对于有误差的物理量进行运算的时候，学习了误差传播的公式，并学会了推导，将笔记记录如下：
设
f
(
a
,
b
,
c
)
f(a,b,c)
f(a,b,c)为某一计算式，其中的变量带有误差形式，分别为
a
=
a
^
±
Δ
a
,
b
=
b
^
±
Δ
b
,
c
=
c
^
±
Δ
c
a=\hat a \pm \Delta a, b= \hat b \pm \Delta b, c= \hat c \pm \Delta c
a=a^±Δa,b=b^±Δb,c=c^±Δc
则
（1） 算术合成误差传递公式
Δ
f
=
∣
∂
f
∂
a
∣
Δ
a
+
∣
∂
f
∂
b
∣
Δ
b
+
∣
∂
f
∂
c
∣
Δ
c
\Delta f = \left| {\frac{{\partial f}}{{\partial a}}} \right|\Delta a + \left| {\frac{{\partial f}}{{\partial b}}} \right|\Delta b + \left| {\frac{{\partial f}}{{\partial c}}} \right|\Delta c
Δf=∣
∣∂a∂f∣
∣Δa+∣
∣∂b∂f∣
∣Δb+∣
∣∂c∂f∣
∣Δc
Δ
f
f
=
∣
∂
ln
f
∂
a
∣
Δ
a
+
∣
∂
ln
f
∂
b
∣
Δ
b
+
∣
∂
ln
f
∂
c
∣
Δ
c
\frac{{\Delta f}}{f} = \left| {\frac{{\partial \ln f}}{{\partial a}}} \right|\Delta a + \left| {\frac{{\partial \ln f}}{{\partial b}}} \right|\Delta b + \left| {\frac{{\partial \ln f}}{{\partial c}}} \right|\Delta c
fΔf=∣
∣∂a∂lnf∣
∣Δa+∣
∣∂b∂lnf∣
∣Δb+∣
∣∂c∂lnf∣
∣Δc
（2）方和根合成法误差传递公式
S
f
=
(
∂
f
∂
a
)
2
Δ
2
a
+
(
∂
f
∂
b
)
2
Δ
2
b
+
(
∂
f
∂
c
)
2
Δ
2
c
{S_f} = \sqrt {{{\left( {\frac{{\partial f}}{{\partial a}}} \right)}^2}{\Delta ^2}a + {{\left( {\frac{{\partial f}}{{\partial b}}} \right)}^2}{\Delta ^2}b + {{\left( {\frac{{\partial f}}{{\partial c}}} \right)}^2}{\Delta ^2}c}
Sf=(∂a∂f)2Δ2a+(∂b∂f)2Δ2b+(∂c∂f)2Δ2c
S
f
f
=
(
∂
ln
f
∂
a
)
2
Δ
2
a
+
(
∂
ln
f
∂
b
)
2
Δ
2
b
+
(
∂
ln
f
∂
c
)
2
Δ
2
c
\frac{{{S_f}}}{f} = \sqrt {{{\left( {\frac{{\partial \ln f}}{{\partial a}}} \right)}^2}{\Delta ^2}a + {{\left( {\frac{{\partial \ln f}}{{\partial b}}} \right)}^2}{\Delta ^2}b + {{\left( {\frac{{\partial \ln f}}{{\partial c}}} \right)}^2}{\Delta ^2}c}
fSf=(∂a∂lnf)2Δ2a+(∂b∂lnf)2Δ2b+(∂c∂lnf)2Δ2c
关于为什么求对数的思考，因为
d
(
ln
f
)
d
f
=
1
f
⇒
d
(
ln
f
)
=
∂
ln
f
∂
a
d
a
+
∂
ln
f
∂
b
d
b
+
∂
ln
f
∂
c
d
c
=
d
f
f
=
Δ
f
f
=
S
f
f
\frac{{d(\ln f)}}{{df}} = \frac{1}{f} \Rightarrow d(\ln f) = \frac{{\partial \ln f}}{{\partial a}}d a + \frac{{\partial \ln f}}{{\partial b}}d b + \frac{{\partial \ln f}}{{\partial c}}d c = \frac{{df}}{f} = \frac{{\Delta f}}{f} = \frac{{{S_f}}}{f}
dfd(lnf)=f1⇒d(lnf)=∂a∂lnfda+∂b∂lnfdb+∂c∂lnfdc=fdf=fΔf=fSf
关于方和根直接开平方的思考，不妨以
f
=
f
(
a
,
b
)
f=f(a,b)
f=f(a,b)为例，
a
a
a和
b
b
b所服从的分布是独立的，则有
ln
f
=
m
ln
a
+
n
ln
b
E
(
d
ln
f
)
=
E
(
m
ln
a
)
+
E
(
n
ln
b
)
E
2
(
d
ln
f
)
=
E
2
(
m
d
ln
a
)
+
E
2
(
n
d
ln
b
)
+
2
E
(
m
d
ln
a
)
E
(
n
d
ln
b
)
=
E
2
(
m
d
ln
a
)
+
E
2
(
n
d
ln
b
)
⇒
d
2
ln
f
=
(
m
d
ln
a
)
2
+
(
n
d
ln
b
)
2
\ln f = m\ln a + n\ln b\\ E(d\ln f) = E(m\ln a) + E(n\ln b)\\ {E^2} (d\ln f) = {E^2}(md\ln a) + {E^2}(nd\ln b) + 2E(md\ln a)E(nd\ln b) = {E^2}(md\ln a) + {E^2}(nd\ln b)\\ \Rightarrow {d^2}\ln f = {(md\ln a)^2} + {(nd\ln b)^2}
