abstract: The students of Arizona State University, under the mentorship of Dr George Karady, have been collaborating with Salt River Project (SRP), a major power utility in the state of Arizona, trying to study and optimize a battery-supported grid-tied rooftop Photovoltaic (PV) system, sold by a commercial vendor. SRP believes this system has the potential to satisfy the needs of its customers, who opt for utilizing solar power to partially satisfy their power needs. An important part of this elaborate project is the development of a new load forecasting algorithm and a better control strategy for the optimized utilization of the storage system. The built-in algorithm of this commercial unit uses simple forecasting and battery control strategies. With the recent improvement in Machine Learning (ML) techniques, development of a more sophisticated model of the problem in hand was possible. This research is aimed at achieving the goal by utilizing the appropriate ML techniques to better model the problem, which will essentially result in a better solution. In this research, a set of six unique features are used to model the load forecasting problem and different ML algorithms are simulated on the developed model. A similar approach is taken to solve the PV prediction problem. Finally, a very effective battery control strategy is built (utilizing the results of the load and PV forecasting), with the aim of ensuring a reduction in the amount of energy consumed from the grid during the “on-peak” hours. Apart from the reduction in the energy consumption, this battery control algorithm decelerates the “cycling aging” or the aging of the battery owing to the charge/dis-charges cycles endured by selectively charging/dis-charging the battery based on need. ii The results of this proposed strategy are verified using a hardware implementation (the PV system was coupled with a custom-built load bank and this setup was used to simulate a house). The results pertaining to the performances of the built-in algorithm and the ML algorithm are compared and the economic analysis is performed. The findings of this research have in the process of being published in a reputed journal. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2018

A spinning reserve assessment technique for a deregulated system has been developed and presented in this thesis. The technique is based on direct search optimization approach. Computer programs have been developed to implement the optimization processes both for transmission loss and without transmission loss.<p> A system commits adequate generation to satisfy its load and export/import commitment. Additional generation known as spinning reserve is also required to satisfy unforeseen load changes or withstand sudden generation loss. In a vertically integrated system, a single entity generates, transmits and distributes electrical energy. As a part of its operational planning, the single entity decides the level of spinning reserve. The cost associated with generation, transmission, distribution including the spinning reserve is then passed on to the customers.<p> In a deregulated system, generation, transmission and distribution are three businesses. Generators compete with each other to sell their energy to the Independent System Operators (ISO). ISO coordinates the bids from the generation as well as the bids from the bulk customers. In order to ensure a reliable operation, ISO must also ensure that the system has adequate spinning reserve. ISO must buy spinning reserve from the spinning reserve market. A probabilistic method called the load forecast uncertainty (LFU)-based spinning reserve assessment (LSRA) is proposed to assess the spinning reserve requirements in a deregulated power system.<p> The LSRA is an energy cost- based approach that incorporates the load forecast uncertainty of the day-ahead market (DAM) and the energy prices within the system in the assessment process. The LSRA technique analyzes every load step of the 49-step LFU model and the probability that the hourly DAM load will be within that load step on the actual day. Economic and reliability decisions are made based on the analysis to determine and minimize the total energy cost for each hour subject to certain system constraints in order to assess the spinning reserve requirements. The direct search optimization approach is easily implemented in the determination of the optimal SR requirements since the objective function is a combination of linear and non-linear functions. This approach involves varying the amount of SR within the system from zero to the maximum available capacity. By varying the amount of SR within the system, the optimal SR for which the hourly total operating cost is minimum and all operating constraints are satisfied is evaluated.<p> One major advantage of the LSRA technique is the inclusion of all the major system variables like DAM hourly loads and energy prices and the utilization of the stochastic nature of the system components in its computation. The setback in this technique is the need to have access to historical load data and spot market energy prices during all seasons. The availability and reliability of these historical data has a huge effect on the LSRA technique to adequately assess the spinning reserve requirements in a deregulated system.<p> The technique, along with the effects of load forecast uncertainty, energy prices of spinning reserve and spot market and the reloading up and down limits of the generating zones on the spinning reserve requirements are illustrated in detail in this thesis work. The effects of the above stochastic components of the power system on the spinning reserve requirements are illustrated numerically by different graphs using a computer simulation of the technique incorporating test systems with and without transmission loss.

