在线符号有向图推理机

符号有向图(Signed Directed Graph,SDG)是一种基于模型方法的故障识别与诊断技术,是一种定性表示因果关系的图论方法。符号有向图(SDG)是定性研究化工过程中变量间关系的有力工具,利用SDG的正向反向推理机制可以快速发掘出系统中变量之间影响的因果关系,不仅可以用于在线的故障诊断,还可以用于离线的危险和可操作性(HAZOP)分析。

水箱体系如下图所示。

对于的SDG模型及推理运行的结果如下：

import sdg
digraph_content=”’digraph buffer_tank {
node [style=filled];
V1->F1[style=solid];
F1->L[style=solid];
L->V3[style=solid];
V2->F2[style=solid];
V3->F3[style=solid];
F2->L[style=dashed];
F2->V2[style=dashed];
F3->L[style=dashed];}
”’
set_nodes_init_state={‘V1′:1}

steps=2
sdg.sdg_run_forward(steps,digraph_content,set_nodes_init_state)

img = pylab.imread(‘digraph.png’)
pylab.imshow(img)
pylab.show()

推理结果

Nodes list:
['F1', 'F2', 'F3', 'L', 'V1', 'V2', 'V3']
Adjacency matrix:
[[ 0  0  0  1  0  0  0]
 [ 0  0  0 -1  0 -1  0]
 [ 0  0  0 -1  0  0  0]
 [ 0  0  0  0  0  0  1]
 [ 1  0  0  0  0  0  0]
 [ 0  1  0  0  0  0  0]
 [ 0  0  1  0  0  0  0]]

连续搅拌全混釜（CSTR）反应器如下图所示。

对于CSTR反应器模型，这是一个稍复杂一些的体系。

import sdg

digraph_content=”’digraph CSTR{
node [style=filled];
size=”8,8″;
F->V[style=dashed];
V->F[style=solid];
V->CA[style=dashed];
V->T[style=solid];
CA->CA[style=dashed];
CA->rA[style=solid];
CA0->CA[style=solid];
F0->V[style=solid];
F0->CA[style=solid];
F0->T[style=dashed];
T->T[style=dashed];
T->rA[style=solid];
T->Tc[style=solid];
rA->CA[style=dashed];
rA->T[style=solid];
T0->T[style=solid];
Fc->Tc[style=solid];
Tc->T[style=solid];
Tc->Tc[style=dashed];
Vj->Tc[style=dashed];
Tc0->Tc[style=solid];
Tj->Tc[style=dashed];
}
”’
init_state={‘F’:1}
steps=2
sdg.sdg_run_forward(steps,digraph_content,init_state)
img = pylab.imread(‘digraph.png’)
pylab.imshow(img)
pylab.show()

推理结果

Nodes list:
['CA', 'CA0', 'F', 'F0', 'Fc',..., 'Tj', 'V', 'Vj', 'rA']
Adjacency matrix:
[[-1  0  0  0  0  0  0  0  0  0  0  0  1]
 [ 1  0  0  0  0  0  0  0  0  0  0  0  0]
 [ 0  0  0  0  0  0  0  0  0  0 -1  0  0]
 [ 1  0  0  0  0 -1  0  0  0  0  1  0  0]
 [ 0  0  0  0  0  0  0  1  0  0  0  0  0]
 [ 0  0  0  0  0 -1  0  1  0  0  0  0  1]
 [ 0  0  0  0  0  1  0  0  0  0  0  0  0]
 [ 0  0  0  0  0  1  0 -1  0  0  0  0  0]
 [ 0  0  0  0  0  0  0  1  0  0  0  0  0]
 [ 0  0  0  0  0  0  0 -1  0  0  0  0  0]
 [-1  0  1  0  0  1  0  0  0  0  0  0  0]
 [ 0  0  0  0  0  0  0 -1  0  0  0  0  0]
 [-1  0  0  0  0  1  0  0  0  0  0  0  0]]