lnf=mlna+nlnbE(dlnf)=E(mlna)+E(nlnb)E2(dlnf)=E2(mdlna)+E2(ndlnb)+2E(mdlna)E(ndlnb)=E2(mdlna)+E2(ndlnb)⇒d2lnf=(mdlna)2+(ndlnb)2
例1：
设有
f
=
4
m
π
d
2
h
,
m
=
m
^
+
Δ
m
,
d
=
d
^
+
Δ
d
,
h
=
h
^
+
Δ
h
f = \frac{{4m}}{{\pi {d^2}h}},m = \hat m + \Delta m,d = \hat d + \Delta d,h = \hat h + \Delta h
f=πd2h4m,m=m^+Δm,d=d^+Δd,h=h^+Δh，求方差传播误差。
解：
首先求对数有
ln
f
=
ln
4
π
+
ln
m
−
2
ln
d
−
ln
h
\ln f = \ln \frac{4}{\pi } + \ln m - 2\ln d - \ln h
lnf=lnπ4+lnm−2lnd−lnh，
分别求偏微分，有
∂
ln
f
∂
m
=
1
m
\frac{{\partial \ln f}}{{\partial m}} = \frac{1}{m}
∂m∂lnf=m1
∂
ln
f
∂
d
=
−
2
d
\frac{{\partial \ln f}}{{\partial d}} = - \frac{2}{d}
∂d∂lnf=−d2
∂
ln
f
∂
h
=
−
1
h
\frac{{\partial \ln f}}{{\partial h}} = - \frac{1}{h}
∂h∂lnf=−h1
所以
S
f
f
=
(
∂
ln
f
∂
a
)
2
Δ
2
a
+
(
∂
ln
f
∂
b
)
2
Δ
2
b
+
(
∂
ln
f
∂
c
)
2
Δ
2
c
=
(
Δ
m
m
)
2
+
(
2
Δ
d
d
)
2
+
(
Δ
h
h
)
2
\begin{array}{l} \frac{{{S_f}}}{f} = \sqrt {{{\left( {\frac{{\partial \ln f}}{{\partial a}}} \right)}^2}{\Delta ^2}a + {{\left( {\frac{{\partial \ln f}}{{\partial b}}} \right)}^2}{\Delta ^2}b + {{\left( {\frac{{\partial \ln f}}{{\partial c}}} \right)}^2}{\Delta ^2}c} \\ = \sqrt {{{\left( {\frac{{\Delta m}}{m}} \right)}^2} + {{\left( {\frac{{2\Delta d}}{d}} \right)}^2} + {{\left( {\frac{{\Delta h}}{h}} \right)}^2}} \end{array}
fSf=(∂a∂lnf)2Δ2a+(∂b∂lnf)2Δ2b+(∂c∂lnf)2Δ2c=(mΔm)2+(d2Δd)2+(hΔh)2
所以
f
=
[
1
±
(
Δ
m
m
)
2
+
(
2
Δ
d
d
)
2
+
(
Δ
h
h
)
2
]
4
m
^
π
d
^
2
h
^
f = \left[ {1 \pm \sqrt {{{\left( {\frac{{\Delta m}}{m}} \right)}^2} + {{\left( {\frac{{2\Delta d}}{d}} \right)}^2} + {{\left( {\frac{{\Delta h}}{h}} \right)}^2}} } \right]\frac{{4\hat m}}{{\pi {{\hat d}^2}\hat h}}
f=⎣
⎡1±(mΔm)2+(d2Δd)2+(hΔh)2⎦
⎤πd^2h^4m^
例2：
设测量到山顶的距离为
247.3
m
247.3m
247.3m，测定角度为
1
0
∘
3
4
′
{10^ \circ }34'
10∘34′，测距误差为
0.05
m
0.05m
0.05m，测角误差为
3
′
3'
3′，试算水平距离的测量误差。
解：
由题意得
L
=
D
⋅
c
o
s
θ
L=D \cdot cos\theta
L=D⋅cosθ， 取对数有
l
n
L
=
l
n
D
+
l
n
c
o
s
θ
ln L =ln D+ln cos \theta
lnL=lnD+lncosθ
所以误差为
S
L
=
L
(
1
D
Δ
D
)
2
+
(
tan
θ
Δ
θ
)
2
=
247.3
∗
cos
(
10
∘
3
4
′
180
∘
π
)
(
1
247.3
×
0.05
)
2
+
(
tan
(
10
∘
3
4
′
180
∘
π
)
×
3
′
180
∘
π
)
2
=
0.0631
m
\begin{array}{l} {S_L} = L\sqrt {{{\left( {\frac{1}{D}\Delta D} \right)}^2} + {{\left( {\tan \theta \Delta \theta } \right)}^2}} \\ = 247.3*\cos \left( {\frac{{{{10}^ \circ }34'}}{{180 \circ }}\pi } \right)\sqrt {{{\left( {\frac{1}{{247.3}} \times 0.05} \right)}^2} + {{\left( {\tan \left( {\frac{{{{10}^ \circ }34'}}{{180 \circ }}\pi } \right) \times \frac{{3'}}{{180 \circ }}\pi } \right)}^2}} \\ = 0.0631m \end{array}
SL=L(D1ΔD)2+(tanθΔθ)2=247.3∗cos(180∘10∘34′π)(247.31×0.05)2+(tan(180∘10∘34′π)×180∘3′π)2=0.0631m
2万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