A spinning reserve assessment technique for a deregulated system has been developed and presented in this thesis. The technique is based on direct search optimization approach. Computer programs have been developed to implement the optimization processes both for transmission loss and without transmission loss.<p> A system commits adequate generation to satisfy its load and export/import commitment. Additional generation known as spinning reserve is also required to satisfy unforeseen load changes or withstand sudden generation loss. In a vertically integrated system, a single entity generates, transmits and distributes electrical energy. As a part of its operational planning, the single entity decides the level of spinning reserve. The cost associated with generation, transmission, distribution including the spinning reserve is then passed on to the customers.<p> In a deregulated system, generation, transmission and distribution are three businesses. Generators compete with each other to sell their energy to the Independent System Operators (ISO). ISO coordinates the bids from the generation as well as the bids from the bulk customers. In order to ensure a reliable operation, ISO must also ensure that the system has adequate spinning reserve. ISO must buy spinning reserve from the spinning reserve market. A probabilistic method called the load forecast uncertainty (LFU)-based spinning reserve assessment (LSRA) is proposed to assess the spinning reserve requirements in a deregulated power system.<p> The LSRA is an energy cost- based approach that incorporates the load forecast uncertainty of the day-ahead market (DAM) and the energy prices within the system in the assessment process. The LSRA technique analyzes every load step of the 49-step LFU model and the probability that the hourly DAM load will be within that load step on the actual day. Economic and reliability decisions are made based on the analysis to determine and minimize the total energy cost for each hour subject to certain system constraints in order to assess the spinning reserve requirements. The direct search optimization approach is easily implemented in the determination of the optimal SR requirements since the objective function is a combination of linear and non-linear functions. This approach involves varying the amount of SR within the system from zero to the maximum available capacity. By varying the amount of SR within the system, the optimal SR for which the hourly total operating cost is minimum and all operating constraints are satisfied is evaluated.<p> One major advantage of the LSRA technique is the inclusion of all the major system variables like DAM hourly loads and energy prices and the utilization of the stochastic nature of the system components in its computation. The setback in this technique is the need to have access to historical load data and spot market energy prices during all seasons. The availability and reliability of these historical data has a huge effect on the LSRA technique to adequately assess the spinning reserve requirements in a deregulated system.<p> The technique, along with the effects of load forecast uncertainty, energy prices of spinning reserve and spot market and the reloading up and down limits of the generating zones on the spinning reserve requirements are illustrated in detail in this thesis work. The effects of the above stochastic components of the power system on the spinning reserve requirements are illustrated numerically by different graphs using a computer simulation of the technique incorporating test systems with and without transmission loss.

The accuracy of load forecast has a significant impact for power companies on executing the plan of power development, reducing operating costs and providing reliable power to the client. Short-term load forecasting is to forecast load demand for the duration of one hour or less. This study presents a new approach to process load forecasting. A Support Vector Machine (SVM) was used for the initial load estimation. Particle Swarm Optimization (PSO) was then adopted to search for optimal parameters for the SVM. In doing the load forecast, training data is the most important factor to affect the calculation time. Using more data for model training should provide a better forecast results, but it needs more computing time and is less efficient. Applications of data mining can provide means to reduce the data requirement and the computing time. The proposed Modified Support Vector Machines approach can be proved to provide a more accurate load forecasting.

With the economic development, the high quality has become a critical issue for service continuous of power companies. To ensure the stable power supply, the asset management of power equipments is applied to prevent the system outage. With voluminous distribution transformers over very wide area, the real time monitoring of temperature has been included in the scope of smart grid. During recent years, the service outage due to transformer overloading has caused customer panic as well as deterioration of service quality. This thesis develops the Transformer Terminal Unit (TTU) by integration of computer chip for power consumption, DSP and sampling circuit of temperature measurement to achieve the functions of real time monitoring of transformer operation condition. When an abnormal operation condition such as overloading or high oil temperature occurs, the TTU can report the contingency back to the control station via the hybrid communication system so that the distribution system operators can take remedy action to prevent the contingency. The actual loading and temperature of transforms are also measured and collected in this study to develop the relationship of temperature and loading levels. By collecting transformer temperature, the power demand of a transformer can be estimated and the load shedding can then be activated to prevent the problem of overloading when the temperature exceeds the operation constraint.

