1
Intelligent Power System Operation and Control: Japan Case Study
In the last twenty years, intelligent systems applications have received increasing attention in various areas of power systems, such as operation, planning, control, and management. Numerous research papers indicate the applicability of intelligent systems to power systems. While many of these systems are still under investigation, there already exist a number of practical implementations of intelligent systems in many countries across the world, such as in Japanese electric utilities.1 In conventional schemes, power system operation, planning, control, and management are based on human experience and mathematical models to find solutions; however, power systems have many uncertainties in practice. Namely, those mathematical models provide only for specific situations of the power systems under respective assumptions. With these assumptions, the solutions of power systems problems are not trivial. Therefore, there exist some limitations for the mathematical-model-based schemes. In order to overcome these limitations, applications of intelligent technologies such as knowledge-based expert systems, fuzzy systems, artificial neural networks, genetic algorithms, Tabu search, and other intelligent technologies have been investigated in a wide area of power system problems to provide a reliable and high-quality power supply at minimum cost. In addition, recent research works indicate that more emphasis has been put on the combined usage of intelligent technologies for further improvement of the operation, control, and management of power systems.
Several surveys on worldwide application of intelligent methodologies on power systems have been recently published.2,3 Considering the experience backgrounds of the authors, the present chapter is focused on the application examples of intelligent technologies in Japanese power system utilities. This chapter is organized as follows: The current situation of intelligent methods application in the Japanese power system in general is described in Section 1.1. Artificial intelligent applications in power system planning and control/ restoration are addressed in Sections 1.2 and 1.3, respectively. Next steps and future implementation are explained in Section 1.4, and finally, the present chapter is summarized in Section 1.5.
1.1 Application of Intelligent Methods to Power Systems
TABLE 1.1
Areas of Intelligent Systems Applications
TABLE 1.2
Applied Intelligent Techniques
Intelligent systems are currently utilized in Japanese utilities as well as other developed countries. Table 1.1 shows the application areas of intelligent technologies in Japanese power system utilities. Many applications have been proposed in literature in those areas to demonstrate the advantages of intelligent systems over conventional systems. A certain number of actual implementations of intelligent systems is already working toward better and more reliable solutions for problems in power systems. Table 1.2 shows the intelligent technologies actually implemented in Japanese utilities for the area of applications shown in Table 1.1.
TABLE 1.3
Objectives of Intelligent Systems Applications
Many applications of other intelligent technologies, such as multiagent systems, simulated annealing, and data mining, have been proposed in literature in those areas of applications. However, up to now, those technologies have not been implemented in actual power systems. The purpose of the intelligent systems applications is classified into several categories, as shown in Table 1.3.
1.2 Application to Power System Planning
1.2.1 Expansion Planning of Distribution Systems
The implemented intelligent systems for expansion planning of distribution systems have been mostly utilized to achieve the following objectives, shown in Table 1.4. The systems are connected to the distribution automation system through the local area computer networks, and the decisions are made by the intelligent systems for reconfiguration of distribution networks and the removal of unnecessary equipment after getting required data from the specified distribution automation system.
After implementation of an intelligent system, the investment for system expansion is reduced because of the efficient utilization of already existing devices and equipment. Detailed and fast evaluations are available for future expansions following the increase the power demand. Loss minimization of the distribution systems has been achieved as the result of the averaging of transformer usages.
TABLE 1.4
Functions of Expansion Planning System and Utilized Intelligent Technologies
1.2.2 Load Forecasting
The intelligent load forecasting system, actually implemented in one of the Japanese utilities, is composed of several neural networks for considering the changes of load configurations, depending on the seasons, such as spring, spring to summer, summer, summer to autumn, autumn, and winter. For the training of each neural network, the records during five years have been utilized, including the maximum power demand, hourly power demand, and weather data. The forecasting procedure of the total power demand is divided into two steps as follows:
Maximum demand is predicted by the system one day ahead based on the weather forecast, including the highest temperature and lowest temperature. The maximum power demands one day before and one week before are also utilized to estimate the maximum power demand of one day ahead.
Based on the maximum power demand in step 1, the hourly power demand is predicted.
By the system, weekly load forecasting is also available. The total power demand is also predicted by the neural networks from the individual load forecasting: light loads such as residential loads and heavy loads such as manufacturing companies and commercial areas. The power loss is also estimated by the neural network. In total, thirty-six neural networks are separately utilized for the detailed load forecasting for the light loads and heavy loads, for different times of 11 a.m., 2 p.m., and 7 p.m., and the seasons. The above procedure can be summarized in the following two steps:
TABLE 1.5
Estimation Error for Daily Power Demand
Power demand is predicted separately for the light loads and heavy loads at 11 a.m., 2 p.m., and 7 p.m. to have the total power demand of one day ahead.
Based on the estimated power demands in step 1, the hourly power demand is predicted by another neural network.
Here, it must be noted that correction of the forecasting by human operators is also available on the implemented load forecasting system.
Table 1.5 shows the accuracy of the load forecasting achieved by the implemented intelligent system based on the neural networks. As shown in the table, the precision of the estimation of power demand is relatively high.
1.2.3 Unit Commitment
For the unit commitment of a group of hydro generators along a river in Japan, an intelligent system has been implemented. The utilized technologies include a rule-based system and intelligent searching scheme. The optimal unit commitment for the grouped hydro generators can be determined based on the water level at each dam, the estimated water flow rate, and so on. All the hydro units should be operated especially during the period when the high fuel cost is expected for the thermal generators. The intelligent system has the functions shown in Table 1.6.
