الفهرس 
Literature Review on Net Zero-Emission in The Deregulated Marke
Classical and Meta-heuristic Techniques for Decarbonized Unit CommitmentOnly 14 pages are availabe for public view
 |  | from | 461 |  |  |
| | | from | 461 | | |
Abstract
In electricity systems, deregulation has primarily been implemented from an economic perspective. Deregulation creates rivalry among providers, which further enhances technological performance. The original plan was to use a tariff that was constantly changing over time rather than a fixed one. from an economic perspective, this implies that less will be consumed if prices are high at a given hour and vice versa. In theory, this means that demand will need to be rescheduled and redistributed in order to move loads from peak hours when prices are highest to lower demand hours when prices are lower. Therefore, prior knowledge of load and price values is essential for security studies and short- and long-term planning strategies. The newly constructed energy power markets are required to publish their prices every five minutes and 24 hours in advance in order to support price determination, unit commitment, and demand response strategies. It is evident that understanding the relationship or values between load and pricing is necessary for the many suggested techniques. As a result, forecasting studies proved to be quite important.
Numerous elements, including past loading, temperature, weather, costs, and so on, influence load forecasting studies. Because of the deregulated market, prices have a disproportionately large impact on load. Both the load and the price time series are intrinsically influenced by other variables. Thus, the estimated prices method is also required to predict the price of the expected load value to construct a load forecasting model. Therefore, developing (load and price), and (wind, and solar) power forecasting models are convenient. Building load and price forecasting models based on load and price time series, which naturally include other parameters like temperature and weather conditions, seasonal effects, and calendar, are appropriate given the challenges in obtaining the other factors affecting these time series when comparing with forecasting load and price whether the affected parameters are considered.
Many algorithms, including machine learning, and deep learning, have been used to study load forecasting. One of the modern machine learning methods lately used in power system research are Adaptive Neural-based Fuzzy Inference Systems (ANFIS), and Artificial Neural Networks (ANN) besides the modern deep learning algorithms are Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRUs). The performance of the machine and deep learning techniques as forecasting tools are directly correlated with the input data and measured by using a set of errors such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Normalized Root Mean Squared Error (NRMSE), and Mean Absolute Percentage Error (MAPE), after selecting the most appropriate optimizer for these algorithms. Preprocessing strategies are therefore functionalized to handle input data. these strategies are such as normalizing and de-normalizing, standardizing, filtering, and smoothing the datasets.
There are four short-term forecasting models which are as follows:
1. The first model involves predicting the Electric Vehicle (EV) charging demand, hourly energy load, and the hourly price of power individually, without relying on each other or other variables like temperatures, and the days of the week (weekdays, weekends, or holidays) in four seasons.
2. In the second model, each of the hourly energy load and the hourly price of electricity are predicted dependently on one another and the other previously discussed elements.
3. The third model involves forecasting the hourly wind and solar power, depending on the effect of seasonality and the temperature.
4. The fourth model is composed of two scenarios which are:
a. In the first scenario, there Forecasting a 24-hour period for a single workday, the last day of every season, while accounting for several factors such as dewpoints, dry bulb usage, workdays, and the associated electricity consumption costs.
b. The second scenario involves projecting a 24-hour period for a single weekend day or special holiday in each season, while accounting for various factors such as dewpoints, dry bulb conditions, max, and min temperature of each hour, off days, and the associated electricity expenditures associated with the consumption within that season.
The Multi-Objective Decarbonized Unit Commitment (MO-DUC) is expressed as a multi-objective function that minimizes network production, zero-net pollution and carbon costs, and transmission line loss without taking capital spending into consideration. By merging solar and wind energy, the emissions from the grid are reduced. On the other hand, since EVs only require grid power to charge, they will help lower vehicle-related pollution. Renewable Energy Resources (RERs) will also assist in lower production costs. The MO-DUC is designed by using dispatchable generators and renewable resources in four seasons for two cases, where the first case includes the direct supply to available load without any network (zero-losses). The second case, which illustrates the effect of losses, considers the IEEE 57 bus system. Four scenarios of EVs charging load profile are applied in four seasons for two types of days in the calendar (workdays, and holidays) in two cases that are:
1. Zero-losses of the system consideration.
2. losses of the system of IEEE 57 bus electric network.
In the following four EV charging load profile scenarios which are explained as follows:
1. Scenario #1: Grid EVs’ load_immediate_full capacity for 24-hour.
2. Scenario #2: Grid EVs’ load_intermediate_balanced for 24-hour.
3. Scenario #3: Grid EVs’ load_immediate_full capacity from hour 11 pm to 8 am.
4. Scenario #4: Grid EVs’ load_intermediate_balanced from hour 11 pm to 8 am.
The deregulated market declares the schedule in response to the wholesale market’s price announcement. Contributing EV owners and residence customers may choose the Demand Response (DR) program based on the elasticity of the hourly load, which allows them to shift their loads in times such as during peak hours, when electricity prices are high, or during rush hour.
In order to minimize overall production costs and emissions and to maximize profit and income, the committed producing units’ hourly distribution from the fossil-fueled generators and the renewable energy resources (wind, and solar) and contribution of demand and the four scenarios for charging load profile of the electric vehicles (EVs) are part of the Demand Response (DR) program. For four scenarios of EVs charging load, the outcomes are compared using the three approaches Water Cycle Learning Algorithm (WCLA), Genetic Algorithm (GA), and Dynamic Programming (DP). The spring months are considered the highest season in the peak load and the highest in the electricity price among other seasons, the DR program is applied in this work in the Spring on both a workday and a special holiday to obtain the optimized scheduling of distributing generators to minimize the total cost and emission and maximize the profit and the revenue.