Guor-Rurng Lii

Long-term generation expansion planning employing dynamic programming and fuzzy techniques

This research proposes a new method for long-term generation expansion planning. The method adopts a multi-aspect optimal approach which considers the capital cost of the newly added units, the maintenance and fuel costs, environmental impact, reliability, etc. To accommodate the growth of power load, the generation capacity needs to expand to meet the load demand. In order to find an optimal alternative to increase the generation capacity and satisfy different constraints economically and efficiently, the optimization technique is employed. The dynamic programming (DP) as the optimization method is used in this study. Since the requirements of environmental standard and power quality are getting more and more strict, economical factors are no more the unique one to weigh for the generation expansion planning. The environmental protection and reliability are also important factors of the problem. However, types of pollution are very complicated and are not easy to incorporate into the solution model. In this research, we apply the fuzzy theory to represent the state of pollution and judge if a combination of investment is acceptable or not. Moreover by employing the fuzzy technique, we can delete a lot of unnecessary paths and states to reduce the computation burden of DP. Finally we use an example to illustrate and prove the applicability and validity of the presented approach.

Reliability design of distribution systems using modified genetic algorithms

Abstract This investigation presents an evolutionary algorithm to reliably design a distribution system. Adopted reliability indices include failure rate and interruption duration, which are both commonly used in distribution systems. The total cost including the apparatus investment cost and the system interruption cost, is the objective function to be minimized. Reliability constraints on a load point are addressed to ensure the adequacy of the power supply of the load point. The evolutionary optimization algorithm is a modified genetic algorithm (GA), which uses binary and floating-point representations. The proposed GA is compared with the conventional generalized reduced gradient (GRG) method. A secondary substation of the Taiwan Power Company is used in an example of the methods application. The discussed method is simple and effective and is useful for expanding the existing systems and constructing new systems.

Optimal capacitor allocation using fuzzy reasoning and genetic algorithms for distribution systems

This paper presents an optimal capacitor allocation method which uses fuzzy reasoning and genetic algorithms for primary distribution systems. In the method, capacitor allocation is applied to correct voltage deviation and reduce power loss for a given load pattern. The problem of capacitor allocation includes determining the location, type (fixed or switched), and size of capacitor. Fuzzy reasoning finds the sensitive buses which are used as the candidate locations for capacitors placement. Genetic algorithms determine the size and type of capacitors to be placed in the system. It is a combinatorial optimization problem which has the objective composed of power losses and capacitor installation costs subject to bus voltage constraints. Three membership functions relating to voltage deviation, real power loss, and reactive power loss are defined for fuzzy application. The proposed approach is demonstrated using two example systems. Computational results show that the proposed method can quickly achieve an optimal or near-optimal solution.

Reliability planning for composite electric power systems

This paper proposes a reliability design approach using network flow technique, genetic algorithms, and Monte Carlo simulation for composite electric power systems. With increased emphasis on reliability design and cost control in electric power system planning and operation, particularly in composite electric power system, we are therefore striving to achieve an optimal reliability design solution under a reliability/cost implemented model. Because the floating-point representation is more efficient than the binary representation in genetic algorithms application, and the former also has more robust operators to locate near optimal solutions in most cases, we will employ the floating-point representation. The proposed method primarily finds out the optimal values of reliability indices for the components such that the objective function composed of interruption cost and installation cost is minimized. The reliability indices adopted include expected demand not served (EDNS) and forced outage rate (FOR). Application of the proposed method is demonstrated using a 23-bus test system.

An induction motor position controller optimally designed with fuzzy phase-plane control and genetic algorithms

Abstract A new controller that employs fuzzy phase-plane control (FPPC) and genetic algorithms (GAs) is presented herein. For optimal position control of an induction motor, it is equipped with a modified evolutionary direction operator (MEDO). As the FPPC technique is applied, the proposed controller has many advantages, such as satisfactory control performance under a wide range of operating conditions is obtained, multifarious expert experiences and defuzziness are not required, and it has a quicker response than conventional fuzzy controllers do. GAs optimize the parameters of fuzzy phase-plane function, and the MEDO determines a better evolutionary direction for solution searching. The proposed technique was successfully developed and applied to the position control of an induction motor. Simulation results confirmed that the approach is computationally efficient and has excellent control performance.

