题目:
作者:
期刊会议:
年份:
论文地址:
代码:
keyword:Euler equations; Machine learning; Neural networks; Conservation laws; Riemann problem; Hidden fluid mechanics
The conservation of mass, momentum and energy for compressible flow in the inviscid limit can be modeled by the Euler equations
∂ t U + ∇ ⋅ f ( U ) = 0 , x ∈ Ω ⊂ R d , d = 1 , 2 , t ∈ ( 0 , T ] \partial_{t} U+\nabla \cdot f(U)=0, x \in \Omega \subset \mathbb{R}^{d}, d=1,2, t \in(0, T] ∂tU+∇⋅f(U)=0,x∈Ω⊂Rd,d=1,2,t∈(0,T]
对于一维
U = ( ρ ρ u ρ E ) , f ( U ) = ( ρ u ρ u 2 + p u ( ρ E + p ) ) U=\left(\begin{array}{c} \rho \\ \rho u \\ \rho E \end{array}\right), \quad f(U)=\left(\begin{array}{c} \rho u \\ \rho u^{2}+p \\ u(\rho E+p) \end{array}\right) U=⎝⎛ρρuρE⎠⎞,f(U)=⎝⎛ρuρu2+pu(ρE+p)⎠⎞
对于二维
U = ( ρ ρ u 1 ρ u 2 ρ E ) , f = ( G 1 , G 2 ) , with G i ( U ) = ( ρ u i δ i 1 p + ρ u 1 u i δ i 2 p + ρ u 2 u i p u i + ρ u i E ) , i = 1 , 2 U=\left(\begin{array}{c} \rho \\ \rho u_{1} \\ \rho u_{2} \\ \rho E \end{array}\right), \quad f=\left(G_{1}, G_{2}\right), \text { with } G_{i}(U)=\left(\begin{array}{c} \rho u_{i} \\ \delta_{i 1} p+\rho u_{1} u_{i} \\ \delta_{i 2} p+\rho u_{2} u_{i} \\ p u_{i}+\rho u_{i} E \end{array}\right), i=1,2 U=⎝⎜⎜⎛ρρu1ρu2ρE⎠⎟⎟⎞,f=(G1,G2), with Gi(U)=⎝⎜⎜⎛ρuiδi1p+ρu1uiδi2p+ρu2uipui+ρuiE⎠⎟⎟⎞,i=1,2
其中, ρ \rho ρ is the density, p p p is the pressure, u u u is the velocity in one dimension or u 1 u_{1} u1, u 2 u_{2} u2 are velocity components in x x x
and y y y directions in two dimensions, and E E E is the total energy, δ i j \delta_{i j} δij is the Kronecker delta. 为了结束这些方程,我们还需要一个方程,即描述压强和能量关系的状态方程。本文考虑了在给定b条件下多变性气体的状态方程
p = ( γ − 1 ) ( ρ E − 1 2 ρ ∥ u ∥ 2 ) p=(\gamma-1)\left(\rho E-\frac{1}{2} \rho\|\boldsymbol{u}\|^{2}\right) p=(γ−1)(ρE−21ρ∥u∥2)
不需要知道不连续点的确切位置,我们只需要对不连续点的区域进行一个粗略的估计,并在不连续点周围使用更多分散的数据。
二维问题的速度和压力的准确性比一维情况下表1中,因为速度和压力在一维情况下平稳而速度和压力在二维情况下不连续
创新: