sign in sign up
<<
Home , Decision support system

Database Systems Journal vol. IV, no. 1/2013 37

Decision Support System in National Power Companies. A Practical Example (Part I)

Adela BÂRA, Iuliana BOTHA, Ion LUNGU, Simona-Vasilica OPREA Academy of Economic Studies, Bucharest Transelectrica Company, Bucharest [email protected], [email protected], [email protected], [email protected]

The paper presents the developing stages of the decision support prototype in which the and the presentation level are built and validated. The paper also extends the results published in the 12th international conference on Informatics in Economy (IE 2013) proceedings and will presents the major steps for developing the data warehouse that integrates the sources from the Wind Power Plants (WPP) from the national parks and also the interface modules that allow managers to analyze data at a central level. Keywords: Data Warehouse, , Analytics, Decision Support System

Introduction integrates the sources from the Wind 1The matter of developing decision Power Plants (WPP) from the national support systems has now become a trend in parks and also some interface modules that the national companies. The large volumes allow managers to analyze data at a central of heterogeneous data gathered from level. different types of measuring devices and different formatting sources represent a 2. Considerations regarding data challenge for developing integration warehouses solutions for extracting, transforming and The solution adopted for the informatics loading data into a central data warehouse system prototype that we are going to from where data can be analyzed and develop is based on data warehouse presented through special analytic tools implementation using a multidimensional like Business Intelligence (BI). This model (MOLAP). The data warehouse scenario is the base for our case study and built at this stage will allow metrics also for the research project in the National aggregation and also cubes processing and Grid Company that has the objective to storage. develop a prototype for decision support Usually the data warehouse architecture system for analyzing the renewable varies depending on the specific situation resources like wind power. Also, these of each organization. For a basic aspects have been presented in a limited architecture (figure 1), data is loaded from manner in the paper [1]. one or more sources, and the users directly The system’s architecture consists in four access the data warehouse [5]. levels: data, model, presentation and The architecture of a data warehouse has telecommunications. The architecture has three main components, as shown in the been presented in previous papers like [2], figure below: [3]. The aim of the present paper is to · Data warehouse itself and its present the developing stage of the system; prototype in which the data warehouse and · Data acquisition system from OLTP the presentation level are built. We’ll systems and other sources; describe some of the major steps for · Data analysis and presentation system. developing the data warehouse that

38 Decision Support System in National Power Companies. A Practical Example (Part I)

Fig. 1 – Main components of a data warehouse Source: Authors

A more complex architecture uses processing in terms of: integrity, precision, integration and cleaning systems and also accuracy, format. multiple systems designed for The data collections level contains each organization department, as indicated processed data, aggregated and loaded into in [5]. multidimensional structures on different As shown in figure 2, such complex levels prepared for being used in analysis. architecture is divided into four different At this level we can multiple data levels to achieve the data, as follows: mart systems for enterprise departments. sources, transformations, collections of The data analysis level involves extracting data, analysis. data from data warehouse and using The data sources level collects business intelligence technologies for the heterogeneous data from various internal analysis and interpretation of the or external operational systems of the information provided. At this level are organization. It uses a process of data used: OLAP products for analysis (so that integration using a separate module called data are presented graphically, in data source module. , embedded in portals etc.), The data transformations level uses a features for data retrieval using more process of extracting, processing (cleaning) criteria, products, complex and and data loading (ETL - Extract, dynamic reports. Transform, Load), which requires data

Fig. 2 – Complete architecture of a data warehouse system Source: Authors

From the above architecture we can obtain The operational module is represented by three functional modules of data organizational data which are usually warehouse development: operational, stored in different formats at different central, and strategic [5]. locations. These data may come from

