along with the implementation of the corresponding optimal co-design strategy.
Note lines 1225 and 1226 (Vehicle.m) and see that I have penalized the local variable gammaSq_i heavily in the cost function used in the decentralized design (at each subsystem i). I found that this is necessary as otherwise, gammaSq_i seems to increase with i . Another way that I saw working is when I constrain each gammaSq_i based on previous gammaSq_j values (j<i) saw in the decentralized design process. Note that gammaSq = max(gammaSq_i). In the centralized design, we had the gammaSq variable directly in our design problem. But here in the decentralized design it is not the case. In all, we have to be careful when designing the cost functions used in the decentralized process.
along with the implementation of the corresponding optimal co-design strategy.
Note lines 1225 and 1226 (Vehicle.m) and see that I have penalized the local variable gammaSq_i heavily in the cost function used in the decentralized design (at each subsystem i). I found that this is necessary as otherwise, gammaSq_i seems to increase with i . Another way that I saw working is when I constrain each gammaSq_i based on previous gammaSq_j values (j<i) saw in the decentralized design process. Note that gammaSq = max(gammaSq_i). In the centralized design, we had the gammaSq variable directly in our design problem. But here in the decentralized design it is not the case. In all, we have to be careful when designing the cost functions used in the decentralized process.