利用人工神经网络储层预测技术，进行隐蔽油气藏预测，有成功的，也有失败的。预测中，如何正确进行时窗有效参数选择、样本建立、样本检验以及作图参数选择，将直接关系到储层预测的成败。我们利用人工神经网络储层预测技术，在HG地区进行隐蔽油气藏预测时，针对隐蔽油气藏的特点，探索出了一套在精细的储层标定及层位解释后，沿层进行时窗选择、样本建立、有效参数提取、样本检验及作图参数选择的工作方法，并取得了可喜的地质效果。 Using artificial neural network reservoir prediction technology to make subtle reservoir prediction has both success and failure. In the prediction, how to correctly select the effective parameters of the time window, sample establishment, sample test and selection of the mapping parameters will directly affect the success or failure of reservoir prediction. We use artificial neural network reservoir prediction technology to detect subtle reservoirs in HG area. Based on the characteristics of subtle reservoirs, we have explored a set of methods for time window selection along the layers after fine reservoir calibration and horizon interpretation , Sample establishment, effective parameter extraction, sample inspection and selection of mapping parameters, and achieved gratifying geological results.