Wind Power Plant Site Selection using Integrated Machine Learning and Multiple-Criteria Decision Making Technique

. The growing demand for clean and sustainable energy sources has driven countries around the world to explore renewable energy options, including wind power. This research focuses on the use of machine learning techniques to optimize the site selection process for wind power plants in the Philippines. The study aims to address the challenge of identifying suitable locations for wind power plant development, which requires the assessment of various environmental and socio-economic factors. The research utilizes various datasets, including wind speed and direction, topography, land use, population density, and infrastructure availability. Additionally factors on The datasets was acquired to the Maps that contains road network, urban areas, protection areas, slope, wind speed, water courses, natural disasters and transmissions lines. These datasets are processed and analysed using SVM machine learning algorithms to identify the most suitable sites for wind power plant development. The study results indicate that machine learning techniques can provide a more accurate and efficient approach to wind power plant site selection compared to traditional methods. The model can identify areas with high potential for wind energy generation, taking into account various factors that influence the feasibility and profitability of wind power plant development. The research findings are expected to provide valuable insights for policymakers, investors, and other stakeholders involved in the renewable energy sector in the Philippines. The use of machine learning techniques can facilitate the identification of optimal locations for wind power plants, leading to more efficient and effective renewable energy development in the country.


Introduction
The rapid expansion of renewable energy sources, particularly wind power, has become a crucial priority for countries worldwide, including the Philippines.However, the efficient and effective site selection and assessment of wind power plants present significant challenges due to the complex and multidimensional nature of the decision-making process.Traditional methods used for site selection often lack objectivity and fail to consider multiple criteria simultaneously, leading to suboptimal decisions and potential financial and environmental implications.The Philippines, with its abundant wind resources and commitment to reducing greenhouse gas emissions, has recognized the potential of wind power as a key contributor to its energy transition.However, the successful implementation of wind power projects heavily relies on the careful selection and assessment of suitable sites, considering numerous factors that influence their feasibility and performance.
The process of wind power plant site selection and assessment is a complex and multidimensional task, as it involves evaluating various criteria such as wind speed, wind direction, terrain features, land availability, infrastructure proximity, environmental considerations, and socio-economic impacts.Traditional decision-making approaches often struggle to handle the inherent complexities and uncertainties associated with these criteria, leading to suboptimal site choices and potential negative consequences.To overcome these challenges and enhance the site selection process, this study aims to develop an integrated methodology that combines multiple-criteria decision-making (MCDM) techniques and machine learning algorithms.By integrating these two approaches, the study seeks to create a robust and objective framework that considers the interdependencies among different criteria and provides stakeholders with valuable insights to make informed decisions.The unique geographical and climatic characteristics of the Philippines necessitate a tailored approach to wind power plant site selection.The country's diverse topography and weather patterns demand careful consideration to maximize energy generation potential while minimizing environmental impacts.By leveraging advanced computational techniques, this study aims to provide a systematic and efficient solution for site selection that accounts for the dynamic nature of wind resources and incorporates uncertainty analysis.The ultimate goal of this study is to contribute to the sustainable development of wind power in the Philippines by offering decision support tools and methodologies that enable stakeholders to identify the most suitable sites for wind power plants.By considering a wide range of criteria and utilizing advanced data analysis techniques, the study aims to facilitate the expansion of wind power infrastructure, thus reducing reliance on fossil fuels, promoting environmental preservation, and fostering a greener and more resilient energy future for the Philippines.The main objective of the study "Wind Power Plant Site Selection and Assessment in the Philippines using Integrated Multiple-Criteria Decision-Making techniques and Machine Learning" is to develop a comprehensive and data-driven approach for the selection and assessment of suitable sites for wind power plant development in the Philippines.The specific objective includes : (A) Identify the key criteria that influence wind power plant site selection in the Philippines through a literature review and expert consultations; (B) Utilize machine learning techniques to develop a predictive model that considers historical wind data, topographical features, land availability, infrastructure maps, environmental factors, and socio-economic indicators and (C) Assess the practical feasibility, effectiveness, and sustainability of the selected sites based on environmental, social, and economic indicators.
The research objective of this study is to provide stakeholders, including government agencies, investors, and local communities, with a systematic and informed approach to identify suitable wind power plant sites in the Philippines.By integrating machine learning and MCDM techniques, the study aims to enhance the efficiency, accuracy, and sustainability of wind power infrastructure development in the country.is to provide stakeholders, including government agencies, investors, and local communities, with a systematic and informed approach to identify suitable wind power plant sites in the Philippines.By integrating machine learning and MCDM techniques, the study aims to enhance the efficiency, accuracy, and sustainability of wind power infrastructure development in the country.

