Метод прогнозування характеристик точності вимірювальних каналів енергетичних систем

Анотація

УДК 621.391

FORECASTING METHOD OF CHARACTERISTICS ACCURACY OF MEASURING CHANNEL POWER

E.A. Reutskyi

Modern energy systems are complex hardware-software complexes, consisting of a set of measuring channels, computer components, communication lines, means registration and display. The work of such systems in the management process is necessary to study the characteristics of individual components to establish and maintain reliable operation.

When converting information into energy system components and end user arrives to question its authenticity, which affect measurement errors, errors in data transmission, imperfect lines and so on. Given that the main factor that affects the accuracy of the information data is the primary measuring converters, urgent and important task is to develop methods of forecasting and control characteristics of precision measuring channels.

The purpose of research - the study forecasting method precision characteristics of measuring channels of energy systems using time series, building a prediction model based testing.

Materials and methods of research. In the analysis we assume that itself contains a number of deterministic and random component. This part can be determined part of the following components: a trend that defines the main trend of the time series; cycles are fluctuations relative to the trend; seasonal components that express periodic fluctuations.

As a forecasting model chosen autoregression model slick average (ARIMA), which is effective at predicting stationary and non-stationary random processes and relatively easy to use.

This model differs iterative approach to the formation predictive model, ie, all its parameters are selected from a plurality of possible thus chosen a model that most accurately describes this time series with minimal dispersion.

To predict the characteristics of the selected accuracy relative error of measuring channel based on test results obtained with the following assumptions:

1. The value of relative error has a normal distribution.

2. Absolute relative error ranged from 0.5 to 0.6.

3. Evaluation of the random variable that will be modeled and prognosis should be effective and appropriate.

The task of forecasting first reduced to a maximum definition and selection each deterministic component time series that is made in particular through the model ARIMA. Their presence, influence and share individual component parts is determined individually for each process, expressed time series.

As a pseudo-random number generator will use true random number generator. Since the process changes the characteristics of precision measuring channels Technical Systems is an unpredictable process, as the model selected random aperiodic process.

Results. As a result, generated using random number generator real time series, based on an analysis which was based on the requirements for accuracy and forecast volume samples using a number

Upon receipt and transmission channels through measuring energy systems held its loss and distortion caused by measurement errors that must be considered. The work was based method of forecasting performance precision measuring channels using time series models autoregression moving average ARIMA, used real random numbers generator. Based on the research the amount of data required parameters of the model and the accuracy of the forecast.

In further studies need to consider ways to improve the prognosis and the possibility of using other alternative methods to be used in practice, which can improve the accuracy in the production process and a comparative analysis of the proposed method of work.