No description available.

This thesis reports findings from a number of modern machine learning techniques applied to electricity market price forecasting. The techniques evaluated were Support Vector Machines, Boosting, Bayesian networks, neural networks and a weekly average method. All techniques were evaluated on seven day into the future forecasting of the Regional Reference (pool) Prices (RRP) for the New South Wales (NSW) region of the Australian National Electricity Market (NEM). Due to highly volatile and non-repetitive nature of the NSW RRP, all complex machine learning methods provided inferior accuracy forecasts compared to a weekly average method. The weekly average method was computationally less expensive and more transparent to the user than any of the machine learning techniques. The Support Vector Machine (SVM) was chosen for its novel application to electricity price forecasting because it is considered to be the next generation to neural networks. The structured SVM training algorithm proved more consistent and reliable than the neural network algorithm. Bayesian networks offer the adaptability of a neural network with the advantage of providing a price forecast with confidence intervals for each half-hour determined from the actual data. The SVM and Bayesian techniques were found to provide acceptable forecasts for NSW demand. An investigation of international electricity markets found that each market was unique with different market structures, regulations, network topologies and ownership regimes. Price forecasting techniques and results cannot be universally applied without careful consideration of local conditions. For instance, price data for the Spanish and Californian electricity markets were investigated and found to have significantly lower price volatility than the NSW region of the NEM. An extensive examination of the NSW RRP showed that the price exhibited no consistent long-term trend. A stationary data set could not be extracted from the price data. Thus, making forecasting unsuited to techniques using large historical data sets. The strongest pattern found for NSW prices was the weekly cycle, so a weekly average method was developed to utilise this weekly cycle. Over 25 weeks of NSW RRP from February to July 2002, the seven day into the future price forecast mean absolute error (MAE) for the SVM technique was 27.8%. The weekly average method was more accurate with an MAE of 20.6% and with a simple linear price adjustment for demand, the error was reduced to 18.1%. The price spikes and uneven distribution of prices were unsuitable for the Boosting or Bayesian network techniques.

No description available.

No description available.

No description available.

No description available.

This thesis reports findings from a number of modern machine learning techniques applied to electricity market price forecasting. The techniques evaluated were Support Vector Machines, Boosting, Bayesian networks, neural networks and a weekly average method. All techniques were evaluated on seven day into the future forecasting of the Regional Reference (pool) Prices (RRP) for the New South Wales (NSW) region of the Australian National Electricity Market (NEM). Due to highly volatile and non-repetitive nature of the NSW RRP, all complex machine learning methods provided inferior accuracy forecasts compared to a weekly average method. The weekly average method was computationally less expensive and more transparent to the user than any of the machine learning techniques. The Support Vector Machine (SVM) was chosen for its novel application to electricity price forecasting because it is considered to be the next generation to neural networks. The structured SVM training algorithm proved more consistent and reliable than the neural network algorithm. Bayesian networks offer the adaptability of a neural network with the advantage of providing a price forecast with confidence intervals for each half-hour determined from the actual data. The SVM and Bayesian techniques were found to provide acceptable forecasts for NSW demand. An investigation of international electricity markets found that each market was unique with different market structures, regulations, network topologies and ownership regimes. Price forecasting techniques and results cannot be universally applied without careful consideration of local conditions. For instance, price data for the Spanish and Californian electricity markets were investigated and found to have significantly lower price volatility than the NSW region of the NEM. An extensive examination of the NSW RRP showed that the price exhibited no consistent long-term trend. A stationary data set could not be extracted from the price data. Thus, making forecasting unsuited to techniques using large historical data sets. The strongest pattern found for NSW prices was the weekly cycle, so a weekly average method was developed to utilise this weekly cycle. Over 25 weeks of NSW RRP from February to July 2002, the seven day into the future price forecast mean absolute error (MAE) for the SVM technique was 27.8%. The weekly average method was more accurate with an MAE of 20.6% and with a simple linear price adjustment for demand, the error was reduced to 18.1%. The price spikes and uneven distribution of prices were unsuitable for the Boosting or Bayesian network techniques.

The deregulation of the electricity industry has introduced many opportunities as well as challenges to the once monopolised industry. This recent reform towards a competitive electricity industry advocates a need for charging energy losses to market participants through a more satisfactory and transparent mechanism. Market participants, whether they are generators or consumers, would want a loss allocation scheme that is able to reflect each market participants' contribution of generation or usage in the network. However, as electricity is an indistinguishable entity, there is no accurate method to trace the flow of electricity thus far. Hence, the issue of power loss allocation within the deregulated market still remains an unresolved setback to progress to a fully competitive electricity market. Many loss allocation methods have been introduced, however, none have been universally accepted. This thesis investigates existing power flow tracing and loss allocation methods in order to critically analyse the advantages and disadvantages of each method. They include loss allocation methods currently employed in Australia’s National Electricity Market (NEM) and Great Britain Market, as well as a selection of better known loss allocation methods that are introduced in the academic research field. Understanding of these methods makes it easier to choose a method that is more suitable for each electricity market. Many researchers believe that a resolution is through a fair and equitable allocation of losses. However, the definition of “fair and equitable” varies from one literature to another. In general, a fair and equitable loss allocation method should meet electrical laws as well as economical laws. This is because market driven transactions have become the new independent decision variables that define the behaviour of electric power systems. This definition is then used as the basis to assess the results obtained from the implementation of each existing method analysed. It was found that a key limitation of existing methods is the lack of a method that is able to trace the usage allocation of each generator to each load in an electrically justifiable manner. Any improvement to existing loss allocation methods should address this limitation. Thus, the main objective of this thesis is to present two transaction based methods that have been developed and tested by the author of this thesis. Fundamentally, both methods hold the capability to analyse losses involved in the transfer of power from one point of the network to another point. The first investigated method is based on the network reduction method, where a system is reduced to the nodes of interest. The second method is based on the loop frame of reference. Instead of representing the network flows through the commonly accepted nodal frame of reference, power flows within the network are instead expressed as the sum of power flows around loops that links loads to active sources. This provides the loop-based method with an advantage in which it allows the power requirements of a load to be viewed as emanating from an active source and also the advantage of assessing the viability of contract agreements within a hybrid market model. The final objective of this thesis is to analytically compare selected existing loss allocation schemes with the proposed loop-based method. As there are no standard means of judging the accuracy of any loss allocation methods, the author of this thesis proposed a different way to distinguish different loss allocation methods. That is, through the type of competition that each method promotes. A wide range of results is obtained in which the loss allocations of some methods are dependent only on the real power injection at each bus. On the other hand, the loss allocations of other methods such as the loop-based method are dependent on network operation efficiency. The comprehension of the different type of competitions each method promotes aims to assist market regulators in recognising the feasibility of employing each loss allocation method.

