An electric power system is an aggregation of electrical systems used to generate, transmit and consume electric power. Power systems engineering is a subset of electrical engineering that deals with the engineering of generation, transmission, distribution, utilization of electric power and related equipments.
Power system is on continuous rise at fast pace for last few decades. As the size of the power system consisting of generators, transmission lines, distribution lines and related equipments increases, the possibility of occurring a fault in the power system increases. In previous years, due to several factors like threat of climate change, energy security, penetration of renewable power resources, increase in peak and base load have been significant cause of concern worldwide to achieve stable and adequate power for convenient living.
The collection of large quantity of input data and system parameters, using this data and parameters to control the complete power system is the basic mechanism for controlling present day power system. Manual calculations, traditional technical evaluation and creating reports are required to successfully operate the power system. As the power system grew, this manual analysis became more tedious to control steady state and transient state of power system. In view of this, application of modern system techniques like Artificial Intelligence (AI) may be of great help in for reliable and optimum performance of power system operation. Though, most of these applications require information about large number of system parameters, which can be provided by latest telecommunications and computing technology, they require sophisticated technology capable to extract this information from the large input data set. It may be noted that, many times such large input data sets are not error free and often contain various types of errors. AI plays a major role in power systems where they solve different problems in power systems such as scheduling, calculating, statistics, forecast, maintenance, protection and real time control.
Artificial Intelligence (AI) has been, in last two and more decades, developed as a science even though it may still be considered in its early stages of development. Depending on the aims and mythologies applied in research, its definition varies.
Before the development of AI, word “intelligence” was associated with the human brain. But as the computer technology advanced, scientists found a way of training computers by the methodology similar to our brain. Thus Artificial Intelligence was invented, which can essentially be defined as intelligence originating from machines. To put it in a simple way, it is providing machines with the ability to “think”, “learn”, and “adapt”.
As a broad description, it may be described as the science of making machines to do things that would require intelligence if done by humans. AI applications are now being implemented in a very wide variety of fields, ranging from social media, humanities, natural and applied sciences and technology. In the context of power systems, application wings of AI like robotics, artificial neural networks (ANNs), fuzzy logic and expert system techniques are used.
Commonly, artificial intelligence is known to be the intelligence performed by machines and software, for example, robots and computer programs. The term is generally used to the project of developing systems possessing the intellectual capabilities, features and characteristics of humans, like the ability to think, reasoning, find the meaning, generalize, distinguish, learn from past experience or rectify their mistakes.
Need for AI in Power System
Power system analysis by conventional techniques becomes more difficult because of:
- Complicated, versatile and vast amount of information used in calculation, diagnosis, analysis of present and past data, control and operation of power systems.
- Increase processing time due to large amount of data.
Peak loads, energy demands and size of power system are continuously increasing. In view of this, the existing power system is overburdened due to peak demand and energy requirements. It is not always capable to cater to the high peak demands and energy needs. Hence, the power system struggles between generation and demand. This situation requires a liberal power system operation and control which is possible only by analyzing the system conditions in more exhaustive manner in present situations than earlier. Sophisticated computer tools are now the available in solving the complex problems like load flow, frequency control, reactive power control, load forecast, equipment maintenance, system protection, real time control and system design, which are part of power system planning, operation, and diagnosis. Among these computer tools, Artificial Intelligence has established itself as most sophisticated software in recent years and has been applied to various areas of power system.
Artificial Intelligence Techniques
There are mainly three techniques used by AI
- Artificial neural networks
- Fuzzy logic systems.
- Expert system technique
(A) ARTIFICIAL NEURAL NETWORKS (ANN)
Artificial Neuron fig 1ps5
Artificial Neural Networks consist of artificial neurons which are artificial model of human neuron as shown in fig 1.
Artificial Neural Networks are biologically inspired systems which convert a set of inputs like X1 to X5 into a set of outputs like ‘Y’ by a network of neurons, where each neuron produces one output as a function of inputs. A neural network is modeled like human brain and can consist of millions of simple processing nodes, called perceptron which are thoroughly interconnected. A single artificial neuron is building block of ANN. A fundamental neuron can be considered as a processor which makes a simple non-linear operation of its inputs producing a single output. The understanding of the working of neurons and the pattern of their interconnection can be used to construct computers for solving real world problems of classification of patterns and pattern recognition.