Following the scheduling determined by the intelligent system, economic operation of the hydro units has been achieved. It has been shown through the actual operation of the implemented intelligent system that the scheduling given by it is proper, acceptable, and economical. Further improvement of the system performance and additional rules for operational constraints will be implemented on the current intelligent system.
TABLE 1.6 Types of Scheduling of Hybrid Unit
1.2.4 Maintenance Scheduling
For the optimization of the maintenance scheduling with 4,500 cases yearly and 600 cases monthly at maximum, the implementation of intelligent technologies such as Tabu search, mathematical planning, neural networks, simulated annealing, and others were investigated. Tabu search was finally selected for this purpose. After implementation of the maintenance scheduling system, a certain amount of cost reduction becomes possible and the required time for the maintenance scheduling is shortened.
1.3 Application to Power System Control and Restoration
1.3.1 Fault Diagnosis
An intelligent system for fault diagnosis has also been operated to support the human operators in Japan. The system is composed of two parts: the first part utilizes the rule-based expert system for the classification of the fault types, and the second part utilizes several neural networks associated with the fault types to give their probabilities. The faults are classified into six types, as shown in Table 1.7.
The accuracy of the identification of fault types is shown in Table 1.8. The accuracy is relatively low for the faults, including sleet jump, galloping, and grounding through the construction machines because of the shortage of the faults of those types.
1.3.2 Restoration
An expert-system-based restoration system was installed more than fifteen years ago in Japan, for the automation of the 110 and 220 kV systems. This was one of the earliest implementations of intelligent technologies. Plenty of simulations were performed to acquire knowledge for the expert systems. For the renewal of the system, modifications of programs and the knowledge base, and also the renewal of the computer system, are inevitable. Currently, the system is not operated as an actual operation. However, the system is utilized for the training of operators.
TABLE 1.7
Types of Faults
TABLE 1.8
Accuracy of Fault Identification
1.3.3 Stabilization Control
Over the years, many optimization methodologies, robust techniques, and expert and intelligent systems have been used to stabilize power systems, and to improve the control performance and operational functions of power utilities during normal and abnormal conditions.4 Among these methodologies, fuzzy logic systems have practically attracted more attention. One of the specific features of the fuzzy logic power system stabilizers is their robustness as they provide a wider stable region even for the fixed fuzzy control parameters. Application of fuzzy logic controllers has been proposed mainly for power system stabilizers, and a prototype of a personal-computer-based fuzzy logic power system stabilizer (PSS) was placed in service on a hydro unit in June 1997. The prototype was replaced by a fuzzy logic PSS made by a manufacturer in May 1999.5 The PSS has been working quite well for nearly ten years, including the PC-based prototype stage. Many other applications have also been proposed in the literature, as shown in Table 1.9, but only a few cases have been implemented.
TABLE 1.9
Application of Fuzzy Logic Controllers
TABLE 1.10
Problems for Future Extension of Intelligent Systems in Real Power Systems
1.4 Future Implementations
As shown in the former sections, there already exist a number of implementations of intelligent systems in Japanese utilities. Some of them are now at their renewal stages. However, because of the reasons listed in Table 1.10, the renewals of some of the intelligent systems will be postponed. Most of these obstacles will be solved by further developments of software/hardware technologies.
Currently, the power system operation and control in all aspects, including automatic generation control (AGC), which is the subject of this book, are undergoing fundamental changes due to restructuring, expanding of functionality, the rapidly increasing amount of renewable energy sources, and the emerging of new types of power generation and consumption technologies. This issue opens the way to realize new/powerful intelligent techniques. The infrastructure of the future intelligent power system should effectively support the provision of ancillary services such as an AGC system from various sources through intelligent schemes.
1.5 Summary
This chapter presents the state of the art of intelligent techniques in Japanese utilities based on the investigation of the Subcommittee of the Intelligent Systems Implementations in Power Systems of Japan. The investigation has been completed, and an investigation of some areas, including the area of automatic generation control, where the actual implementation of intelligent techniques will be expected in the future, has been started.
The rest of this book is focused on the intelligent automatic generation control issue, and several intelligent control strategies are developed for simultaneous minimization of system frequency deviation and tie-line power changes to match total generation and load demand, which is required for the successful operation of interconnected power systems.
References
1. T. Hiyama. 2005. State-of-the art intelligent techniques in Japanese utilities. In Proceedings of International Conference on Intelligent Systems Application to Power Systems—ISAP, Arlington, VA, 425–28.
2. Z. Vale, G. K. Venayagamoorthy, J. Ferreira, H. Morais. 2010. Computational intelligence applications for future power systems. In Computational Intelligence for Engineering Systems, 180–97. New York: Springer.
3. M. Fozdar, C. M. Arora, V. R. Gottipati. 2007. Recent trends in intelligent techniques to power systems. In Proceedings of 42nd International Universities Power Engineering Conference (UPEC), Brighton, UK, 580–91.
4. H. Bevrani. 2009. Power system control: An overview. In Robust power system frequency control, 1–13. New York: Springer.
5. T. Hiyama, S. Oniki, H. Nagashima. 1996. Evaluation of advanced fuzzy logic PSS on analog network simulator and actual installation on hydro generators. IEEE Trans. Energy Conversion 11(1):125–131.