Database Systems Journal vol. IV, no. 1/2013 39 applications or operational systems from which is due to the advantages offered by companies or from external sources. the various processes of decision making Regardless of their source, the data must and analysis. By using different ways to be collected and brought into a consistent access data and use processing format to be used. This process of data technologies available, users can get transformation is the basis for building a information that will help in the process of consistent data warehouse. Data establishing company strategy. In this way, transformation is a complex process of data are prepared for interpretation and extracting, filtering, cleaning, analysis using specific instruments compounding, validating and loading data (graphics, presentations, dynamic reports, (ETL). web browsers, data etc.). The central data warehouse module is represented by software products, servers 3. Developing the data warehouse on which they run, chosen implementation As we shown in [1], to implement the for the data warehouse. This module can model we have chosen Oracle Warehouse be implemented in two versions: 1) Builder 11g development environment implementation of a distributed system, (documentation available at [4]) and we where data is stored in decentralized, have followed specific steps: creating the independent units (Independent Data source module, designing destination Marts), each containing data relevant to a module, making mappings, creating and particular aspect of organization validating objects. operations; 2) implementation of a unique, Developing the data source module means centralized data source, which can be actually importing definitions accessed by users from all departments of from external sources of data (figure 3). the institution. All data sources need to be mapped to the The strategic module (business module) data warehouses objects which are created offers the final value of a data warehouse when designing the destination module.

Fig. 3 – Creating the data source module Source: [1]

Depending on analysis requirements of the structural entities that will be part of the decision makers, we have identified the organization and selection of dimension tables, which are the main multidimensional data. The dimensions are

40 Decision Support System in National Power Companies. A Practical Example (Part I) organized on several levels, so as to permit Central tables containing measures typical OLAP operations: rotation, attributes (metrics) and foreign keys to sections, navigation in detail (drill down) dimension tables are called data cubes. and aggregation (drill up / roll up). They usually contain numeric data that can For example, DIM_CENTRALA be aggregated and analyzed based on dimension contains attributes related to existing levels in dimensions [5]. wind turbines and wind power stations and Particularly important for subsequent is based on a virtual table created on analysis and data presentation is relational tables CENTRALE_T, CUB_PRODUCTIE cube that contains STATII_RACORD_T and TURBINE_T: metrics related to energy produced by wind turbines. This cube is based on a virtual CREATE OR REPLACE VIEW table created on tables PRODUCTIE_T dim_centrale AS and TURBINE_T: SELECT s.id_statie_racord, s.denumire denumire_statie, s.id_locatie, c.id_centrala, CREATE OR REPLACE VIEW c.denumire denumire_centrala, fact_productie AS c.id_investitor, SELECT t.serie_turbina, c.putere_instalata PI_Centrala, t.data_punere_fct, t.data_revizie, c.nr_contract, t.serie_turbina, p.data, p.viteza, p.directie, t.puterea_instalata PI_turbina, p.energie_produsa, t.data_punere_fct, p.stare_turbina, t.id_loc, t.id_producator, t.imagine t.id_producator FROM turbine_t t, centrale_t c, FROM productie_t p, turbine_t t statii_racord_t s WHERE WHERE t.id_centrala=c.id_centrala p.serie_turbina=t.serie_turbina; AND c.id_statie_racord=c.id_statie_rac The design of data warehouse schema is ord; made using joins between dimension and fact tables within cubes. In our case, a In this case, the hierarchical levels in snowflake data warehouse schema model DIM_CENTRALA dimension are: was built (figure 4), because there are joins L_STATIE - L_CENTRALA - L_TURBINA. also between dimensions [1].

Fig. 4 – of the data warehouse Source: [1] All objects of the data warehouse are procedures and functions for creating stored in database, for each object being attributes, hierarchical levels of the created a set of specific PL/SQL

Database Systems Journal vol. IV, no. 1/2013 41 dimensions, measures and joins within OTS to analyze the renewable production. cubes. The application consists in two components: one for current activities and 4. Developing the presentation level another one for decision support made by a The presentation level consists in two set of analytical reports. major components: interface modules for Data from the wind power plants will be local Wind Power Plants and interface presented in different formats for the modules for central management executives, like: reports, dashboards, pivot (executives and operative department from tables, charts, and interactive maps. central level). Each module has two In the following sections we’ll present only components: one for current activities like some components of the BI modules that production, consumption, resource allow managers to analyze the evolution of allocation, short term prediction, and wind generation, consumption and another component for data analysis based allocation in different periods of time in on OLAP model for historical and the selected zones and parks. The predictive reporting. interfaces consists of a set of dashboards The modules for current management are and reports that presents data in a graphical developed using JDeveloper environment and intuitive way and available through and the modules for analysis are developed mobile devices. using Oracle Business Intelligence Suite. Another facility that we’ve implemented is a geographical representation for an Interface level for the national operator intuitive analysis of renewable resources in In this step we developed the analyses different areas and wind parks. reports and dashboards for the national

Fig. 5 - Interactive maps for analyzing WPP production Source: [1]

The map in figure 4 visually depicts the Figure 5 shows an interactive map for the energy consumption and production for current studies and notices correlated with each connected station. When selecting a the actual contracts between the national geographic area and then the station OTS and the investors. marked the specific indicator is presented in detail.

