Ruholla Jafari-Marandi

Fluid Genetic Algorithm (FGA)

Abstract Genetic Algorithm (GA) has been one of the most popular methods for many challenging optimization problems when exact approaches are too computationally expensive. A review of the literature shows extensive research attempting to adapt and develop the standard GA. Nevertheless, the essence of GA which consists of concepts such as chromosomes, individuals, crossover, mutation, and others rarely has been the focus of recent researchers. In this paper method, Fluid Genetic Algorithm (FGA), some of these concepts are changed, removed, and furthermore, new concepts are introduced. The performance of GA and FGA are compared through seven benchmark functions. FGA not only shows a better success rate and better convergence control, but it can be applied to a wider range of problems including multi-objective and multi-level problems. Also, the application of FGA for a real engineering problem, Quadric Assignment Problem (AQP), is shown and experienced.

Self-Organizing and Error Driven (SOED) artificial neural network for smarter classifications

Abstract Classification tasks are an integral part of science, industry, business, and health care systems; being such a pervasive technique, its smallest improvement is valuable. Artificial Neural Network (ANN) is one of the strongest techniques used in many disciplines for classification. The ANN technique suffers from drawbacks such as intransparency in spite of its high prediction power. In this paper, motivated by learning styles in human brains, ANN’s shortcomings are assuaged and its prediction power is improved. Self-Organizing Map (SOM), an ANN variation which has strong unsupervised power, and Feedforward ANN, traditionally used for classification tasks, are hybridized to solidify their benefits and help remove their limitations. The proposed method, which we name Self-Organizing Error-Driven (SOED) Artificial Neural Network, shows significant improvements in comparison with usual ANNs. We show SOED is a more accurate, more reliable, and more transparent technique through experimentation with five different datasets.

A System of System Approach for Smart Complex Energy System Operation Decision

The emerging technology in smart grids enables a bi-directional communication between buildings and power grids. Not only can a building request energy from power grid, but it is also able to sell surplus energy back to power grid or to its neighborhood buildings. The vision is that multiple residential houses can freely form a smart complex to share energy and exchange information, which is expected to reduce energy consumption and achieve significant energy cost savings. However, the study of the operation decisions of a smart complex with multiple residential buildings is limited in the literature. To address this research gap, this paper proposes a System of Systems (SoS) approach to investigate the smart complex operation that considers a residential complex setting where all of the households are connected to one another through the complex virtual decision making level, while maintaining their preferred comfort level. The core objective of the proposed model is to minimize the cost of energy for all of the households in the complex which can freely share energy and exchange information. Each household has the flexibility to maintain its comfort level while minimizing the energy cost. Two mathematical models are presented: (i) at complex level, and (ii) at household level. Although each household is given the freewill to set its comfort levels while optimizing energy costs, the interconnection of these houses with shared renewable energy systems and a complex size battery can boost the load shifting. To derive the operation decision for the smart complex, a genetic algorithm (GA) is developed, and a case study is used to test the efficiency of this GA. It was found that the GA provides acceptable levels of convergence within a reasonable time frame, thereby exhibiting potential for use in real-time decision making in the smart grid context.Copyright

Analyzing the robustness of an array of wireless access points to mobile jammers

A risk metric for a wireless network under attack by jammers is proposed.A multi-period mixed integer programming interdiction model is developed.Signal strength trend is negative-exponential relative to jammers and locations.UCSD dataset indicates a fairly robust network in spite of jamming.Lagrangian relaxation often finds near-optimal solutions for jammer placement. We present an approach for measuring the vulnerability of a wireless network. Our metric, n-Robustness, measures the change in a networks total signal strength resulting from the optimal placement of n jammers by an attacker. Toward this end, we develop a multi-period mixed-integer programming interdiction model that determines the movement of n jammers over a time horizon so as to minimize the total signal strength of users during a sustained jamming attack. We compared several solution approaches for solving our model including a Lagrangian relaxation heuristic, a genetic algorithm, and a stage decomposition heuristic. We tested our approach on a wireless trace dataset developed as part of the Wireless Topology Discovery project at the University of California San Diego. We found that the Lagrangian approach, which performed best overall, finds a close-to-optimal solution while requiring much less time than solving the MIP directly. We then illustrate the behavior of our model on a small example taken from the dataset as well as a set of experiments. Through our experiments we conclude that the total signal power follows a sigmoid curve as we increase the number of jammers and access points. We also found that increasing access points only improves network robustness initially; after that the benefit levels off. In addition, we found that the problem instances we considered have an n-Robustness of between 39 and 69%, indicating that the value of the model parameters (e.g., number of jammers, number of time periods) has an effect on robustness.