Neural Networks find extensive applications in areas where traditional computers don’t perform satisfactory. Like, for problem statements where instead of programmed outputs, you’d like the system to learn, adapt, and change the results in accordance with the data you’re feeding to it. Neural networks also find applications whenever we are dealing with incomplete data. And honestly, most of the input data to be presented to an ANN is mostly incomplete.
ANN are classified by their architecture, number of layers and connectivity pattern. A typical structure of an ANN is shown in fig 2.
This is described as below.
Input Layer: The nodes are input units which do not process the data and information but distribute this data and information to next layer i.e. hidden layer.
Hidden Layers: The nodes in hidden layers are not directly evident and visible. They work as processing unit of ANN and converts input from input nodes and provide output to output nodes as per the algorithm. They provide the networks the ability to map or classify the nonlinear problems. Hidden layers may be single layer or more than one layer.
In case of any hardware like OR gate, the outputs are defined by inputs in traditional computer. But in case of human brain or ANN, output shall not follow the defined output but decide according to its thought process and reasoning.
Output Layer: The controlled outputs from hidden layers are transferred to output nodes existing in output layer.
(A1) Application of ANN in Power Systems
The one of the important function in a power system is to maintain the power quality in terms of permissible voltage and frequency limits. This task has to be performed with highest level of accuracy and consistency. Fig – 3 shows a simplified diagram of the main data flow in a power system where real-time measurements are stored in a database.
The inputs to the database consist of generated active and reactive power, sending end voltage and current, angle between sending end voltage and current, overhead line parameters, load side active and reactive power, receiving end voltage and currents, angle between receiving end voltage and currents and surge impedance loading. These parameters are equivalent to inputs to the node of input layer of ANN. Estate estimation and real time power system model are equivalent to hidden layers of ANN. Output to the output layer of ANN is in the form of security assessment of power system. This generates automated control actions to be applied to the power system for operation of power system within permissible limits of frequency and voltage. As long as frequency is within permissible limits, the normal operation of power system takes place. If these limits are crossed and power system goes into unsafe zone, control actions come into action and other conditions like utility outage, islanding and house load operation may occur. In this way, based on the estimated values the existing mathematical model of the power system is established. In general, there are two types of security assessments: static security assessment and dynamic security assessment. In both types different operational states are defined as follows.
• Normal or secure state: In this state, all customer demands are met and operating parameters like voltage and frequency are within permissible limits.
• Alarming or near critical state: In this state the system voltage and frequency are still within limits but further disturbances can lead off normal operation resulting in unstable operation of power system.
• Off normal or unstable operation: In this case, the power system enters the unstable mode of operation upon crossing permissible limits of security parameters. If severity of this off normal state is high, cascading effect may occur in power system, resulting in segregation of power system in several parts with generation in some parts and loads in remaining parts.
The forecasting of load demand in a power system has become one of the most important issue for research and development, as it plays a significant role in financial planning and capacity addition planning. Load forecasting is however a difficult job. First, because the load forecast is tedious and depend upon several factors. The load at a given hour is dependent not only on the load at the previous hour, but also on the load at the same hour on the previous day. In addition to this, there are other significant variables that must be considered for accurate load forecasting like whether-related variables. Generally, we can maintain the load forecast into three groups: Short-term load forecasting over an interval ranging from a few minutes to a week. Mid-term load forecasting generally is considered from one month to five years, and is mostly used to buy adequate fuel for power plants. Long-term load forecasting covers the range from 5 to 20 years or more, used by Government agencies to outline future capacity addition planning. ANN is most suited for load forecasting because of the availability of historical load data available with the generator databases. The majority of the projects using ANNs have considered lots of factors such as historical data, off load periods, holidays and weather condition. Weather condition is considered because in winter or rainy season load demand increases and in summer season load demand increases. Application of ANNs is useful because, it can handle large number of inputs at a time. Other positive points of ANNs designed for load forecasting are as: Capability of ANN to handle input parameters that do not have direct relationship between them such as weather conditions and load profile and Process of ANN model can be conducted off-line without time constrains.
