Main Autobiography Master's work Library Links Report about thr search The individual task Contacts

Theme: “The working out of the computer subsystem of analysis and forecasting of power resources consumption”

1. Introduction. Basing of the topic actuality

Aggrasation of the competition in conditions of permanent prices elevation for row-materials, transport and power resources makes enterprises to hold the necessary level of profitableness. There are several ways to low the working expencies. This work is dedicated to one of them and exactly – to the lowering of power-resources expencies in production. It is not a secret for everyone that this kind of production costs is the quarter of cost price of many goods and services.
 One of the main expencies articles in budget of any industrial enterprise is power resources expencies. That is why getting out the complete information about the usage of all kinds of energy the possibility of analysis of this information, forecasting and management of power resources consumption on each step of production have strategic importance.
This is a rule known by any leader. Nevertheless at the present not every enterprise use means of effective control and management of power resources. The reasons are different: lack of money, other priorities or just misunderstanding of the importance of this question. But life shows the truth and facts prove a lot. Those enterprises where there are workable automatic systems of control and analysis of heat and power resources have low power inputs. The usage of such systems is a tool the right expolation of which helps to reach the goal.

2. Goals and tasks of work

Information and analytical system (IAS) “Control and analysis of heat and power resources” is determined to the automation of control and analysis of resources consumption, financial indicators of payment, calculations of company’s management in power – usage and making up of limits
The important condition for effective work of  system is automation of power consumption and payment data receiving process.
Information placed in system database can be successfully used for solving of analytical issues and the support of decision making in questions of enterprise management.
The main goal of developing and using of IAS is automation of control and analysis tasks of heat and power resources consumption, limits of power and money savings.
The general conception of a system is shown in picture 1.

 

3. Offered scientific innovation

While studing the question of power resources consumption it should be taken into consideration different parameters such as seasons, prices and others).
It should be determined how they are connected with the consumption and given model should be realized in a program product. It is also important to decide which of the methods of the analysis and forecasting’s the most optimum under received to models.

4. Offered practical value

The fulfillment of master’s degree work supposes the security of the following values:

1. Collection of the information about the usage of all kinds of heat and power resources. Balance calculations of consumption and payment of resources.
2. Holding the database for keeping the facts about consumption, payment ant other theme departments of statistic information.
3. Calculation of power saving measures and the establishments of their effectives.
4. Forecasting of  planned consumption indexes with the calculation of  normative indexes and forecasting indexes of external factors.
5. Support of decision making in limit nominations for departments.
6. Balance calculations of planned indexes of heat and power consumption.
7. Integration of information of all indexes, statistical analysis in order to find dependences, regularities and tendencies and so on.
8. Making up and development of models for the idea of organizations unity as an object of management.
9. Revision of the first information given by the departments with the external information.
10. Picking out (on the basis of calculating methods working on the statistical data) the “candidates” of organizations for the immediate holding of complete power audit.
11. Fulfillment of the complex of typical and arbitrary tasks of giving up the information of IAS to the departments as well as the presentation of the information with the help of schemes, graph’s and diagrams; preparing of the ordinary forms of an account as well as new ones. Documentation of all the results of the IAS information.
12. Connection with external information systems integration into the united automatic system Minenergy.
13. Keeping and giving of the normative documentation.
14. Publishing and control over the fulfillment of the management instructions.

5. The review of existing forecasting methods

Forecasting process consists of several steps sequences of which is not always simple. But nevertheless one logic line may be found there:
- choosing of factors and forecasting parameter;
- collection of data;
- advanced data processing;
- placement of visual data;
- choosing of forecasting model;
- choosing of adequate methods of evaluation of forecasting model parameters;
- making of the models;
- evaluation of adequacy of the made models;
- choosing of the best model;
- forecasting process;
- monitoring of data and adaptation of the model with the account of new data.
So make adequate models and competent forecasting it is necessary to have:
- theoretical base;
- working experience in statistical program ensuring;
- experience of making models and forecasting.
Each task should be done by several ways. Forecasting is not an exception of this rule. While making a forecast the analyst may take different steps and use in practice the most effective one. For the examination of forecasting qualities of the model a cross-examination is used. Model is built on the excerption of a cut “tail” and then the “tail” and forecast are compared. Besides it can be examined if the forecast is stable. It is done by means of removing of some observations from the final excerption and finally even if the quality of forecasting model is good the process of monitoring should not be neglected.

Holt’s and Brown’s methods

In the middle of last century Holt offered an improved method of exponent smoothing out which was named after its creator. In this algorithm the indexes of level and trend are smoothed out by means of exponent smoothing out. The smoothing parameters are different.
The first equation shows the smoothed out row of general level. The second equation serves for the trend evaluation. The third equation determines the forecast forward. Constants of smoothing out in Holt’s method play the same role as a constant in a simple exponent smoothing out. They are chosen by means of running over on these parameters with any step. More simple equations (with less calculations) can be also used. The thing is that the parameters giving the exactness to the model can be always found in tests and then used in a read forecasting. Particular case of Holt’s method is Brown’s method when  a=?.