The thesis investigates the optimal expansion planning of substations for the distribution system of Taipei City District of Taiwan Power Company. The small area load forecasting is executed with the support of Outage Management System(OMS) database. The capacity expansion of distribution substations is obtained by considering the annual load growth of each service area to achieve the cost effectiveness of substation investment. The geographic information of each service zone has been retrieved form the OMS data. With the land use planning of Taipei City Government, the load density of each small area for the target year is derived according to the final floor area and development strength of the land base. The load forecasting of each small area is then solved by considering the load growth of each customer class, which is then used for the expansion planning of substations. After determining the small area load forecasting for the final target year, the center of gravity method is applied to find the geographic blocks of all substations and the corresponding service areas at the target year. The power loading of each small area is used to calculate the power loading loss of which service area to solve the optimal location within the block for each substation. Based on the annual load forecasting of all small areas, the expansion planning of distribution substations for Taipei City District is derived by Dynamic Programming(DP) and Immune Algorithm(IA) to achieve minimization of power loading loss with subject to the operation constraint. By the proposed methodology, the unit commitment of distribution substations is determined to meet the load growth of service area and achieve power loading loss minimization of distribution systems.

Despite the extensive deployment of smart-meters (SMs) at the low-voltage (LV) level, which are either fully operational or will be in the near future, distribution network operators (DNOs) are still relying on a limited number of permanently installed monitoring devices at primary and secondary medium-voltage (MV) substations, for purposes of network operation and control, as well as to inform and facilitate trading interactions between generators, distributors and suppliers. Accordingly, improved and sufficiently developed models for the analysis of aggregate demands at the MV-level are required for the correct assessment of load variability, composition and time-dependent evolution, necessary for: addressing issues of robustness, security and reliability; accomplishing higher penetration levels from renewable/distributed generation; implementing demand-side-management (DSM) schemes and incorporating new technologies; decreasing environmental and economic costs and aiding towards the realisation of automated and proactive ''smart-grid'' networks. The analysis of MV-demand measurements provides an independent source of information that can capture network characteristics that do not manifest in the data collected at the LV-level, or when such data is restricted or altogether unavailable. This information describes the supply/demand interactions at the mid-level between high-voltage (HV) transmission and LV end-user consumption and opens possibilities for validation of existing bottom-up aggregation approaches, while addressing issues of reliance on survey-based data for technical and economic power system studies. This thesis presents improved and novel methodologies for the analysis of aggregate demands, measured at MV-substations, aimed at more accurate and detailed load profiling, temporal decomposition and identification of the drivers of demand variability, classification of grid-supply- points (GSPs) according to consumption patterns, disaggregation with respect to customer-classes and load-types and load forecasting. The developed models are based on a number of traditional and modern analytical and statistical techniques, including: data mining, correlational and regression analysis, Fourier analysis, clustering and pattern recognition, etc. The approaches are demonstrated on demand datasets from UK and European based DNOs, thus providing specific information for the demand characteristics, the dependencies to external parameters and to socio-behavioural factors and the most likely load composition at the corresponding geographical locations, while the approaches are also intendent to be easily adaptable for studies at equivalent voltage and demand aggregation levels.

