Fuzzy time series forecasting (FTSF) is a typical forecasting method with
wide application. Traditional FTSF is regarded as an expert system which leads
to lose the ability to recognize undefined feature. The mentioned is main
reason of poor forecasting with FTSF. To solve the problem, the proposed model
Differential Fuzzy Convolutional Neural Network (DFCNN) utilizes convolution
neural network to re-implement FTSF with learnable ability. DFCNN is capable of
recognizing the potential information and improve the forecasting accuracy.
Thanks to learnable ability of neural network, length of fuzzy rules
established in FTSF is expended to arbitrary length which expert is not able to
be handle by expert system. At the same time, FTSF usually cannot achieve
satisfactory performance of non-stationary time series due to trend of
non-stationary time series. The trend of non-stationary time series causes the
fuzzy set established by FTSF to invalid and cause the forecasting to fail.
DFCNN utilizes the Difference algorithm to weaken the non-stationarity of time
series, so that DFCNN can forecast the non-stationary time series with low
error that FTSF cannot forecast in satisfactory performance. After mass of
experiments, DFCNN has excellent prediction effect, which is ahead of the
existing FTSF and common time series forecasting algorithms. Finally, DFCNN
provides further ideas for improving FTSF and holds continued research value.

通过使用微分模糊卷积神经网络，本文提出的新模型具有学习能力和对非静态时间序列的预测准确性，其预测效果优于现有的模糊时间序列预测方法以及普通的时间序列预测算法，可进一步提高模糊时间序列预测的准确率。