基本電費繳交的多寡，主要是依據所簽訂的契約容量。契約容量訂太高，若實際用電量遠低於契約容量，則會平白多繳納未使用到之電費；訂太低，若瞬間最高用電量超出契約容量，則需繳交高額的超約罰款，因此契約容量訂的恰不恰當，關係到電費支出的高低。目前大多數用戶仍是依歷史用電資料求解出最適契約容量，再將此契約容量套用至來年使用，但過往用電特性未必符合未來用電之趨勢，因此發展預測模式預測未來之最適契約容量有其必要性。目前研究最適契約容量的方法，各有不同，各有其所長。本研究將以統計檢定及迴歸分析，預測出未來夏季及非夏季最高用電量之機率分配，據此建構符合檢定結果之用電成本數學模型，在依數值積分求解出最適經常及非夏月契約容量，以達全年總期望用電成本最小化之目標。經個案研究證實，若採用本研究所提之預測模式可有效幫助企業降低電費7.35%。

Demand charge depends on the contract capacity that, was written on the contract between Taiwan Power Company and enterprises. If it is set too high, the electricity user will pay higher demand charge. If it is set too low, the maximum electricity capacity will exceed the contract capacity, which leads to penalty charges. Therefore, the setting of suitable contract capacity is quite important for enterprises. In the past, most enterprises found their optimal contract capacity by calculating from their history records, but it may not reflect the actual future electricity use. Therefore, it is necessary to develop predictive models to predict the optimal contract capacity of future. For so many years, great multitude of studies have been conducted to find the optimal contract capacity. In this study, statistical tests and regression analysis were used to predict the probability distribution of future electricity use in summer and non-summer peak period. According to test results to construct electricity costs mathematical model. Then, the numerical integration was employed to find the optimal contract capacity of the summer and non-summer peak period. Therefore, the electricity costs are the cheapest all year. Based on the case in this study, their electricity costs dropped by 7.35% with the help of the mathematic models we proved.