The method of Winters

Although method of Holt given below is not very simple, it can not take into consideration the season vibrations while forecasting. Speaking more exactly this method can not “see” them in prehistory. There is an expansion of Holt’s method to the 3-parameters exponent smoothing out. This algorithm is called the method of Winters. An attempt is made to take into account the season components of data. The system of equation describing Winters method is the following:
fraction of the first equation serves for the removing of seasons from Y(t). After this algorithm works with “pure” data where there are not season vibrations. They appear in the final forecast when the “pure” forecast is calculated almost by Holt’s method and is multiplied by season coefficient.

Box – Jenkins methods

In the middle of goes of last century a new and rather powerful class of algorithms for forecasting of temporal rows was developed.
The largest part of methodology receach and models checking was made by 2 statysts: C.E.P. Box and C.M. Jenkins. Since that time making of such models and forecasting made on their basic are called methods of Box-Jenkins. We shall see the hierarchy of algorithms in detail later and now let us say that this family consists of many algorithms, the most known and used of them is ARIMA. It is placed in almost every specialized package for forecasting. In classic variant ARIMA does not use free variables. The models are based on the information of forecasting rows and this limits the algorithm possibilities. At present time scientific literature contains the variants of ARIMA models which allow to take into account free variables. We shall not see them but the classic variant only. ARIMA does not suppose any exact model for forecasting of given temporal series. The common class of models is given which describes the temporal row and allows to show the current index of variable by means of its previous indexes. Then algorithm chooses the most suitable model of forecasting. As it was already said below, there is a hierarchy of models of Box-Jenkins. It should be logically determined as the following:

AR(p)+MA(q)->ARMA(p,q)->ARMA(p,q)(P,Q)->ARIMA(p,q,r)(P,Q,R)->...

6. The list of non-solved problems

The enterprise has the following tasks according to the power resources consumption policy:

- to lessen human factor in the process of placing of the information into database of analytical system;
- to lessen human factor in the process of information process;
- to broaden the information level of management and staff about current state of power resources consumption;
- to make documents work of staff easier (processing, analysis making up and others);
- to provide the reliable forecast of power resources consumption in the future on the basis of different forecasting and analysis methods.

7. Conclusion

The variant of creating an individual system on the basis of the ready neuropackage is suitable for small companies and private persons – investors, traders and owners. Though there are some examples of large groups which have chosen this variant and reached success. For example, company DuPont made new material – safety glass, having used in such large western banks as Citibank, Security Pacific National Bank, The World Bank, Lloyds Bank, The Federal Reserve Board, Federal Reserve Bank of New York; in insurance companies – Royal Insurance, Presidential Life Insurance, New York Life Insurance and others.
 It is company is large there is an importance of developing of its own program product which can increase the activities of the company. In the given product there is sense to realize analytical system together with account and others subsystems.

8. The list of references

М.: Институт энергодиалога “Восток–Запад”, 2003. 128 с. "Долгосрочное прогнозирование газового рынка"

Журнал "Обзор энергетика"

"Комплексная автоматизация энергоучета на промышленных предприятиях и хозяйственных объектах"

Вести, Нац. техн. ун-та "ХПИ", выпуск №12, 2002г., с.299

"Нейросетевая модель прогнозирования потребления газа в жилищно-бытовом секторе"

Вести, Харьк. гос. политехн. ун-та, выпуск 113, с.363

"Нейросетевая модель связного потребления тепловой и электрической энергии крупным жилым массивом города"

Новосибирский государственный технический университет "Практическая реализация нечёткой нейронной сети при краткосрочном прогнозировании электрической нагрузки"

Х.: Изд-во «Харьков», 2002.— 240 с.

"Синтез прогностической модели связного потребления электричества и тепла на базе искусственной нейронной сети"

ЭНЕРГОПРОГРЕСС - ЖУРНАЛ "ЭНЕРГЕТИК", 2004. № 6

"Учет влияния метеорологических факторов при прогнозировании электропотребления энергообъединений"

ЭНЕРГОПРОГРЕСС -- ЖУРНАЛ "ЭЛЕКТРИЧЕСКИЕ СТАНЦИИ ", 2004. № 6

"Информационная структура и программные средства обработки и хранения данных технологического оборудования и режимных параметров"

Россия, Москва, Министерство образования РФ,
Московский энергетический институт (технический университет)

"Учёт и контроль потребления топливно-энергетических ресурсов минобразования России"