since the industrial revolution. An exponential growth of industrial and residential power use is encountered with the technological revolution. Cogenerated and self produced energy is a solution that allows the reuse of heat produced, decreases transmission investments, and reduces carbon emissions and decreases dependency on energy resource owners. The mass production sites, health centers, big residential sites and more can use the system. In this chapter, the focus is given to industrial auto-producers. Power market balance is based on the day-ahead declarations; therefore, the production is to be planned in detail to avoid penalties. A recurrent Artificial Neural Network model is constructed in order to predict the day ahead energy supply. The model considers energy resource price, demand from multiple sites, production cost, the amount of energy imported from the grid and the amount of energy exported to the grid. In order to achieve the energy production rate with the least error rate possible, an energy demand forecasting model is constructed for a paper producing company, using a Nonlinear Autoregressive Exogenous Model (NARX) network implemented in Matlab. Three parameters of the forecasting model are tuned using the Particle Swarm Optimization (PSO) algorithm: the number of layers, the number of nodes in hidden layers and the number of delays in the network. Error level is measured using the Minimum Absolute Percentage Error between the predictions and the actual output. Results indicate that NARX is an appropriate tool for forecasting energy demand and the algorithm yields better results when the system parameters are tuned.

Recent Models and Solution Methodologies for Optimization Problems in Supply Chain Management Under Fuzziness

Supply chain (SC) involves collaborating with business partners which uniquely specialize on only a few key strategic activities. The network structures formed in SC’s have emerged in the last decade with the accelerated developments in globalization, outsourcing and information technology. The complex network structures have introduced novel problems to both industry and academia while traditional complications are yet investigated. The intensification points of SC problems are mainly configuration of distribution networks, forming distribution strategies, trade-off analyses, managing inventory and cash-flow. One of the main challenges in modeling and solving these problems is to deal with the uncertainties involved in the complex nature of SC. Demand has been the main uncertain aspect of the problems of the related literature followed by internal parameters, supplier related parameters, environmental parameters and price. The uncertainty issues have been commonly dealt with fuzzy approaches in the literature. Fuzzy approaches become beneficial under uncertainties such as the absence of data, use of qualitative data or the need for subjective judgments. Hence, fuzzy techniques in SC optimization problems are vastly implemented in the literature. The purpose of this study is basically to summarize the fuzzy techniques employed for SC optimization models, their past applications, solutions algorithms and offer directions for future research.

Socio-Effective Value of Bio-Diesel Production

Increasing air pollution in urban areas has accelerated the interest in biodiesel and vehicles that consume biodiesel. As a caution, majority of the developed countries have started using biodiesel in transportation or determined goals and targets for the near future. Brazil has been a pioneer in the field, whereas the European Union has set the objective of utilizing 10 % of all vehicles using biodiesel by 2020. While the utilization and implementation of biodiesel-based systems severely contribute to economical and environmental savings, the antecedent production process has its own adverse effects such as the demolishment of agricultural sites. This chapter aims to analyze these effects as well as to propose a model for balancing the trade-offs by minimizing the negative consequences and maximizing the positive ones. The related model involves nonlinear constraints and objectives which are dependent of different uncertain scenarios and expectations. A particle swarm optimization (PSO) and self-organizing maps (SOMs) approach are implemented to attain appropriate solutions of the model. This proposition will also provide a new perspective for both academia and investors in the biodiesel field.

Collective intelligence for evaluating synergy in collaborative innovation

Collaborative innovation is an unavoidable need for the small and medium enterprises (SME) both in terms of economic scale and technological knowledge. Risks and the innovation power are analyzed for the wealth of collaboration. This paper aims to present the synergy indexas a multiplier of the innovation power of research partners to construct a successful collaboration. The proposed index can be used with different number of companies in collaboration cluster and the synergy maximization is guaranteed by using a new particle swarm algorithm, Foraging Search. This paper will give the formulation and criteria of the synergy index in detail. A sample synergy index application for the Turkish SMEs will clarify the steps to follow.

Evolutions in Particle Swarm Optimization: Benchmark on Continuous Cases

This chapter aims to present the results of developing a novel swarm optimization method that responds to the need of using multidimensional parameters. This meta-heuristic optimization approach is inspired by the ecological system of animals and their hierarchical relationship. Animal Food Chain in the nature is known to have three groups: herbivores (plant eaters), omnivores (both plant and meat eaters) and carnivores (meat eaters). In the food pyramid, the number of herbivores is higher than the number of omnivores which is higher than the number of carnivores in ratios depending on the environment. Furthermore, herbivores are known to be slowest and carnivores are known to be the fastest of the chain. These features are represented by conditional and multidimensional parameters in the Foraging Search algorithm. Tests are run on continuous and non-linear benchmark problems with various numbers of constraints. Cross validations of the results are realized by comparison of classical and Predator–Prey based Particle Swarm algorithms. Test results emphasize the power of the Foraging Search as an optimization tool.