42 Decision Support System in National Power Companies. A Practical Example (Part I)

Fig. 6 – The studies and notices correlated with the actual contracts Source: Authors

The results are presented through an OLAP The figure 7 presents the connection report with navigation facilities between stations distributed over the regions having hierarchies (based on 1:n relationship): marked the production and the drill-down, roll-up, slice/dice. consumption.

Fig. 7 – Analytical report with the connection stations over the geographic regions Source: Authors

Analytical reports at central level environment that offers different facilities Based on the data warehouse we’ve for multidimensional analyses and developed a set of analytical reports for graphical elements based on the virtual executives from the national level in order data warehouse. to observe and analyze the wind power In this step, by developing the analytical plants connected in the system. reports for executives, we can validate the For developing we used the Oracle data warehouse functionalities and we can Business Intelligence Discoverer establish new requirements and facilities

Database Systems Journal vol. IV, no. 1/2013 43 that will be implementing in the next · Reports for analyzing the consumption phase. in periods of time and different At this stage, by making useful reports for measuring points; the executives, the data warehouse · Reports for analyzing the functioning functionality is validate, and new of national power over different requirements and features may be periods of time; implemented in the next phase of the · Reports for wind predictions correlated prototype. We have built several sets of with wind generation in different reports for different types of activities WPPs. pursued in the analysis: For example, in the figure 8 the predictions · Reports for grid connection, studies in for wind generation for the current month progress and contracts in progress for are represented in the same chart with the WPPs; actual production and one can observe that · Reports for wind generation and the differences recorded are small. The production over time and in different predictions are made by using data mining regions; techniques described in [3].

Fig. 8 - Wind generation predictions vs. actual production Source: [1]

Reports for current production at regions and meteorological conditions. The national level over time analyses are realized for different types of The production is observed and analyzed at turbines and producers due to noticeable national level in correlation and over differences.

44 Decision Support System in National Power Companies. A Practical Example (Part I)

Fig. 9 – The correlation between different WPPs Source: Authors

The report presented in figure 9 allows processing cubes, proper data validation users to analyze the correlation between and analysis. hourly productions from WPPs at national The analytical module validates the data level by selecting the quarter, month and warehouse functionalities and offers an day as initial parameters. The reports are intuitive and interactive interface for flexible, dynamic and allow the user to executive and high-level managers to select the WPP for which the Scatter analyze the major components of WPPs: diagram is presented. production, consumption and resource allocation. Conclusions and future work The current research was conducted within Acknowledgment a project won through national This paper presents some results of the competition. In order to achieve rapid research project PN II, TE Program, Code validation of data warehouse 332: “Informatics Solutions for decision functionalities, the 2011 phase of the making support in the uncertain and research project has used virtual objects unpredictable environments in order to and a ROLAP model (Relational OLAP). integrate them within a Grid network”, Thus, we designed the star / snowflake financed within the framework of People schema of data warehouse using Oracle BI research program. Discoverer. But the solution adopted for the prototype References implementation is based on a real data [1] Adela Bâra, Iuliana Botha, Ion Lungu, warehouse which use multidimensional Simona-Vasilica Oprea, “Prototype for model (MOLAP), implemented using a Decision Support System in National Oracle Warehouse Builder 11g. In this way Power Companies”, Proceedings of the the system performance is enhanced, and 12th International Conference on are allowed in an advanced manner Informatics in Economy (IE 2013), aggregation of measures, storage and 2013, pp. 381-385, ISSN: 2284-7472

