12Advanced sensor and communication technologies can make natural gas supply systems 13 smarter than ever before, in both system management and operation. This paper presents the 14 development of a novel data-driven Demand Side Management, whose framework includes demand 15 forecasting, customer response analysis, prediction of dynamic condition of the gas network, quick 16 supply reliability evaluation, multi-objective optimization and decision-making. The aims of this 17 DSM method are to smooth load profiles, improve company profit and enhance system reliability, 18 by means of a dynamic pricing strategy. To verify the effectiveness of the developed framework, a 19 case study is considered, concerning the management of a relatively complex gas supply system, 20wherein four different pricing periods are introduced for comprehensively testing. The results in the 21 case study show that the DSM framework is able to effectively achieve the targets of peak shaving 22 and valley filling. Besides, it can significantly and stably improve the system efficiency and 23 reliability, for different pricing periods. Finally, pricing period determination is discussed in relation 24 to the features of performance. 25
Multi-objective optimization 27Nomenclature CPP Critical Peak Pricing RR risk reduction DR Demand Response SR risk of natural gas shortage DSM Demand Side Management Supply Capacitytotal,max s optimal F representative solution in family s vi,t, ki,t positive constant factor G objective value cumulative standard Gaussian distribution all the permanent efficiency improvement methods [7], for example, equipment replacement and 58 system update. DR methods can control the patterns of the customers' load [8]-[10]. Furtherly, 59 according to the way to influence the customers, the DR methods can be classified as Incentive-60 based DR and Time-based DR. The Incentive-based DR are triggered by specific situations [6], [11]: 61 for example, special contracts for some specific customers with limited sheds, voluntary behaviors 62 to emergency signals [12], customers bidding for curtailing at reasonable prices, etc. The Time-63 based DR methods are mainly performed by changing the price of gas in time to desired demand. 64 According to the rules to adjust the price, the Time-based DR methods can be grouped by Time-of-65 Use (TOU) method [13], Critical Peak Pricing (CPP) method [14] and Real-Time Pricing (RTP) 66 method [15]. From the literature review, it emerges that the Time-based DR methods are the most 67 effective DSM strategies, because their inherent characteristics are more suitable to the real-world 68 130 Accurate gas demand forecasting is essential to the effectiveness of the DSM strategy. The 131 researches of energy demand forecasting have explored different kinds of methods, including time 132 series model [5], [37], regression model [38], [39] and artificial neural network [40]-[43]. Recently, 133 some hybrid forecasting methods, integrating different models to overcome the relevant problems 134 of the diffe...