Calculation of line parameters
ANNs operate on artificial simulation of human biological models and perform biological simulation of real world problems. Hence, the input parameters of power system like generation, transmission and distribution systems can be fed to the ANNs so that a suitable output solution can be obtained. In spite of limitations and incompleteness of parameters of a practical transmission and distribution system, the exact values of line parameters can be determined with the use of ANN. For example, the value of inductance, capacitance and resistance in a transmission line can be numerically calculated by ANNs taking in various factors like environmental factors, unbalancing conditions, line loading, frequency and voltage variations etc.
(A2) Advantages of artificial neural networks
∙ Speed of processing.
∙ They do not need any appropriate knowledge of the system model.
∙ They have the ability to handle situations of incomplete data and corrupt data.
∙ They are fault tolerant.
∙ Artificial neural networks are fast and robust.
(A3) Disadvantages of artificial neural networks
∙ Large dimensionality.
∙ Results are always generated even if the input data are unreasonable.
∙ They are not scalable i.e. once an artificial neural network is trained to do certain task, it is difficult to extend for other tasks without retraining the neural network.
(B) FUZZY LOGIC
Fuzzy logic or Fuzzy systems are logical systems for standardization and formalization of approximate reasoning. It is similar to human decision making with an ability to produce exact and accurate solutions from certain or even approximate information and data. The reasoning in fuzzy logic is similar to human reasoning. Fuzzy logic is the way like which human brain works, and we can use this technology in machines so that they can perform somewhat like humans.
Diagram of fuzzy logic controller is shown in fig 4.
Fuzzification provides higher quality of expressive power, higher generality and a better capability to solve complicated problems at economical cost. Fuzzy logic allows a particular level of ambiguity throughout an analysis. Because this ambiguity can specify available information and minimize problem complexity, fuzzy logic is useful in many applications. For power systems, fuzzy logic is suitable for applications in many areas where the available information involves uncertainty and ambiguity. For example, a problem might involve logical reasoning, but can be applied to numerical, other than symbolic inputs and outputs. Fuzzy logic provide the conversions from numerical to symbolic inputs, and back again for the outputs.
(B1) Application in Power Systems
- Fault diagnosis
Fuzzy expert system for fault diagnosis in a transmission line is given in fig 5.
Database and knowledge base work as the input layer, fault network identification, hypotheses & calculation and inference engine work as hidden layers and fault determination work as output layer. If any fault occurs in the transmission line, the fault detector detects the fault and feeds it to the fuzzy system. Only three line fault currents and the angular difference between fault and pre-fault currents are sufficient to implement this technique and are used as inputs to the fuzzy system. The fuzzy system is used to obtain the output in the form of the fault type. Hence Fuzzy systems can be successfully used for fault diagnosis.
- Fuzzy logic can be used for designing the physical components of power systems. They can be used in anything from small circuits to large systems. They can be used to increase the efficiency of the components used in power systems. As most of the data used in power system analysis are approximate values and assumptions, fuzzy logic can be of great use to derive a reasonable, exact and ambiguity-free output.
- Reactive power control
For reactive power control with the objective of enhancing the voltage profile of power system, fuzzy logic has been most suitable. The voltage deviation and controlling variables are converted into fuzzy set or fuzzy system notations to construct the relations between voltage deviation and controlling ability of the controlling devices. The main control variables are generator excitation, transformer taps and VAR compensators. A fuzzy system provides information about correct values of generation excitation, correct setting of transformer taps and calculate correct values and location of VAR compensators.
- A soft computing fuzzy technique is employed to maximize the efficiency from solar panel to give maximum power output.
- The performance of various energy storage systems’ life time can be improved by utilizing fuzzy logic controllers in a hybrid power system using renewable sources.
(C) EXPERT SYSTEMS
Structure of an Expert system is shown in fig 6.
