为了提高黄金价格预测精度，提出等维动态马尔可夫\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}预测模型，引入取新去旧的数据处理方法，使用等维动态实现数据优化。等维动态马尔可夫\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}预测模型是将等维动态\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}模型与马尔可夫链结合起来，在等维动态\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}模型的预测结果上再进行状态划分与转移，重新得到预测值。选取2018年1月~2019年4月共16组黄金价格数据，将动态等维的维数定为13，数据处理时选用2018年1月~2019年1月的13个黄金价格数据，预测2019年2月的黄金价格，再依次预测2019年3月和4月的黄金价格。以2019年2~4月的3个黄金价格预测数据作为拟合，预测2019年5月的黄金价格。通过比较灰色\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}预测模型、等维动态\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}模型与等维动态马尔可夫\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}预测模型的精度，可知等维动态\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}预测模型的精度较\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}模型有所提高，等维动态马尔可夫\mathrm{S}\mathrm{C}\mathrm{G}\mathrm{M}（\mathrm{1}，\mathrm{1}{）}_{\mathrm{C}}模型的拟合精度最高，达到一级，相对误差平均值为0.85%，符合预测要求，应用该模型预测的2019年5月的黄金价格为1 314.78美元/盎司，实际黄金价格为1 295.55美元/盎司，价格较为接近。

In order to improve the accuracy of gold price prediction，an equal dimensional dynamic Markov SCGM（1，1）_C forecasting model was proposed.Prediction has high requirements for the selection of data，and the latest data can improve the prediction accuracy.The equal dimensional dynamic Markov SCGM（1，1）_C model is a composite model which combines the equal dimensional dynamic SCGM（1，1）_C model with the Markov chain.On the basis of the prediction results of the equal dimensional dynamic SCGM（1，1）_C，the grey fitting accuracy index is divided into states，and the state of the monthly gold price is determined.On this basis，the next transition direction is determined according to the transition probability matrix，and finally the predicted data is obtained.In this paper，the data processing method of take the new one and remove the old one was introduced，and the equal dimension dynamic data optimization was used.Because the grey SCGM（1，1）_Cprediction model is also a grey model，the grey model is characterized by less original data，so a large number of original values are not needed in this paper.A total of 16 groups of gold price data from January 2018 to April 2019 were selected，and the dimension of dynamic equal dimension was determined to be 13.When SCGM（1，1）_Cmodel data were processed，13 gold price data from January 2018 to January 2019 were selected to predict the gold price in February 2019，and then the gold price of March 2019 and April 2019 was predicted as above.The prediction data from February 2019 to April 2019 were used as fitting data to observe whether the accuracy of the prediction model is the best.The grey SCGM（1，1）_Cmodel was predicted directly with all 16 known data.By comparing the grey SCGM（1，1）_Cprediction model，the equal dimensional dynamic SCGM（1，1）_Cmodel and the equal dimensional dynamic Markov SCGM（1，1）_Cprediction model it is know that the accuracy of the equal dimensional dynamic SCGM（1，1）_Cmodel is higher than the SCGM（1，1）_Cmodel.The fitting accuracy of the equal dimensional dynamic Markov SCGM（1，1）_Cis the highest，reaching the first order，the average relative error is 0.85%，which meets the prediction requirements，and the gold price in May 2019 is predicted to be $1 314.78.Although the grey SCGM（1，1）_Cmodel has the lowest accuracy，it is simple to calculate and all the predicted values can be obtained by one calculation.The equal dimensional dynamic Markov SCGM（1，1）_Cmodel is the most complex，but its predict results are the most accurate.Compared with the neural network and other methods，the equal dimensional dynamic Markov SCGM（1，1）_Cmodel is simpler，so the model can be used to predict the gold price.The gold price in May 2019 is $1 295.55.Which Contrast with the predict is very close.