CC BY
7References
1. Flood Modeller's help file. - URL: www.floodmodeller.com (date of access: 10.10.2021).
2. Flood Modeller website. Case Studies section. - URL: www.floodmodeller.com (date of access: 10.10.2021).
3. Web site AppAdvice LLC. - URL: https://appadvice.com/app/flood-alert/420666016 (date of access: 10.10.2021).
4. Woodhouse Joanne. Flood impacts brought to life through science and technology // www.jbaconsulting.com. 23rd October 2017. - URL: https://www.jbaconsulting.com/knowledge-hub/flood-impacts-science-technology (date of access: 10.10.2021).
УДК 004.9
doi :10.Ш20/SPBPШ/id21 -409
Руссков Олег Владимирович1,
соискатель;
Сараджишвили Сергей Эрикович2,
доцент, канд. техн. наук, доцент
ЦИФРОВАЯ МОДЕЛЬ ДЛЯ ПРОГНОЗИРОВАНИЯ МАКСИМАЛЬНОГО ЭНЕРГОПОТРЕБЛЕНИЯ ЭНЕРГЕТИЧЕСКОГО РЕГИОНА
1 2
' Россия, Санкт-Петербург, Санкт-Петербургский политехнический
университет Петра Великого, 12 zenit-che@mail.ru, ssaradg@mail.ru
Аннотация. Многие процессы в промышленности описываются неравномерными временными рядами. Трудности прогнозирования неравномерных временных рядов создают необходимость разработки новых моделей прогнозирования. В статье обоснованы процессы цифровой трансформации отрасли. Показана важность создания и применения цифровых моделей прогнозирования. Рассмотрены преимущества и недостатки существующих моделей временных рядов. Предложен методический подход к прогнозированию неравномерных и изменчивых временных рядов. Рассмотрена цифровая модель прогнозирования неравномерного энергопотребления. Алгоритм работы модели показан в деталях. Описаны испытания и внедрение данной модели на промышленном предприятии. Анализируются преимущества предлагаемой модели по сравнению с существующими. Описаны экономические и экологические аспекты применения модели. Показано снижение углеродного следа в производимой продукции в результате использования предложенной модели в промышленности. Сделан вывод о применимости предложенного метода для прогнозирования неравномерных и изменчивых временных рядов.
Ключевые слова, неравномерный временной ряд, прогноз электроэнергии, цифровая модель, энергопотребление, энергосбережение.
Oleg Russkov1, PhD Student;
Sergei Saradgishvili2,
Associate Professor, Candidate of Technical Sciences
DIGITAL MODEL FOR FORECASTING THE MAXIMUM
POWER CONSUMPTION OF ENERGY REGION
1 2
' Peter the Great St. Petersburg Polytechnic University,
St. Petersburg, Russia, 1 2 zenit-che@mail.ru, ssaradg@mail.ru
Abstract. There are many uneven time series in industry. The difficulties of forecasting uneven time series create the need to develop new forecasting models. The processes of industry digital transformation are justified. The importance of creating and applying digital forecasting models is shown. The advantages and disadvantages of existing time series models are considered. The methodical approach to forecasting uneven and volatile time series is proposed. Digital model for forecasting uneven power consumption is considered. The algorithm of the model is shown in the details. Tests and implementation of this model at industrial enterprise are described. The advantages of the proposed model in relation to existing ones are analyzed. The economic and environmental aspects of the model application are described. The reduction of the carbon footprint in manufactured products as a result of using the proposed model in industry is shown. The conclusion about the applicability of proposed method for forecasting uneven and volatile time series is made.
Keywords: uneven time series, electrical power forecast, digital model, power consumption, energy saving.
Introduction
Nowadays information technologies have reached a level that allows them to be used in previously difficult to imagine areas. The whole world is passionate about digitalization. IT specialists and system architects work hard and magnates invest huge amount of money in digital technologies elaboration. In particular, digital industry [1] and digital energetics [2] are rapidly developing aimed at monitoring and managing all significant processes in the context of each unit with hourly discreteness. All units are characterized by time series data. Consequently the most important scientific and economic task is to predict time series with a horizon of one hour [3]. The most popular time series forecasting models [4] now are neural network models [5], autoregressive models [6], GARCH models [7], maximum similarity models [8] and exponential smoothing models [9]. It is known that the unevenness of time series increases with the hourly forecast horizon (herewith the series may be quite even in the context of the day). Therefore existing time series models drastically reduce their effectiveness [10], forecasting error significantly increases [11]. Alternative solution is the development of fundamental mod-
els [12] that take into account absolutely all factors affecting the studied series even the human factor [13,14]. But development of fundamental forecasting model for energy region consumption is still difficult at the current level of digital transformation. Therefore in order to overcome the increasing uneven time series forecast error it is necessary to develop new approaches and methods.