An expert system obtains the knowledge of a human expert in a narrow specified domain into a machine implementable form. Expert systems are computer programs which have proficiency and competence in a particular field. This knowledge is generally stored separately from the program’s procedural part and may be stored in one of the many forms, like rules, decision trees, models, and frames. They are also called as knowledge based systems or rule based systems. Expert systems use the interface mechanism and knowledge to solve problems which cannot be or difficult to be solved by human skill and intellect.
- It is permanent and consistent.
- It can be easily documented.
- It can be easily transferred or reproduced.
Expert Systems are unable to learn or adapt to new problems or situations.
(C3) Applications of Expert Systems
1. Expert systems can be used to improve the performance of a transmission line. The environmental sensors sense the environmental and atmospheric conditions and give them as input to the expert systems. The expert systems are computer programs written by knowledge engineers which provide the value of line parameters to be used as the input. The ANNs are trained to change the values of line parameters over the given ranges based on the environmental conditions. Training algorithm has to be given to ANN. After training is over, neural network is tested and the performance of updated trained neural network is evaluated. If performance is not up to the desired level, some variations can be done like varying number of hidden layers, varying number of neurons in each layer. The processing speed is directly proportional to the number of neurons. These networks take different neurons for different layers and different activation functions between input and hidden layer and hidden and output layer to obtain the desired output. In this way the performance of the transmission line can be improved.
2. Many areas of applications in power systems match the abilities of expert systems like decision making, archiving knowledge, and solving problems by reasoning, heuristics and judgment. Expert systems are especially useful for these problems when a large amount of data and information must be processed in a short period of time.
3. Since expert systems are basically computer programs, the process of writing codes for these programs is simpler than actually calculating and estimating the value of parameters used in generation, transmission and distribution. Any modifications even after design can be easily done because they are computer programs. Virtually, estimation of these values can be done and further research for increasing the efficiency of the process can be also be performed.
Exhaustive List of Application and Features of AI in Power Systems
Several problems in power systems cannot be solved by conventional techniques, as they are based on several requirements which may not feasible all the time. In these situations, artificial intelligence techniques are the obvious and the only option. Areas of application and features of AI in power systems are:
∙ Replacing human workers for dangerous and highly specialized operations, such as live maintenance of high voltage transmission lines using robotic technology.
∙ Operation in hazardous environments, such as radioactive locations in nuclear plants, access to narrow spaces, such as cable ducts and cooling pipes and precise positioning of measuring equipments is carried out by using of Robotic technology. In case of Nuclear Reactors, precise fuelling operation is done by using three dimensional robotic technology based on AI.
∙ Expert systems use the interface mechanism and knowledge to solve problems which cannot be, or difficult to be, solved by human skill and intellect.
∙ Results are permanent and consistent and can be easily documented. Results can be easily transferred and reproduced.
∙ The understanding of the working of neurons and the pattern of their interconnection can be used to construct computers for solving real world problems of classification of patterns and pattern recognition.
∙ Operation of power system like unit commitment, hydro-thermal coordination, economic dispatch, congestion management, maintenance scheduling, state estimation, load and power flow.
∙ Planning of power system like generation expansion planning, power system reliability, transmission expansion planning, reactive power planning.
∙ Control of power system like voltage control, stability control, power flow control, load frequency control and reduction in transmission line losses.
∙ Automation of power system like restoration, management, fault diagnosis, network security.
∙ Applications of distribution system like planning and operation of distribution system, demand side response and demand side management, operation and control of smart grids, network reconfiguration.
∙ Applications of distributed generation like distributed generation planning, solar photovoltaic power plant control, wind turbine plant control and other renewable energy resources control. Control of power plants like fuel cells power plant control, thermal power plant control.
The main feature of power system design and planning is reliability and excellent performance, which was earlier evaluated using conventional methods. Moreover, conventional techniques do not fulfill the probabilistic essence of power systems. This leads to increase in operating and maintenance costs. Plenty of research has been performed to develop AI for power system applications. A continuous research is yet to be carried out to achieve maximum advantages of this developing technology for improving the efficiency of power system, optimum use of capital investment, sophisticated control and monitoring, optimum performance of renewable energy resources etc.