The problem of accounting data recovery on chemical enterprise
Relevance of the work is caused by the presence of missing data in the readings of energy meters. The main aim of the research is to study the choice of method for recovering missing data on energy consumption in industry. The methods used in the study: the models are calculated using the applicatio...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | Russian |
| Published: |
Tomsk Polytechnic University
2019-05-01
|
| Series: | Известия Томского политехнического университета: Инжиниринг георесурсов |
| Subjects: | |
| Online Access: | http://izvestiya-tpu.ru/archive/article/view/1372 |
| Summary: | Relevance of the work is caused by the presence of missing data in the readings of energy meters. The main aim of the research is to study the choice of method for recovering missing data on energy consumption in industry. The methods used in the study: the models are calculated using the application Curve Fitting Toolbox of the software complex "Matlab 7.0". The library of graphical models Curve Fitting Toolbox includes an application cftool, which allows defining a parametric model (such as, exponential function Exp, polynomial Polynomial, rational RAT, as well as the sum of sinusoidal functions SumSin), selecting parameters, analyzing approach suitability, displaying the result graphically. In the library of graphical models Curve Fitting Toolbox the models from more than 50 different mathematical functions are determined by search method. The results: The paper describes the features of simple and complex data recovery methods with further estimation of their errors and indicates the ways to improve the accuracy of n-factor models. The authors have studied direct and inverse dependences of recovering lost accounting data for a chemical enterprise. The optimal limits of initial research data samples are proved. The paper also provides options for defining the best methods for value recovery in cases of their absence. |
|---|---|
| ISSN: | 2500-1019 2413-1830 |