1. The Uneven Time Series Forecasting Method
Developed by authors methodological approach to forecasting uneven time series is a compromise between time series models and fundamental ones. The authors published models implemented on its basis previously [15]. The scheme of the approach is presented at the Fig. 1. The essence of the method is establishing a relationship of researched series with adjacent ones easier to forecast.
Fig. 10. Forecasting uneven time series methodical approach.
Feedback will allow calculating the future values of researched uneven time series.
2. Hourly Power Consumption Forecasting Model
The example of the method implementation is forecasting power consumption of an industrial enterprise. The volume of overall power consumption is determined by averaging the maximum power volumes in n working days of the month [16]. Herewith, the hours of maximum volumes are un-
known to the consumer in advance because this maximum relates to the local power system of Russian Federation region:
P — Average (Pmaxl, Pmax2> ■■■ > P-maxn), (1)
where P is overall power consumption volume; Pmaxi..Pmaxm are maximum hourly power consumption of working days.
Graphs of hourly power consumption of all consumers sum up and form a general graph of the local energy region (Fig. 2).
MW
Power volume per hour N accounts in the market for payment after averaging
End
3
Fig. 2. The scheme of forming the volume of power consumption
The hour of maximum power consumption is the hour of power control for the particular consumer. It is quite possible to accurately forecast a consumption profile of energy region using existing forecast models in the context of a day [17-20]. But this does not lead to success in forecasting the number of hour because hourly power consumption of the region is variable (Fig. 3).
Fig. 3. The variability of region hourly power consumption
Presence of similar peaks within one day does not allow determining the maximum hour with satisfactory forecast error (Fig. 4). That is why existing time series forecast models do not give expected result in this case. A new approach is required to solve this science problem.
2 4 6 8 10 12 14 16 18 20 22 24 Hour
Fig. 4. The errors of hourly maximum power forecast
Therefore time series of energy region power consumption consisting of 24*N values (where N is the number of working days in a month) can be mapped to adjacent series consisting of N numbers of factual maximum consumption hours (Fig. 5).
июнь июль август
17-19 18-19 19-21 19-21 20-21
January February 1 March April May September October November December
18 19 19 19 19 20-21 18-20 17-19 17-19
20 18 19 19 19 19 19 19 19
19 20 21 19 19 19 19 19 19
19 20 19 19 19 19 19 19 19
19 20 19 20 19
20 19 20 ■-♦— —■- -пкг
Fig. 5. Hours statistics of energy region per 3 years
The analysis of this adjacent time series shows that there are small (1-2 hours) groups of prevailing hours numbers in most months. Pareto diagram is most convenient tool for hours analysis (Fig. 6). Pareto principle is implemented: 20% of hours numbers determines 80% significance for overall forecast.
Fig. 6. Pareto diagram per September 2018-2020
Determining the set of hours for decreasing power consumption is a first step. Next step is calculating the volume of possible decrease. This volume (V) has not to make a disturbance in statistics of energy region maximum hours numbers:
V<V±- V2, (2)
where is the average daily maximum of hourly power consumption of energy region, v2 is the next largest volume of power consumption after v1 (Fig. 3).
A brief forecasting scheme is shown in Fig. 7. Pareto diagrams shown at the right side of Figure 7 are similar Pareto diagram in Figure 6.
cr"; Start
u
3 yea rs
Analysis of power peak
hours statistics of previous --- y/
3 years
Analysis of power peak hours statistics of previous 2 years
Analysis of power peak hours statistics of previous year
Determining the prevailing maximum hours
Analysis of the difference
between the peaks of region power consumption
Power reduction by calculated difference in forecasted hour (hours)
C^^End^^^ Fig. 7. Forecasting hours of region power consumption maximum
3 years is an optimal period covering both established and new trends. Period more than 3 years contains outdated data and trends. So the forecast module analyzes the statistics of energy region maximum hours per 3 years first. Then the set of hours per 2 previous years is analyzed. Set of hours per 1 year is analyzed finally. If the hours sets differ significantly then the hours from the period closer to the forecast year are taken into account. Thus the algorithm of forecast model accounts the latest trends. There is a transition from the unsuccessful forecasting uneven time series of energy region power con-
2 yea rs
sumption to much simpler forecasting even series of maximum hour numbers (Fig. 8).
