结合求解和dsolve来求解具有微分和代数方程的方程组

时间:2013-07-17 13:54:38

标签: matlab differential-equations

我正在尝试解决方程系统,其中包含代数和微分方程。为了象征性地这样做,我需要结合dsolve和solve(是吗?)。

考虑以下示例: 我们有三个基本方程式

a == b + c; % algebraic equation
diff(b,1) == 1/C1*y(t); % differential equation 1
diff(c,1) == 1/C2*y(t); % differential equation 2

求解两个微分方程,消除int(y,0..t)然后求解c = f(C1,C2,a)得到

C1*b == C2*c   or   C1*(a-c) == C2*c
c = C1/(C1+C2) * a

我如何说服Matlab给我这个结果?这是我试过的:

syms a b c y C1 C2;
Eq1 = a == b + c; % algebraic equation
dEq1 = 'Db == 1/C1*y(t)'; % differential equation 1
dEq2 = 'Dc == 1/C2*y(t)'; % differential equation 2
[sol_dEq1, sol_dEq2]=dsolve(dEq1,dEq2,'b(0)==0','c(0)==0'); % this works, but no inclusion of algebraic equation
%[sol_dEq1, sol_dEq2]=dsolve(dEq1,dEq2,Eq1,'c'); % does not work
%solve(Eq1,dEq1,dEq2,'c') % does not work
%solve(Eq1,sol_dEq_C1,sol_dEq_C2,'c') % does not work

我尝试过的方程式或解决方案中没有求解和/或解析的组合给了我一个有用的结果。有什么想法吗?

1 个答案:

答案 0 :(得分:0)

现在我假设您希望代码相当通用,所以我使它能够处理任何给定数量的方程式和任何给定数量的变量,并且我没有手动计算。

请注意,符号工具箱的工作方式每年都会发生巨大变化,但希望这对您有用。现在可以将等式Eq1添加到dSolve的输入列表中,但有两个问题:一个是dSolve似乎更喜欢字符输入,第二个是dSolve 1}}似乎没有意识到有3个独立变量abc(它只看到2个变量,b和{{1} })。

为了解决第二个问题,我区分了原始方程以得到一个新的微分方程,存在三个问题:第一个是Matlab评估c相对于a的导数为t,因此我必须将0替换为a,例如a(t)b(我称c a(t)的长版{ {1}})。第二个问题是Matlab使用了不一致的表示法,而不是将a的导数表示为a,它将其表示为Da,因此我不得不用前者替换后者,例如diff(a(t), t)b;这给了我c。最后一个问题是系统现在已经确定了,所以我必须得到初始值,这里我可以解决Da = Db + Dc,但Matlab似乎对使用a(0)感到满意。

现在回到最初的第一个问题,解决我必须将每个sym转换回char。

这是代码

a(0) = b(0) + c(0)

一些小注释,我将function SolveExample syms a b c y C1 C2 t; Eq1 = sym('a = b + c'); dEq1 = 'Db = 1/C1*y(t)'; dEq2 = 'Dc = 1/C2*y(t)'; [dEq3, initEq3] = ... TurnEqIntoDEq(Eq1, [a b c], t, 0); % In the most general case Eq1 will be an array % and thus DEq3 will be one too dEq3_char = SymArray2CharCell(dEq3); initEq3_char = SymArray2CharCell(initEq3); % Below is the same as % dsolve(dEq1, dEq2, 'Da = Db + Dc', ... % 'b(0)=0','c(0)=0', 'a(0) = b(0) + c(0)', 't'); [sol_dEq1, sol_dEq2, sol_dEq3] = dsolve(... dEq1, dEq2, dEq3_char{:}, ... 'b(0)=0','c(0)=0', initEq3_char{:}, 't') end function [D_Eq, initEq] = ... TurnEqIntoDEq(eq, depVars, indepVar, initialVal) % Note that eq and depVars % may all be vectors or scalars % and they need not be the same size. % eq = equations % depVars = dependent variables % indepVar = independent variable % initialVal = initial value of indepVar depVarsLong = sym(zeros(size(depVars))); for k = 1:numel(depVars) % Make the variables functions % eg. a becomes a(t) % This is so that diff(a, t) does not become 0 depVarsLong(k) = sym([char(depVars(k)) '(' ... char(indepVar) ')']); end % Next make the equation in terms of these functions eqLong = subs(eq, depVars, depVarsLong); % Now find the ODE corresponding to the equation D_EqLong = diff(eqLong, indepVar); % Now replace all the long terms like 'diff(a(t), t)' % with short terms like 'Da' % otherwise dSolve will not work. % First make the short variables 'Da' D_depVarsShort = sym(zeros(size(depVars))); for k = 1:numel(depVars) D_depVarsShort(k) = sym(['D' char(depVars(k))]); end % Next make the long names like 'diff(a(t), t)' D_depVarsLong = diff(depVarsLong, indepVar); % Finally replace D_Eq = subs(D_EqLong, D_depVarsLong, D_depVarsShort); % Finally determine the equation % governing the initial values initEq = subs(eqLong, indepVar, initialVal); end function cc = SymArray2CharCell(sa) cc = cell(size(sa)); for k = 1:numel(sa) cc{k} = char(sa(k)); end end 更改为==,因为这似乎是我们的Matlab版本之间的差异。我还在=中添加了t作为自变量。我还假设你知道细胞,数字,线性指数等等。