Computational specialist MathWorks has introduced a new version of the MATLAB and Simulink software families. R2022b includes two fresh products and various other features that simplify and automate model-based design.
Simscape battery, which is one of the additions, provides design tools and parameterized models for electric vehicle battery systems. It enables creation of digital twins, virtual tests of battery pack architectures, design of battery management systems, and evaluation of battery system behavior across normal and fault conditions. Crucially, Simscape Battery also automates the formation of simulation models to specific pack topologies. Furthermore, it includes cooling plate connections for the testing of electrical and thermal responses.
“We’re excited to launch Simscape Battery as innovation in battery management systems is at an all-time high. The new product includes many design tools intended to simplify and automate model-based design, including the battery pack model builder that lets engineers interactively create and evaluate different battery pack architectures,” said Graham Dudgeon, principal product manager, electrical systems modeling, MathWorks.
Fundamentally R2022b features updates to some of the widely used MATLAB and Simulink tools, which augment the development and testing process in several ways.
The Autosar Blockset enables development of services-oriented applications using client-server ARA methods and deployment on embedded Linux platforms. The tool lets users define data types and interfaces in an architecture model.
The fuzzy Logic Toolbox supports design, analysis, and simulation of fuzzy inference systems (FIS) interactively using the updated fuzzy logic designer app. In addition, the enhanced toolbox enables engineers and researchers to design type-2 FIS using command-line functions or the fuzzy logic designer app.
The HDL coder offers generation of optimized SystemC code from MATLAB for high-level synthesis and use of the frame-to-sample conversion for model and code optimization.
The model predictive control toolbox uses neural networks as prediction models for nonlinear model predictive controllers. Additionally, the toolbox now facilitates users implementing model predictive controllers that meet ISO 26262 and MISRA C standards.
The System Identification Toolbox also supports creation of deep learning-based nonlinear state space models using neural ordinary differential equations. Machine learning and deep learning techniques can also represent nonlinear dynamics in nonlinear ARX and Hammerstein-Wiener models.
