WHY Solver not applicable (cplex)

[凌梓]

Dear Löfberg,

     I want to use cplex to solve the problem of energy interconnection optimization.

    When I finished all the programs, it was found that Solver not applicable (cplex). I guess it might be that my constraints are problematic, and when I delete some constraints(See below), CPLEX can be applicable.
But these constraints are the most core part of my entire program. What should I do?

for k = 1:24

Constraints =[Constraints,Beta(k)*(Pe_net(k)+Pw(k)-P_p2g(k)+2.934*Gamma(k)*(Pg_net(k)+0.0987*P_p2g(k)-Pg_dc(k)))-Pe_dc(k)==Pe_load(k)]; %电功率平衡约束
Constraints =[Constraints,0.9*(1-Beta(k))*(Pe_net(k)+Pw(k)-P_p2g(k)+2.934*Gamma(k)*(Pg_net(k)+0.0987*P_p2g(k)-Pg_dc(k)))+6.651*Gamma(k)*(Pg_net(k)+0.0987*P_p2g(k)-Pg_dc(k))+7.9218*(1-Gamma(k))*(Pg_net(k)+0.0987*P_p2g(k)-Pg_dc(k))-Ph_dc(k)==(Pc_load(k)/1.38)+Ph_load(k)]; %冷热功率平衡约束
 end

for k = 1:24
  Constraints = [Constraints, 0<=Gamma(k)*(Pg_net(k)+Pg_dc(k)-0.0987*P_p2g(k))<=204.499];%微燃机额定容量约束
  Constraints = [Constraints, 0<=(1-Gamma(k))*(Pg_net(k)+Pg_dc(k)-0.0987*P_p2g(k))<=71.515];%燃气锅炉额定容量约束
  Constraints = [Constraints, 0<=(1-Beta(k))*(Pe_net(k)+Pw(k)-P_p2g(k)+2.934*Gamma(k)*(Pg_net(k)+0.0987*P_p2g(k)+Pg_dc(k)))<=500];%电锅炉额定容量约束
end

Thank you!

[Johan Löfberg]

You have loads of nonconvex products between decision variables. That is not allowed in *linear* (or convex quadratic) models

Maybe you have things like turning on and off stuff, which you never should do using products

https://yalmip.github.io/example/unitcommitment/

[凌梓]

thanks a lot .I will try to modify it.

[liuyan...@gmail.com]

同学你好，我是做电力系统规划的，新用yalmip，最近也碰到这个问题，是不是约束里不能变量相乘求解呢？你的这个问题解决了么？求指导。。谢谢啦

[凌梓]

您好。不好意思刚刚看见，约束中变量是不能相乘的，你可以看看Johan Löfberg给的这个例子

https://yalmip.github.io/example/unitcommitment/

我的解决思路是尝试重新选择变量，按照例子中的思路，换种放式来确定约束的，避免了变量相乘。

[王韬]

朋友你好，请问可不可以具体讲讲你是怎么避免变量相乘的呢？我也遇到了同样的问题。谢谢

[Johan Löfberg]

This is an English speaking forum,

however, the questions is discussed a lot here

https://yalmip.github.io/example/unitcommitment/

[王韬]

Sorry for that

I do run into a trouble and I'm fresh in this field, so I got a little bit anxious.

I've read that example twice before, but not got much help.

gonna read that again, and if I still couldn't solve it, hope you can give me some help.

SO MUCH THX!

[Johan Löfberg]

to model x*y where x is continuous and y is binary, you introduce new variable z and implies(y, x==z)+implies(1-y, z == 0)