INTEGRATION OF CONTINUOUS SIGNAL MODULE INSIDE MALTLAB-SIMULINK ENVIRONMENT
Stikanov Valerii, Brytov Oleksii, Artuhov Vitalii, Guiorguizova Viktoriia, Kyriusha Bogdan · Zenodo (CERN European Organization for Nuclear Research) · 2022
Tools of mixed digital-analogue co-simulation of Matlab Simulink package allow enrich developers’ capabilities by user designed simulation system components. We tested possibilities of this integration by transferring optimization module from CAD software ALLTED developed for previous generation of computer systems. In spite of the difference in generations, high performance optimization methods of with variable time step implemented in that system may serve as a useful addition to the built-in optimization components of matlab. Continuous signal part of the model in Matlab is realized in the form of system partial deferential equations. The interface part of the Matlab model is of a procedural type and is used for discrete event processing. While module from Allted is organized vice-versa: procedural model is internal and is being controlled by continuous signal solver. As Allted module is unchangeable binary code co-simulation even driven procedure must be added. There are three classic approaches to the above: 1. integrated co-simulation kernel, in which all models are constructed from other existing models; 2. combined co-simulation with two algorithms sharing access to the same model parameters and states, controlled by even processing unit; 3. separated co-simulation when discrete and linear sub-models are independent and interface operating through external module; On its original forme, none of these three is enough. Custom synchronization procedure was designed; in our case it will be event driven layer between matlab and Allted model. This layer should initiate or pause computations inside of that model subject to external events. The basic structure of the algorithm consists of: 1. If there is no time differences between internal and continuous signal simulation the signal can be transferred “as it is”. 2. If there is a time difference between discrete and continuous signal simulation, we stop internal real time module and interpolate absent data till times became even. 3. In case of a critical difference between interpolated and ALLTED module acquired happens, we adopt parameters of interpolation closer to most trustful source (internal module). The module has been successfully tested by optimization of MEMS gyroscope parameters from matlab examples.