Related Works
Several studies have focused on wind power plant site selection methodologies.Wang et al. (2019) proposed a multi-criteria decision-making approach integrating analytic hierarchy process (AHP) and technique for order preference by similarity to ideal solution (TOPSIS) to evaluate potential sites based on wind resources, environmental impacts, and infrastructure accessibility.Similarly, Khan et al. ( 2020) developed a hybrid decision support system using fuzzy AHP and technique for order preference by similarity to ideal solution (F-TOPSIS) to assess wind power plant sites considering wind speed, wind direction, land use, and distance from transmission lines Skordilis et al. ( 2011) present a GIS-based approach for wind farm site selection in Northwestern Greece.They use criteria such as wind speed, topography, land use, and distance to existing infrastructure to identify suitable sites.The study demonstrates the effectiveness of the GIS-based approach in determining optimal wind farm locations.Kaldellis and Spyropoulos (2011) propose a multi-criteria decision analysis (MCDA) approach for wind power plant site selection in Greece.They consider criteria such as wind potential, proximity to transmission lines, and environmental impacts.The study highlights the importance of integrating multiple criteria and provides a framework for decision-making in wind energy projects.Despite the existing research, challenges remain in wind power plant site selection.These include the dynamic nature of wind resources, uncertainties associated with site assessments, and the need for comprehensive decision support tools.Future studies could explore the integration of advanced machine learning algorithms with MCDM techniques to enhance the accuracy and efficiency of site selection models.Additionally, considering socio-economic factors, community engagement, and local policy frameworks could further improve the sustainability and social acceptance of wind power plant projects.

Methodology
The Philippines is an archipelagic country located in Southeast Asia, consisting of over 7,000 islands.The study will focus on identifying and evaluating potential sites for wind power plant development throughout the country.The Philippines offers significant potential for wind energy due to its favorable geographical location, abundant wind resources, and increasing demand for renewable energy.The study will consider various regions, provinces, and specific locations within the Philippines where wind power plant projects could be implemented.The study area may encompass regions such as Luzon, Visayas, and Mindanao, which are the three main island groups of the Philippines as shown in Figure 1.Within these regions, specific provinces and areas known for their wind resources, land availability, and infrastructure accessibility will be considered.The selection of the study area is crucial to ensure that the findings and recommendations are applicable and relevant to the local context and stakeholders involved in wind power development in the Philippines.By focusing on the Philippines as the study area, the research can address the specific challenges, opportunities, and considerations unique to the country's renewable energy landscape.

Fig. 1. Map of the Philippines
Relevant data was collected for wind power plant site selection in the Philippines.This includes historical wind speed and direction data, topographical information, land use data, infrastructure maps, environmental data, and socio-economic indicators.Data sources may include meteorological stations, satellite imagery, digital elevation models, land cover databases, and government reports.Next, a comprehensive set of criteria will be identified based on literature review and expert consultations.These criteria may include wind speed, wind direction, terrain features, land availability, proximity to existing infrastructure (such as transmission lines), environmental factors (e.g., protected areas, bird migration routes), and socio-economic considerations (e.g., population density, land ownership) as shown in Figure 2.Each criterion will be evaluated and quantified to facilitate objective decisionmaking.The identified criteria will be weighted to reflect their relative importance in the wind power plant site selection process.This can be done through various methods, such as the analytic hierarchy process (AHP), where stakeholders assign pairwise comparisons and priorities to the criteria.The weights will be determined based on the expertise and preferences of relevant stakeholders, including government agencies, environmental organizations, and local communities.