At the same time relationship between both series is obvious and the result of forecast is the desired hour of uneven time series maximum. This forecast performs in full accordance with the methodological approach shown in Figure 1.
Fig. 8. Uneven and adjacent time series relationship
3. Results
The tests of offered model have been carried out at the metallurgical enterprise in Volgograd region of Russian Federation. The model forecasts a set of maximum power consumption hours of Volgograd region each month during 2020 year. Another forecast result was the maximum possible reduction of power consumption of the enterprise during these hours. This information used by specialists of energy power and steelmaking shop. They have developed the procedures to reducing the power consumption by optimizing the start of steel smelting without affecting the production indicators of enterprise.
The confirmed economic effect of this model implementation is 672 500 $ (with VAT) in 2020 year (exchange rate October 2021), the confirmed reduction of power consumption during the peak hours of Volgograd region is 73.6 MW. Comparison of achieved efficiency of the proposed forecasting method with efficiency of conventional forecasting (the ratio of the number of correctly forecasted hours to their total number) is shown at Fig. 9. Old method used one hour as forecasting result; new one uses set of hours as forecasting result. Its minimum forecast efficiency (55 %) was recorded in July 2020, the maximum (89 %) — in May 2020.
Fig. 9. The results of maximum power hours forecast per months 2020 year
The day production program of the plant is primary and has no reference to the hour marks of the day. Therefore it is not possible to forecast a power consumption decrease in accordance with production program. Therefore actions being developed to reduce power consumption should not concern the day program of the plant.
4. Discussion and conclusions
The electricity is the basis of the modern economy. So using a power consumption forecasting model at the industrial enterprises allows to relieve the Unified energy system of Russia during periods of power shortage in addition to obtaining an economic effect for power consumers. The application of the model helps to reduce the atmospheric emissions from inefficient generating stations. It saves non-renewable energy sources they operate on (gas, fuel oil, coal) and therefore reduce the carbon footprint of Russia products [21]. The enterprises using the described model to forecast their uneven unregulated electricity and power consumption get a fundamental opportunity to partici-
pate in new perspective market called "Energy net". The consumer using offered model will become an active participant in the energy market. This will allow keep enterprises workplaces when the market paradigm will be changed. This means that enterprises continue to participate in development of the region's economy because the most of large power consumers are city-forming plants. The algorithm developed on the basis of the described method is implemented by forecast models at three Russian metallurgical enterprises in 2020 year. However, the offered model is fundamentally applicable for any power consumers but metallurgical enterprises. With increasing the number of consumers using this model the economic, environmental and social impacts will grow up on a national scale.
References
1. The Factory of the Future. Industry 4.0 - The Challenges of Tomorrow. - URL: https://home.kpmg/content/dam/kpmg/pdf/2016/05/factory-future-industry-4.0.pdf (date of access 20.10.2021).
2. EnergyNet Market. - URL: https://energynet.ru/home-eng (date of access 20.10.2021).
3. Russkov O., Saradgishvili S. The Electricity Market Prices Forecast as Energy Efficient Procedure for an Industrial Monotown Enterprise. // Procedia Engineering. - 2015. -Vol. 117. - Pp. 309-316.
4. Yang J. F. Power System Short-term Load Forecasting: Thesis for Ph.d. degree. Elektrotechnik und Informationstechnik der Technischen Universitat, Darmstadt, 2006.
5. Haykin S. Neural networks. A comprehensive foundation, 2 edition. - Hoboken: Prentice Hall, 1999.
6. Draper N.R., Smith H. Applied Regression Analysis. - New York: Wiley, 1981.
7. Garcia R.C., Contreras J., Akkeren M., Garcia J.B.C. A GARCH forecasting model to predict day-ahead electricity prices. // In: IEEE Transactions on Power Systems. - IEEE, New York, 2005. - Vol. 20 (2). - Pp. 867-874.
