河道一维污染源控制方程的神经网络解法

A Neural Network Solution to the Control Equation of One-dimensional River Contamination Source

目的 河道污染物的分布模拟是一个求解偏微分方程的问题 ,研究用于求解一维污染源控制方程近似解析解的新方法 .方法 将神经网络和数理方程相结合 ,用三层前向神经网络逼近河道一维污染源控制方程 .结果与结论 得到了不同条件下的对流扩散方程的近似解析解 ,从而验证了河道一维污染源控制方程的神经网络解法的可行性和合理性 .

Aim The distribution simulation of river contamination is a problem of solving partial differential equation, a new method used for seeking approximate analytic solution of one-dimensional contamination source is studied. Methods Approximate analytic solution is solved by integration of neural network method with mathematical physics method and a three-layer feedforward neural network is utilized to approximate the control equation of one-dimensional contamination. Results and Conclusion The approximate analytic solutions are successfully obtained under three different conditions, the feasibility and rationality of applying neural network to river contamination simulation is hereby proved.