催化裂化反应器流程如下图所示。

对于催化裂化反应器模型，这是一个比较复杂的体系。

import sdg

#PCA-SDG FCCU
digraph_content=”’digraph FCCU{
node [style=filled];
size=”8,8″;
Qrgc->Qm[style=solid];
Qm->dTr[style=solid];
dTr->Tr[style=solid];
Patm->Fv12[style=dashed];
V12->Fv12[style=solid];
k12->Fv12[style=solid];
Fv12->dP5[style=dashed];
Fv11->dP5[style=dashed];
dP5->P5[style=solid];
P5->Fv12[style=solid];
P5->P4[style=solid];
Fv13->dP5[style=solid];
P4->Prb[style=solid];
Qout->dTr[style=dashed];
Qslurry->Qout[style=solid];
Qcracking->Qout[style=solid];
Qcatout->Qout[style=solid];
Qff->Qout[style=solid];
F1->Fcoke[style=solid];
F2->FB[style=solid];
FB->Fcoke[style=solid];
Psif->FB[style=solid];
WSHV->Fcoke[style=dashed];
Fcoke->dCsc[style=solid];
F3->Qcracking[style=solid];
F3->Qff[style=solid];
F3->Qm[style=solid];
F3->FB[style=solid];
F3->WSHV[style=solid];
F3->Fcoke[style=solid];
F3->rhoris[style=solid];
F3->Vris[style=solid];
F3->Fwg[style=solid];
F3->T2_ss[style=dashed];
Fwg->dP5[style=solid];
Tr->Fwg[style=solid];
Tr->Tsc[style=solid];
Tr->deltaHcrack[style=solid];
Tr->Qslurry[style=solid];
Tr->Qcatout[style=solid];
Tr->Qff[style=solid];
deltaHcrack->Qcracking[style=solid];
UAf->T2_ss[style=solid];
UAf->dT3[style=dashed];
T2_ss->dT2[style=solid];
F4->Qslurry[style=solid];
F4->Qcracking[style=solid];
F4->Fwg[style=solid];
F4->FB[style=solid];
F4->Fcoke[style=solid];
F4->WSHV[style=solid];
F4->rhoris[style=solid];
F4->Vris[style=solid];
taur->Fcoke[style=solid];
dWr->Wr[style=solid];
Wr->dCsc[style=dashed];
dCsc->Csc[style=solid];
Crgc->dCsc[style=solid];
Csc->dCsc[style=solid];
Fsc->dCsc[style=solid];
Fsc->dWr[style=dashed];
Frgc->dWr[style=solid];
Frgc->dCsc[style=dashed];
Frgc->rhoris[style=solid];
Frgc->Vris[style=solid];
Frgc->Wris[style=solid];
Frgc->Qcatout[style=solid];
Frgc->taur[style=dashed];
rhoris->Prb[style=solid];
Vris->rhoris[style=dashed];
Vris->Prb[style=solid];
Vris->Wris[style=solid];
Wris->WSHV[style=dashed];
Wris->taur[style=solid];
hris->Wris[style=solid];
dT2->T2[style=solid];
T2->dT2[style=dashed];
T2->Tlm[style=dashed];
T2->Qm[style=solid];
T1->T2_ss[style=solid];
T1->Tlm[style=solid];
Tlm->T2_ss[style=solid];
Tlm->dT3[style=dashed];
T3->Tlm[style=solid];
T3->Qloss[style=solid];
dT3->T3[style=solid];
Qloss->dT3[style=dashed];
F5->Qloss[style=solid];
F5->dT3[style=solid];
F5set->F5[style=solid];
}
”’

init_state={‘Qrgc’:1}

steps=2
sdg.sdg_run_forward(steps,digraph_content,init_state)
img = pylab.imread(‘digraph.png’)
pylab.imshow(img)
pylab.show()

推理结果

Nodes list:
['Crgc', 'Csc', 'F1', 'F2', 'F3', 'F4', 'F5',..., 'rhoris', 'taur']
Adjacency matrix:
[[0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 ..., 
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]
 [0 0 0 ..., 0 0 0]]