8. Hastie T., Tibshirani R., Friedman J. The Elements of Statistical Learning. - Heidelberg: Springer, 2017.
9. Zahan M., Kenett R.S. Modeling and Forecasting Energy Consumption in the Manufacturing Industry in South Asia. // International Journal of Energy Economics and Policy. - 2013. - Vol. 1. - Pp. 87-98.
10. Beiden S., Smirnov S., Matveeva M. Electricity market risks and price forecast methods. // Energy market. - 2004. - Vol. 4. - Pp. 22-28.
11. Makoklyuev B.I., Yoch V.F. The forecast precision and the consumption graph variaty correlation. // Electric stations. - 2005. - Vol. 5. - Pp. 64-67.
12. Thomsett M. Fundamental Analysis of the Energy Market. // In: Investing in Energy - 2014. - Pp. 189-209.
13. Kaneman D. Think slow... solve fast. - Moscow: AST, 2013.
14. Rabin M. Psychology and economics. // Journal of economic literature. - 1998. -Vol. 36 No. 1. - Pp. 11-46.
15. Russkov O.V., Saradgishvili S.E. A digital method for correcting planned electric energy consumption as a step to the Energynet market. // In: SHS Web of Conf. - 2018. -Vol.44. - Paper 00073.
16. Regulations for determining the volume of purchase and sale of power in the wholesale market. - URL: https://www.np-sr.ru/ru/regulation/joining/reglaments/1978 (date of access 20.10.2021).
17. Campillo J., Wallin F., Torstensson D., Vassileva I. Energy demand model design for forecasting electricity consumption and simulating demand response scenarios in Sweden. // In: International Conference on Applied Energy ICAE 2012. - 2012. - A10599.
18. Bercu S., Proia F. A SARIMAX coupled modelling applied to individual load curves intraday forecasting. // Journal of Applied Statistics. - 2013. - Vol. 40, No. 6. -Pp. 1333-1348.
19. Adepoju G., Ogunjuyigbe S., Alawode K. Application of Neural Network to Load Forecasting in Nigerian Electrical Power System. // The Pacific Journal of Science and Technology. - 2007. - Vol. 8, No. 1. - Pp. 68-72.
20. Zhou X., Tang Y., Xie Y., Li Y., Zhang H. A Fuzzy Probability-based Markov Chain Model for Electric Power Demand Forecasting of Beijing, China. // Energy and Power Engineering. - 2013. - Vol. 5. - Pp. 488-492.
21. Russkov O.V., Saradgishvili S.E. IT-Method for uneven energy consumption planning. // In: 2017 International Conference on Industrial Engineering, Applications and Manufacturing IEEE. - 2017. - Pp. 1-4.
УДК 330.15:332:338:504:502.6:550. 8.028 :10.18720^РВРи/2М21 -410
Вильдероттер Клаус 1,
профессор, доктор естественных наук, доцент;
Кононова Мария Юрьевна1, приглашённый профессор, доктор технических наук, доцент;
Кнёферл Марина3, студентка 3 курса
ГЕОЭКОЛОГИЧЕСКИЙ МАРКЕТИНГ И СЕЛЬСКОЕ ХОЗЯЙСТВО 4.0: ПЕРСПЕКТИВЫ И РИСКИ ЦИФРОВИЗАЦИИ
В ЭКОНОМИКЕ ГЕРМАНИИ
1 2 Германия, Розенхайм, Технический университет прикладных наук Розенхайма,
1 Klaus.Wilderotter@th-rosenheim.de,
2 Mariia.Kononova@th-rosenheim.de;
3 Германия, Эссен/Мюнхен, ФОМ Университет Мюнхена/Эссена
Аннотация. После формирования согласованного понимания об Индустрии 4.0 закономерным становится её внедрение во все отрасли народного хозяйства и главное в сельское хозяйство 4.0 с перспективой и активами на сельское хозяйство 5.0. Требования геоэкологического маркетинга и оценки экологического следа сельскохозяйственной деятельности возможны при повсеместном экологическом учёте на базе данных дистанционного зондирования Земли и ГИС сопровождении. Перспективам и рискам данных инноваций в сельском хозяйстве посвящён данный проект.
