giomoretto / eco Goto Github PK

%% Info Open source repository for ECO

• ECO stands for economic combustion optimization
• ECO solves for the optimal injector inputs of a direct-injection compression-ignition engine
• The continuous-time OCP is formulated with the software packages acados and casadi and discretized by an explicit fourth-order Runge-kutta integration scheme and multiple shooting. The resulting NLP is solved by sequential quadratic programming (SQP) using HPIPM with a condensed horizon of 5. Furthermore, the Hessian is regularized by a Levenberg-Marquardt Regularization.

%% Additional Software Packages Please install additional software in the folder '04_external'

• casadi v.3.4.5 (https://web.casadi.org/)
• acados v0.1.9 or later (https://github.com/acados/acados)

%% Exemplary scripts In order to run the exemplary scripts, please do the following

1. add all folders and subfolders to the working path command window: run path_set.m
2. create s-functions necessary to run ECO in simulink command window: run main_InjOpt.m
3. run exemplary file for use of ECO in matlab command window: run mainMATLAB.m
4. run exemplary file for use of ECO in Simulink command window: run mainSimulink.m

The exemplary file for matlab highlights how ECO can be used to obtain optimal injector inputs under varying constraints which are set by the user. Make sure to initialize the solver with initial values which are close to the solution previously obtained. Otherwise, robust convergence to a feasible solution is not guaranteed.

The exemplary file for simulink highlights how the solutions of ECO can be used in a real-time controller as feedforward control inputs. The bounds, which are not met, can be adapted by a simple PI-controller. The simulink model presented can be used for compilation on embedded systems

%% Cite the code