Case Study: Cummins achieves faster, more precise engine performance and emissions modeling with MathWorks MATLAB-based AI

Cummins Uses AI-Based Reduced Order Modeling to Predict Engine Performance and Emissions Approach Enhances the Speed and Precision of Engine Performance Models

Cummins, a global leader in engine development, needed a faster and more accurate way to simulate engine cycles for predicting efficiency and emissions. The company’s 3D-to-1D simulations often took more than 20 times longer than real time, making model development slow and resource-intensive. Using MathWorks’ MATLAB environment and tools such as Deep Learning Toolbox and Statistics and Machine Learning Toolbox, Cummins sought a lower-code approach that would let its technical experts spend more time on analysis and less on coding.

MathWorks helped Cummins build LSTM-based neural networks to model 26 engine responses, including pressure, temperature, and engine brake torque. The result was a major speedup in engine cycle simulation, reducing runtime to one-eighth of real time, while also cutting cost, effort, and memory footprint. MathWorks’ platform enabled faster end-to-end AI model development, and Cummins plans to extend the models to real hardware and control components.

Cummins

Shakti Saurabh

Cummins