Reliability planning employing genetic algorithms for an electric power system

This paper applies network flow method, genetic algorithms, and Monte Carlo simulation to optimal reliability design for a composite electric power system. Genetic algorithms are general purpose optimization techniques based on principles inspired from the biological evolution using three main operations of reproduction, crossover, and mutation, which could locate near optimal solutions in most cases. The proposed method primarily adopted Monte Carlo simulation method, maximum-flow minimum-cut theorem, and optimization techniques to find out the optimal values of reliability indices, such that the optimal reliability design for the system can be achieved. The objective function to be optimized is composed of interruption cost and installation cost. The reliability indices mainly used include expected demand not served (EDNS) and forced outage rate (FOR). An application of the proposed method conducted on an IEEE five-bus test system is presented.

Fuzzy logic based voltage control for a synchronous generator

Abstract This paper proposes a novel method using the fuzzy technique and gain tuning for constant voltage control of synchronous generators. The gain tuning technique of the control scheme is designed to be applicable to various load conditions. The fuzzy controller, implemented by two lookup tables for coarse control and fine control, applies the method of gain tuning to quickly converge the terminal voltage of a generator to the set value. The controller has a wider range of operating conditions and smaller settling time compared with that of the conventional controllers using fixed gain. The proposed method was demonstrated using a 50 kVA synchronous generator. Simulation results show that the approach is computationally efficient and has good control performances. Furthermore, because of the use of the fuzzy technique, no transfer function development is required for the implementation of the method. Thus the proposed controller has the advantages of ease of design and flexibility.

POSITION CONTROL EMPLOYING FUZZY-SLIDING MODE AND GENETIC ALGORITHMS WITH A MODIFIED EVOLUTIONARY DIRECTION OPERATOR

This study presents a novel controller by employing fuzzy-sliding mode control and genetic algorithms equipped with a modified evolutionary direction operator for optimal position control with an induction motor. Applying the sliding mode control technique provides the proposed controller with many advantages, such as a small overshoot, tiny steady-state error, rapid response, and adaptability to equipment parameters variation and external disturbance. The genetic algorithm optimizes the parameters of fuzzy membership functions defined according to expert experience, and its supplementary modified evolutionary direction operator determines improved search directions. The proposed approach is applied to a position servo system. Computational results indicate that the proposed controller exhibits excellent performance with all the advantages mentioned above. Moreover, the proposed controller is immune to the chattering problem that typically affects general controllers.This study presents a novel controller by employing fuzzy-sliding mode control and genetic algorithms equipped with a modified evolutionary direction operator for optimal position control with an induction motor. Applying the sliding mode control technique provides the proposed controller with many advantages, such as a small overshoot, tiny steady-state error, rapid response, and adaptability to equipment parameters variation and external disturbance. The genetic algorithm optimizes the parameters of fuzzy membership functions defined according to expert experience, and its supplementary modified evolutionary direction operator determines improved search directions. The proposed approach is applied to a position servo system. Computational results indicate that the proposed controller exhibits excellent performance with all the advantages mentioned above. Moreover, the proposed controller is immune to the chattering problem that typically affects general controllers.

A neuro-fuzzy method for tracking control

The purpose of this paper is to propose a new approach to be used in optimal position control. This method uses fuzzy control system and works with genetic algorithms (GAs) to meet the requirement of optimal position control. Based on the unsupervised training of self-organizing neural network, the fuzzy expert experiences are learned. The neuro-fuzzy controller (NFC) then applies, these experiences learned to determine the output control force. By virtue of the evolution rule of genetic algorithms, the best expert experiences are extracted and employed to achieve the optimal position control. Application of the proposed method to the inverted pendulum system is also presented. The simulation results show that the controller has satisfactory performance.