Database Systems Journal vol. IV, no. 1/2013 45

[2] Iuliana Botha, Adela Bâra, Simona- trends in , Vasilica Oprea, Ticiano Costa-Jordão, 2012, pp. 123-129, ISSN: 1790-5190, “Integrating XML Technology with ISBN: 978-1-61804-134-0 Object-Relational into [4] Oracle Warehouse Builder User’s Decision Support Systems”, Database Guide, 11g Release 1, B31278-06, Systems Journal [Online], vol. III, no. Available: 1/2012, pp. 11-20, ISSN: 2069–3230. http://docs.oracle.com/cd/B28359_01/o Available: wb.111/b31278.pdf http://dbjournal.ro/archive/7/7_2.pdf [5] Ion Lungu (coord.), Adela Bâra, [3] Simona-Vasilica Oprea, Iuliana Botha, Constanţa Bodea, Iuliana Botha, Vlad Adela Bâra, Anda Velicanu, “Prototype Diaconiţa, Alexandra Florea, Anda of a Decision Support System for Velicanu, “Tratat de baze de date. Vol analyzing and forecasting the Wind I. Baze de date. Organizare, proiectare Energy Production in Romania”, şi implementare”, Bucharest, ASE Proceedings of the 1st International Publishing House, 2011, ISBN 978- Conference on Information Technology 606-505-472-1, ISBN vol. 978-606- and Computer Networks (ITCN '12), 505-481-3, pp. 402-408. Vienna, Austria, Recent advances in computer engineering Series, Latest

Adela BÂRA is a Lecturer at the Economic Informatics Department at the Faculty of , and Economic Informatics from the Academy of Economic Studies of Bucharest. She has graduated the Faculty of Economic Cybernetics in 2002, holds a PhD diploma in Economics from 2007. She is the author of 7 books in the domain of economic informatics, over 40 published scientific papers and articles (among which over 20 articles are indexed in international databases, ISI proceedings, SCOPUS and 10 of them are ISI indexed). She participated as team member in 3 research projects and has gained as project manager one research contract, financed from national research programs. She is a member of INFOREC professional association. From May 2009, she is the director of the Oracle Excellence Centre in the university, responsible for the implementation of the Oracle Academy Initiative program. Domains of competence: Database systems, Data warehouses, OLAP and Business Intelligence, Executive Information Systems, Decision Support Systems, Data Mining.

Iuliana BOTHA is an Assistant Lecturer at the Economic Informatics Department at the Faculty of Cybernetics, Statistics and Economic Informatics from the Academy of Economic Studies of Bucharest. She has graduated the Faculty of Cybernetics, Statistics and Economic Informatics in 2006 and the Databases for Business Support master program organized by the Academy of Economic Studies of Bucharest in 2008. Currently, she is a PhD student in the field of Economic Informatics at the Academy of Economic Studies. She is co-author of 5 books, 12 published articles (2 articles ISI indexed and the other 10 included in international databases), 18 scientific papers published in conferences proceedings (among which 6 paper ISI indexed). She participated as team member in 4 research projects that have been financed from national research programs. From 2007 she is the scientific secretary of the master program Databases for Business Support and she is also a member of INFOREC professional association. Her scientific fields of interest include: Databases, Database Management Systems, Design of Economic Information Systems, Grid , e-Learning Technologies.

46 Decision Support System in National Power Companies. A Practical Example (Part I)

Ion LUNGU is a Professor at the Economic Informatics Department at the Faculty of Cybernetics, Statistics and Economic Informatics from the Academy of Economic Studies of Bucharest. He has graduated the Faculty of Economic Cybernetics in 1974, holds a PhD diploma in Economics from 1983 and, starting with 1999 is a PhD coordinator in the field of Economic Informatics. He is the author of 22 books in the domain of economic informatics, 57 published articles (among which 2 articles ISI indexed) and 39 scientific papers published in conferences proceedings (among which 5 papers ISI indexed and 15 included in international databases). He participated (as director or as team member) in more than 20 research projects that have been financed from national research programs. He is a CNCSIS expert evaluator and member of the scientific board for the ISI indexed journal Economic Computation and Economic Cybernetics Studies and Research. He is also a member of INFOREC professional association and honorific member of Economic Independence academic association. In 2005 he founded the master program Databases for Business Support (classic and online), who’s manager he is. His fields of interest include: Databases, Design of Economic Information Systems, Database Management Systems, Decision Support Systems, Executive Information Systems.

Simona Vasilica OPREA is a Senior Engineer and Project Manager at Transelectrica National Power Grid Company. She has graduated the Polytechnic University in 2001, holds a Master Diploma in Infrastructure Management Program, Yokohama National University, Japan in 2007 and a PhD diploma from 2009. She is the author of over 20 articles, from which 3 ISI Web of Science indexed and 2 included in SCOPUS international database. Domains of competence: wind farm, investment opportunity analysis, studies of prognosis, stationary and dynamic regimes, short circuit calculations.