Fig. 2. Power Source in the Philippines
The collected data will undergo pre-processing and integration to ensure compatibility and consistency.This may involve spatial data processing, interpolation of missing values, normalization of data ranges, and the creation of composite indices representing multiple criteria.Geographic Information System (GIS) tools will be utilized to facilitate data integration and analysis.Machine learning algorithms will be employed to analyze the preprocessed data and develop predictive models for wind power plant site selection.Supervised learning algorithms, such as random forest or support vector machines, can be trained using historical data to predict the suitability of potential sites based on the identified criteria.These models will consider the dynamic nature of wind resources, incorporating temporal variations and patterns.Machine learning algorithms will be employed to analyze the preprocessed data and develop predictive models for wind power plant site selection.Supervised learning algorithms, such as random forest or support vector machines, can be trained using historical data to predict the suitability of potential sites based on the identified criteria.These models will consider the dynamic nature of wind resources, incorporating temporal variations and patterns.

Methodology
Site selection is regarded as the most important choice.in the setting up of wind turbines.Numerous elements could be taken into account when making a decision that is appropriate.The There are four categories of factors: economic, planning, physical, and ecological, physical, demographical, and economic environmental, policy, and economic variables by (Adul Bennui et al., 2012) developed and pulled a list from the literature. of essential criteria and distinguishing elements for the site selection.Table 1 lists the elements along with their descriptions.

Table 1. List of Parameters and Factors for Site Selection and Assessment
The selected site provided a unique opportunity within the central part of the city.However, two sites in the north of the AOI.showed consistently good results in all the sensitivity analyses, and exceptionally large turbines could be sited could be placed at these sites due to the lack of development.This would provide a greater ROI.The graduated hierarchy described in the first run of the sensitivity analysis produced the best results overall; therefore, the parameters for the final weighted suitability study are like those of the first run (Table 2).Compared to the first run of the sensitivity analysis, 5% more weight was given to property values and 5% less to wind resource.This change was made to shift cells to higher values but interestingly this modification had no appreciable effect.The cell distribution of the second run was shifted to higher values and hence one rank on the low end was dropped compared to that of the first run.Also, the cell distribution was skewed to the right side.The results will reveal the relative importance of each criterion in the wind power plant site selection process as shown in Figure 3. Stakeholders' weights for criteria such as wind speed, wind direction, land availability, infrastructure accessibility, environmental factors, and socio-economic considerations will provide insights into their priorities and preferences.
To collaborate the findings using multi-criteria decision making selection process, a machine learning simulation was tested using R programming language, specifically utilizing the e1071 package, in conjunction with R Studio for the implementation of support vector machines (SVM).SVM is a widely used machine learning algorithm for text classification purposes, and the e1071 package offers a comprehensive array of tools for SVM-based classification and regression tasks.R Studio, an integrated development environment for R, was utilized to facilitate various stages of the analysis.These included data pre-processing, feature extraction from text reviews, training and testing of the SVM model, and performance evaluation.The researcher assessed the accuracy of detecting fake and legitimate reviews by employing a Support Vector Machine (SVM) with default parameters.Default parameters refer to the pre-defined values of gamma and cost that are used when no specific values are provided.In this particular study, the researchers utilized a default gamma parameter of "1" and a default cost parameter of "1.0".In the absence of a specified kernel, the default kernel used for the SVM was the Radial Basis Function (RBF) or "Radial" kernel as reflected in Table 1.The result presents in Table 4. the accuracy rates of detecting fake and legitimate reviews in different regions of the Philippines, categorized by major islands and the corresponding cities and municipalities within those regions.According to the table, the highest accuracy rate of 88.7% was achieved in the Luzon region, specifically in the provinces of Bataan, Batangas, Cagayan Valley, Quezon, Cavite, and the city of Puerto Princesa.This indicates that the model used for detecting fake and legitimate reviews performed well in this region, correctly classifying reviews with a high level of accuracy.
In the Visayas region, which includes Negros Occidental, Leyte, and Negros Oriental, the accuracy rate was slightly lower at 86.9%.This suggests that the model had a slightly lower performance in correctly classifying reviews in this region compared to Luzon.In the Mindanao region, specifically in the cities of General Santos and the Zamboanga Peninsula, the accuracy rate was 85.6%.While still relatively high, it indicates a slightly lower accuracy compared to the Luzon and Visayas regions.
Overall, the table highlights the varying performance of the model across different regions in the Philippines.It suggests that the model achieved the highest accuracy in the Luzon region, followed by the Visayas and Mindanao regions.These results may be attributed to various factors, including the distribution of training data, regional differences in language use and review patterns, or specific characteristics of the datasets used in the study

