A nonlinear mixed effects model for the prediction of natural gas consumption by individual customers

Abstract This study deals with the description and prediction of the daily consumption of natural gas at the level of individual customers. Unlike traditional group averaging approaches, we are faced with the irregularities of individual consumption series posed by inter-individual heterogeneity, including zeros, missing data, and abrupt consumption pattern changes. Our model is of the nonlinear regression type, with individual customer-specific parameters that, nevertheless, have a common distribution corresponding to the nonlinear mixed effects model framework. It is advantageous to build the model conditionally. The first condition, whether a particular customer has consumed or not, is modeled as a consumption status in an individual fashion. The prediction performance of the proposed model is demonstrated using a real dataset of 62 individual customers, and compared with two more traditional approaches: ARIMAX and ARX.

[1]  F. Diebold,et al.  Comparing Predictive Accuracy , 1994, Business Cycles.

[2]  Gwilym M. Jenkins,et al.  Time series analysis, forecasting and control , 1972 .

[3]  Remy Cottet,et al.  Bayesian Modeling and Forecasting of Intraday Electricity Load , 2003 .

[4]  Antoni Espasa,et al.  Forecasting from one day to one week ahead for the Spanish system operator , 2007 .

[5]  José Ramón Cancelo,et al.  Forecasting the electricity load from one day to one week ahead for the Spanish system operator , 2008 .

[6]  Salvador Gil,et al.  Generalized Model of Prediction of Natural Gas Consumption , 2004 .

[7]  R. Wolfinger,et al.  Generalized linear mixed models a pseudo-likelihood approach , 1993 .

[8]  Kung-Sik Chan,et al.  Time Series Analysis: With Applications in R , 2010 .

[9]  E. F. Sánchez-Úbeda,et al.  Modeling and forecasting industrial end-use natural gas consumption☆ , 2007 .

[10]  A. Harvey,et al.  Forecasting Hourly Electricity Demand Using Time-Varying Splines , 1993 .

[11]  M. Medeiros,et al.  Modeling and forecasting short-term electricity load: A comparison of methods with an application to Brazilian data , 2008 .

[12]  Reinaldo Castro Souza,et al.  A smooth transition periodic autoregressive (STPAR) model for short-term load forecasting , 2008 .

[13]  Vitor Hugo Ferreira,et al.  Input space to neural network based load forecasters , 2008 .

[14]  Marek Brabec,et al.  A statistical model for the estimation of natural gas consumption , 2008 .

[15]  Jeffrey Bailey.  Load Profiling for Retail Choice: Examining a Complex and Crucial Component of Settlement , 2000 .

[16]  Denzil G. Fiebig,et al.  Regional End‐Use Gas Demand in Australia , 1996 .

[17]  S. R. Searle,et al.  Generalized, Linear, and Mixed Models , 2005 .

[18]  G. Jenkins,et al.  Time series analysis, forecasting and control , 1971 .

[19]  D. Bates,et al.  Mixed-Effects Models in S and S-PLUS , 2001 .

[20]  Nonlinear Models for Repeated Measurement Data , 1996 .

[21]  James W. Taylor.  An evaluation of methods for very short-term load forecasting using minute-by-minute British data , 2008 .

[22]  Rong Chen,et al.  A semi-parametric time series approach in modeling hourly electricity loads , 2006 .

[23]  F. Gubina,et al.  Determining the load profiles of consumers based on fuzzy logic and probability neural networks , 2004 .

[24]  S. R. Searle.  Linear Models , 1971 .

[25]  N. Breslow,et al.  Approximate inference in generalized linear mixed models , 1993 .

[26]  Johan A. K. Suykens,et al.  Modelling the Belgian gas consumption using neural networks , 1996, Neural Processing Letters.

[27]  Jayanta K. Ghosh,et al.  An Introduction to Bayesian Analysis , 2006 .

[28]  S. Koopman,et al.  An Hourly Periodic State Space Model for Modelling French National Electricity Load , 2007 .

[29]  D. Rubin,et al.  Statistical Analysis with Missing Data. , 1989 .

[30]  Mark F. J. Steel,et al.  Estimating end-use demand: A Bayesian approach , 1994 .

[31]  William N. Venables,et al.  Modern Applied Statistics with S-Plus. , 1996 .