This thesis reports findings from a number of modern machine learning techniques applied to electricity market price forecasting. The techniques evaluated were Support Vector Machines, Boosting, Bayesian networks, neural networks and a weekly average method. All techniques were evaluated on seven day into the future forecasting of the Regional Reference (pool) Prices (RRP) for the New South Wales (NSW) region of the Australian National Electricity Market (NEM). Due to highly volatile and non-repetitive nature of the NSW RRP, all complex machine learning methods provided inferior accuracy forecasts compared to a weekly average method. The weekly average method was computationally less expensive and more transparent to the user than any of the machine learning techniques. The Support Vector Machine (SVM) was chosen for its novel application to electricity price forecasting because it is considered to be the next generation to neural networks. The structured SVM training algorithm proved more consistent and reliable than the neural network algorithm. Bayesian networks offer the adaptability of a neural network with the advantage of providing a price forecast with confidence intervals for each half-hour determined from the actual data. The SVM and Bayesian techniques were found to provide acceptable forecasts for NSW demand. An investigation of international electricity markets found that each market was unique with different market structures, regulations, network topologies and ownership regimes. Price forecasting techniques and results cannot be universally applied without careful consideration of local conditions. For instance, price data for the Spanish and Californian electricity markets were investigated and found to have significantly lower price volatility than the NSW region of the NEM. An extensive examination of the NSW RRP showed that the price exhibited no consistent long-term trend. A stationary data set could not be extracted from the price data. Thus, making forecasting unsuited to techniques using large historical data sets. The strongest pattern found for NSW prices was the weekly cycle, so a weekly average method was developed to utilise this weekly cycle. Over 25 weeks of NSW RRP from February to July 2002, the seven day into the future price forecast mean absolute error (MAE) for the SVM technique was 27.8%. The weekly average method was more accurate with an MAE of 20.6% and with a simple linear price adjustment for demand, the error was reduced to 18.1%. The price spikes and uneven distribution of prices were unsuitable for the Boosting or Bayesian network techniques.

The deregulation of the electricity industry has introduced many opportunities as well as challenges to the once monopolised industry. This recent reform towards a competitive electricity industry advocates a need for charging energy losses to market participants through a more satisfactory and transparent mechanism. Market participants, whether they are generators or consumers, would want a loss allocation scheme that is able to reflect each market participants' contribution of generation or usage in the network. However, as electricity is an indistinguishable entity, there is no accurate method to trace the flow of electricity thus far. Hence, the issue of power loss allocation within the deregulated market still remains an unresolved setback to progress to a fully competitive electricity market. Many loss allocation methods have been introduced, however, none have been universally accepted. This thesis investigates existing power flow tracing and loss allocation methods in order to critically analyse the advantages and disadvantages of each method. They include loss allocation methods currently employed in Australia’s National Electricity Market (NEM) and Great Britain Market, as well as a selection of better known loss allocation methods that are introduced in the academic research field. Understanding of these methods makes it easier to choose a method that is more suitable for each electricity market. Many researchers believe that a resolution is through a fair and equitable allocation of losses. However, the definition of “fair and equitable” varies from one literature to another. In general, a fair and equitable loss allocation method should meet electrical laws as well as economical laws. This is because market driven transactions have become the new independent decision variables that define the behaviour of electric power systems. This definition is then used as the basis to assess the results obtained from the implementation of each existing method analysed. It was found that a key limitation of existing methods is the lack of a method that is able to trace the usage allocation of each generator to each load in an electrically justifiable manner. Any improvement to existing loss allocation methods should address this limitation. Thus, the main objective of this thesis is to present two transaction based methods that have been developed and tested by the author of this thesis. Fundamentally, both methods hold the capability to analyse losses involved in the transfer of power from one point of the network to another point. The first investigated method is based on the network reduction method, where a system is reduced to the nodes of interest. The second method is based on the loop frame of reference. Instead of representing the network flows through the commonly accepted nodal frame of reference, power flows within the network are instead expressed as the sum of power flows around loops that links loads to active sources. This provides the loop-based method with an advantage in which it allows the power requirements of a load to be viewed as emanating from an active source and also the advantage of assessing the viability of contract agreements within a hybrid market model. The final objective of this thesis is to analytically compare selected existing loss allocation schemes with the proposed loop-based method. As there are no standard means of judging the accuracy of any loss allocation methods, the author of this thesis proposed a different way to distinguish different loss allocation methods. That is, through the type of competition that each method promotes. A wide range of results is obtained in which the loss allocations of some methods are dependent only on the real power injection at each bus. On the other hand, the loss allocations of other methods such as the loop-based method are dependent on network operation efficiency. The comprehension of the different type of competitions each method promotes aims to assist market regulators in recognising the feasibility of employing each loss allocation method.