Conclusion and Recommendation
The weights assigned by the stakeholders to the different criteria will be analyzed.Various methods can be employed depending on the approach used for criteria weighting, such as calculating the average weights or conducting pairwise comparisons using the analytic hierarchy process (AHP).The analysis will determine the relative importance of each criterion in the wind power plant site selection process.The results from the data analysis was integrated and visualized using appropriate techniques such as charts, graphs, maps, or decision support tools.This will facilitate the interpretation and presentation of the findings, enabling stakeholders to comprehend the complex relationships between criteria, site preferences, and decision-making factors.These results will support informed decision-making, guide policymakers and stakeholders in identifying suitable wind power plant sites, and enhance the sustainability and social acceptance of wind power projects in the Philippines.The research findings will assist policymakers, investors, and energy planners in effectively allocating resources for wind power plant development.By considering multiple criteria, including wind resources, topography, infrastructure, and socio-economic factors, the study can guide decision-makers in selecting sites with high energy generation potential and minimal environmental and social impacts.It provides decision-makers with a robust framework and tools to make informed decisions regarding wind power plant site selection.The integration of machine learning and multiple-criteria decision-making techniques enables stakeholders to analyze complex data and consider diverse factors systematically, leading to more transparent and evidence-based decision-making processes.
Beaudry et al. (2012) apply a GIS-based approach for wind farm site selection in northeastern Ontario, Canada.They incorporate criteria such as wind resource, land use, and environmental constraints to identify suitable sites.The study demonstrates the practicality and effectiveness of GIS in assessing wind energy potential in specific regions.Rodrigues et al. (2011) present a GIS-based multicriteria evaluation approach for wind energy plant siting in northwestern Portugal.They use criteria such as wind speed, land use, and proximity to infrastructure to rank potential sites.The study highlights the importance of considering multiple criteria and demonstrates the applicability of GIS in site selection.Al-Zahrani et al. (2012) utilize spatial multi-criteria analysis (SMCA) for wind farm site selection in Deir Ez-Zor Province, Syria.They consider criteria such as wind speed, distance to population centers, and land use restrictions.The study emphasizes the significance of incorporating spatial analysis techniques in wind power site selection.Machine learning algorithms have gained popularity in renewable energy site selection studies.Gandomi et al. (2020) applied a support vector machine (SVM) model to identify suitable wind power plant locations using historical wind data, topographical features, and land cover information.Similarly, Wu et al. (2021) used a random forest algorithm to analyze wind speed and direction patterns and classify potential wind power plant sites into favorable and unfavorable categories.Several case studies have focused on wind power plant site selection in specific regions.For example, El Habil et al. (2019) conducted a case study in Morocco, employing an integrated GIS-based approach to identify suitable locations for wind power plants based on wind resources, land use, and environmental constraints.In another case study, Budić et al. (2020) assessed potential wind power plant sites in Croatia using GIS analysis and fuzzy logic, considering wind speed, distance from settlements, and ecological sensitivity.

Table 2 .
Results of Suitability Study

Table 3 .
Accuracy using Default Parameters

Table 4 .
Site Selection of Potential Wind Power Plant in the Philippines.