利用灰色GM(1,N)模型預測灌溉渠道微型水力發電量

Abstract

Grey systems theory was proposed by Dr. Julong Deng in 1982 for relational analysis and model construction in systems with uncertainties or few data. The method relies on prediction and decision making mechanisms to understand the entire system. There are many uncertain factors that have impact on the output of a micro hydropower system, and hence such a system can be considered as a dynamic grey one. Grey prediction is therefore a proper tool to forecast the electric output of micro hydropower systems in irrigation channels. This research employs GM(1, N) method with N = 1, 2, 3 to build rolling models realized by MATLAB for this purpose, and the experimental results are evaluated by the mean absolute percentage error (MAPE). The best settings and their corresponding results for the various value of N in GM(1, N) are as follows: optimal modeling number = 7 and MAPE = 12.444% for GM(1, 1), optimal modeling number = 60 and MAPE = 9.734% for GM(1, 2) including rotational speed, optimal modeling number = 60 and MAPE =9.676% for GM(1, 2) including water flow, and optimal modeling number = 60 and MAPE = 11.003% for GM(1, 3). Since the resulting MAPE values are all less than 20%, the prediction result of every model is either moderate or highly accurate. The findings of this study can provide useful recommendations for setting macro hydropower systems in irrigation channels.