An optimum ANN-based breast cancer diagnosis: Bridging gaps between ANN learning and decision-making goals

Abstract It is difficult to overestimate the importance of appropriate breast cancer diagnosis, as the disease ranks second among all cancers that lead to death in women. Many efforts propose data analytic tools that succeed in predicting breast cancer with high accuracy; the literature is abundant with studies that report close-to-perfect prediction rates. This paper shifts the focus of improvement from higher accuracy towards better decision-making. Quantitatively, we have shown more accuracy does not always lead to better decisions, and the process of Artificial Neural Networks (ANN) learning can benefit from the inculcation of decision-making goals. We have proposed a decision-oriented ANN classification method called Life-Sensitive Self-Organizing Error-Driven (LS-SOED), which enhances ANN’s performance in decision-making. LS-SOED combines the supervised and unsupervised learning power of ANN to handle the inconclusive nature of hidden patterns in the data in such way that the best possible decisions are made, i.e. the least misclassification cost (the minimum possible loosing of life) is achieved. The learning power of SOED matches, if not excels, the best performances reported in the literature when the objective is to achieve the highest accuracy. When the objective is to minimize misclassification costs, we have shown, on average, in one dataset more than 30 years of life for a group of 283 people, and in another more than 8 years of life for a group of 57 people can be saved collectively.

A simulated Annealing-based optimal design of STATCOM under unbalanced conditions and faults

This paper presents an approach for optimal design of Static Synchronous Compensator (STATCOM) in electric power systems. The underlying optimization problem is to identify the STATCOM controllers parameters using the Simulated Annealing (SA) optimization technique. The performance of the proposed SA-based controller design is compared with Particle Swarm Optimization (PSO) technique and Genetic Algorithm (GA)-based STATCOM design under different operating conditions and faults. The optimal design of the controller with SA optimization approach provides an acceptable post-disturbance and post-fault performance to recover the system to its normal situation. The advantage of the proposed technique to enhance the voltage profile in steady state operation and under different possible disturbances is confirmed through the time-domain simulations with MATLAB/SIMULINK platform.

Application of STATCOM for MVDC shipboard power system

This paper proposes a solution to deal with the unsymmetrical transient faults in the Medium Voltage DC Shipboard Power System (MVDC SPS), which is employing Static Synchronous Compensator (STATCOM) on the AC load side of the SPS. An improved Genetic Algorithm (FGA) approach is also introduced to design the STATCOMs controller. The performance of the proposed FGA-based controller design compared with the GA-based STATCOM design under different operating conditions and faults. The simulation results confirm that the optimal design of the controller with FGA optimization technique provides an acceptable post-disturbance and post-fault performance to recover the system to its normal situation.

Real-time implementation of MVDC shipboard power system reconfiguration

This paper proposes a novel approach to implement a real-time reconfiguration strategy to minimize the effect of the faults in a Shipboard Power System (SPS). In order to deal with the reconfiguration problem during fault situations in Medium Voltage DC (MVDC) SPS, a real-time Simulated Annealing (SA)-based reconfiguration technique is designed and implemented in the Real-Time Digital Simulator (RTDS). To validate the proposed approach, an MVDC SPS model with four zonal loads is used to simulate several fault scenarios. The simulation results demonstrate the effectiveness of the proposed approach to reconfigure the system under different fault situations.