深层卤水可采资源量评价的时间序列神经网络模型研究

深层地下卤水资源量的评价是国内外迄今尚未很好解决的课题,由于深部地质及水文地质参数难以准确获取,因此也难以对深层卤水资源量进行正确评价.从深层卤水开采量的时间序列出发,提出了一类基于神经网络的评价模型.首先分析了卤水开采量的时间序列特点,再建立了神经网络的拓扑结构,并设计了相应的评价算法,最后通过对某储卤构造的单井评价实例,对模型进行了验证.与传统的ARMA时间序列模型相比,其预测性能更好,对剩余可采资源量计算结果也更为准确.

Research on Time Series Neural Network Model of Recoverable Deep Brine Resources Evaluation

Evaluation of the deep brine resources is an issue still unresolved worldwide.As it is difficult to obtain accurate deep geologic and hydrogeologic parameters,it is not easy to correctly evaluate the deep reserves of brine.Starting from the time series of deep brine production,this paper proposes an evaluation model based on neural network.The paper first analyzes the time series features of the brine production,and then establishes the topology of neural network and designs the evaluation algorithms.Finally,a single well of a construction of reservoir brine is evaluated,and the evaluation mode is verified.The predicted performance of neural network is better than that of the traditional ARMA time series model,and the calculation results of remaining recoverable resources are also more accurate.