Research On Prediction Model Of The Potential Of Energy Saving And Emission Reduction In Textile And Apparel Enterprises

Textile and apparel industry consume a large amount of energy,water resources,and produces a large number of greenhouse gases,waste gases,wastewater and other pollutants.The 13 th Five-Year Development Plan of Textile Industry points out that energy saving and emission reduction of textile industry should be promoted from a new height of building ecological civilization,and low-carbon,green and circular textile economy should be developed to promote the transformation and upgrading of the industry.At present,a large amount of data in China's textile and apparel industry only stay on the statistical function or energy efficiency calculation,lack of deep data mining,and fail to transfer the data into information that managers need to play an important role in energy conservation and emission reduction.By using data mining technology and historical environmental pollution data,the prediction model was established,and the potential of energy saving and emission reduction was predicted.This paper studied the potential of energy saving and emission reduction,defined its connotation and calculation formula through literature research method,divided the potential of energy saving and emission reduction into three quantitative indicators: energy consumption,emissions and product output.Four single prediction models included ARIMA model,VAR model,GM(1,1)model and Neural Network model,were established by screening the influencing factors through different correlation analysis.Combing model was established by the two single model which had small prediction error,and conclusion was proved by empirical experiments.Finally,the potential of energy saving and emission reduction was calculated and analyzed.The main conclusions of this paper include below:(1)The quantitative indicators of the potential of energy saving and emission reduction were given.The conceptual framework was defined,and the energy conservation or energy saving rate per unit product output,and emission reduction or emission reduction rate per unit product output were taken as quantitative indicators.The prediction of the potential of energy saving and emission reduction was transformed into the prediction of total energy consumption,greenhouse gas emissions(GHG)or pollutant emissions and product output.(2)The influencing factors of the potential of energy saving and emission reduction were studied.Through literature review,enterprise research and data mining method,it was found that the main influencing factors of energy consumption come from different types of energy and its energy consumption values.The main factors affecting GHG emissions are the use of electricity,steam,natural gas and their GHG emissions.The main influencing factors of water degradation footprint(WDF)were pollutant concentration(such as COD,AN,etc.),sewage discharge,pollutant discharge and WDF of pollutants.The main influencing factors of product output were consumption cost of various types of energy,sewage treatment cost and financial indicators.(3)The prediction model of the potential of energy saving and emission reduction were established.By comparing the prediction accuracy to evaluate the prediction model,it was found that for GHG/pollutant emissions,energy consumption and product output,the prediction error of Neural Network model was the smallest for the single model;for the combined model,the prediction accuracy of VAR model modified by Neural Network was the best,followed by the Neural Network model based on Grey theory,and finally the Grey Prediction model modified by Neural Network(4)The method of the prediction model of the potential of energy saving and emission reduction were applied.Through the application,it was found that the disadvantage of the VAR-Neural Network combination model was that the collinear variables cannot coexist in the same model which resulting in less variables and increasing errors.Therefore,when variables are too few,combination model which called Neural Network model based on Grey theory is recommended;and single model which named Neural Network prediction model is suggested for which can input all variables,obtain better prediction accuracy by self-learning,and basically control the average absolute error below 1%.The main innovations of this paper are as follows:(1)The potential of energy saving and emission reduction of textile and apparel enterprises was defined,and characterized by energy-saving rate per unit product output and emission-reduction rate per unit product output.The calculation formula was given,and the influencing factors was initially determined.(2)The main influencing factors were extracted and screened through the interdisciplinary research method which based on the data mining method.On this basis,the prediction model of the potential of energy saving and emission reduction of textile and apparel enterprises was constructed by Time Series,Neural Network and Grey Prediction method.(3)The application of the prediction model was carried out according to the three-level measurement data of field investigation,and the potential of energy saving and emission reduction of textile and apparel enterprises was analyzed,which improved statistics and utilization of energy saving and emission reduction data.