[pt] Os métodos de análise de séries temporais têm se revelado uma importante ferramenta de apoio à tomada de decisões, com importância crescente em um mundo cada vez mais globalizado. Esse fato pode ser ilustrado, entre muitos outros, através de um convênio firmado entre o CEPEL, o Núcleo de Estatística Computacional da PUC/RJ e a Eletrobrás, para se avaliar a utilidade dessas ferramentas nas etapas do planejamento do setor elétrico brasileiro. A metodologia em Espaço de Estado proporcionou o surgimento de duas importantes classes de modelos de previsão e análise de séries temporais completamente alternativas (os modelos estruturais e os modelos de inovações em espaço de estado), e, por isso, podem por vezes, causar dúvidas quando se fala em métodos de previsão em espaço de estado sem se especificar sobre qual das duas se está falando. Foi escolhido uma técnica específica e facilmente executável em softwares comerciais para cada classe de modelos: O desenvolvimento clássico de Harvey implementado no software STAMP, representando os modelos estruturais; e o desenvolvimento de Goodrich implementado no software FMP, representando os modelos de inovações. Essas técnicas estão tratadas de uma forma aprofundada, para proporcionar um melhor entendimento teórico das diferenças existentes entre ambas. Com o intuito de se avaliar a performance frente às outras técnicas existentes, são comparados os resultados das previsões entre as metodologias a partir de um sistema de comparação baseado nas estatísticas MAPE (Mean Absolute Percentage Error), RMSE (Root Mean Squared Error) e U-Theil. Para tanto são vistos sucintamente as técnicas: Alisamento Exponencial (Holt-Winters), Box & Jenkins e Redes Neurais. Todas as técnicas foram aplicadas aos dados de consumo de energia elétrica das 32 empresas concessionárias do setor no Brasil, além de comparadas com as previsões realizadas por essas concessionárias. A novidade deste trabalho para o projeto em andamento está na aplicação multivariada possível através da metodologia de Goodrich. / [en] The analysis of time series is, nowadays one of the most important tools in the decision making process, due mainly to the globalization of the world. As an illustration of that we can mention the recent contract signed between NEC/PUC-Rio and CEPEL/Eletrobrás, where time series techniques are to be used in the planning process of the brazilian sector. The state-space approach forms the basis of two important forecasting models to time series analysis the structural model and the state space innovation model. Because of that one finds it difficult to have a clear cut definition of either one of them. These two models formulation were implemented in comercial softwares: the structural model of A. Harvey in STAMP and the state space innovation of R. Goodrich in FMP. In order to check the perfomance of these state space approaches vis-à-vis the traditional forecasting techniques, it was used the following statistics: MAPE (Mean Absolute Percentage Error), RMSE (Root Mean Squared Error) and U-Theil. The traditional approaches used in the comparison were: Holt-Winters, Box & Jenkins and Backpropagation Neural Network. All the methods, included the state space ones were applied to the demand series of 32 electrical utilities which form the brazilian electrical distribution system. If was also attempted the multivariate state-space formulation of R. Goodrich which is included in FMP software.

[pt] Esta tese propõe um modelo para o tratamento de observações diárias e é aplicado na área do setor elétrico, no problema de previsão de carga horária. O modelo proposto é basicamente um modelo estrutural onde a sazonalidade anual (movimentos periódicos dentro do ano) é modelada utilizando a técnica de Splines. Esta técnica também é utilizada na estimação do efeito não linear de uma variável explicativa. O modelo desenvolvido nesta tese também leva em conta os feriados dada a grande influência dos mesmos no consumo de energia elétrica. A metodologia proposta é aplicada à três concessionárias do Sistema Interligado Brasileiro: LIGHT (Estado do Rio de Janeiro); CEMIG (Estado de Minas Gerais) e COPEL (Estado do Paraná). A estimação é levada a cabo utilizando o software STAMP conjuntamente com módulos desenvolvidos no utilitário MATLAB. / [en] This thesis presents a model that deals with daily obsevations applied to the problem of forecasting daily elecricity demand. This approach is basaed on a structural time series model with the annual seasonal pattern being modelled by a Periodic Sppline. The methods of Splines was first used in Harvey and Koopman (1993) to analyse hourly load observations, including temperature used an explanatory variable which is also modelled by a Spline. The main contribuition of this thesis is the treatment of holidays and the temperature response modelled by a spline which considerss the possible vsariations that the effect of temperature has on electricity demand within the year. The methodology is applied to three companies of the Brazilian electrical system: LIGHT (State of Rio de Janeiro), CEMIG (State of Minas Gerais) and COPEL (state of Paraná).

Traditional forecasting approaches forecast the total system load directly without considering the individual consumer's load. With the introduction of the smart grid, lots of renewable energy resources such as wind and solar are added to the system from consumer side fluctuates the system load and makes forecasting more complex. Thus, it is necessary to forecast individual consumers load. Here, a framework is presented in which individual customer loads is forecasted rather than the system load. At first, a hierarchical cluster analysis is performed to classify daily load patterns into different groups for all the individuals. Then an association analysis is performed to determine critical influential factors that affect the load curve for given day. The next step is the application of a decision tree to establish classification rules between the different groups of the load curve and the critical influential factors. Then, appropriate forecasting models are chosen for different load patterns and the individual load is forecasted. Finally, the forecasted total system load is obtained through an aggregation of an individual load forecasting results. The relative error of forecasting the system load using this framework is compared with the relative errors using SVM regression and this framework had better accuracy. This framework is also used for forecasting the power output of the renewable generation. Also, the results of the day ahead forecast of system load and renewable generation is used for economic power scheduling for the microgrid and peak shaving for the utilities.