随着全球石油和天然气行业的快速发展和需求增长，出砂问题对设备和管道的影响显著，导致生产效率下降和安全风险增加。为准确预测出砂量，降低油气开采风险和成本，提出一种基于灰狼优化算法（GWO）和人工神经网络（ANN）结合的出砂量预测模型。针对传统模型误差较大的问题，提出的GWO-ANN出砂量预测模型通过灰狼优化算法优化神经网络的权重和偏差，可提高模型的预测精度和鲁棒性。在实验设计部分，通过振动传感器采集气-砂两相流的出砂信号，并利用希尔伯特-黄变换（HHT）分析出砂信号的频带特征，用有限冲激响应（FIR）滤波器对噪声进行滤除。使用主成分分析（PCA）方法减少信号特征的复杂度，将主要特征输入GWO-ANN模型进行训练和预测。实验结果显示，GWO-ANN模型在测试集上最大相对误差较小，表明GWO-ANN模型能够有效地监测出砂量，具有较高的准确性和可靠性。

With the rapid development and increasing demand in the global oil and gas industry, sand production issues have a significant impact on equipment and pipelines, leading to reduced production efficiency and increased safety risks. To accurately predict sand production volume and reduce the risks and costs associated with oil and gas extraction, a sand production volume prediction model combining the grey wolf optimizer(GWO) and artificial neural network(ANN) is proposed. Addressing the problem of large errors in traditional models, the proposed GWO-ANN sand production volume prediction model optimizes the weights and biases of the neural network using the grey wolf optimizer, which can improve the prediction accuracy and robustness of the model. In the experimental design section, sand production signals from gassand two-phase flow were collected using vibration sensors, and the Hilbert-Huang transform(HHT) was used to analyze the frequency band characteristics of the sand production signals. A finite impulse response(FIR)filter was employed to remove noise. Principal component analysis(PCA) was used to reduce the complexity of signal features, and the main features were input into the GWO-ANN model for training and prediction.Experimental results show that the GWO-ANN model has a small maximum relative error on the test set,indicating that the GWO-ANN model can effectively monitor sand production volume with high accuracy and reliability.