Driving cycles play a fundamental role in the design of components, in the optimization of control strategies for drivetrain topologies, and in the identification of vehicle properties. The focus on a single or a few test cycles results in a risk of non-optimal or even poor design regarding the real usage profiles. Ideally, multiple different driving cycles that are representative of the real and scattering operating conditions are used. Therefore, tools for the stochastic generation of representative driving cycles are required, and many works have addressed this issue with different approaches. Until now, the stochastic generation of representative testing cycles has been limited to low dimensionality, and only a few works have studied higher dimensionality using Markov chain theory. However, it is mandatory to create tools that can stochastically generate multidimensional cycles incorporating all relevant operating conditions and maintaining signal dependency at the same time. For this purpose, a new method to synthesize multidimensional and representative testing cycles that can handle constraints and is suited for many evaluation criteria is presented in this study. The performance of the new method is demonstrated for a typical use case and is compared to the existing method as a reference. It is shown that the new method can outperform the existing multidimensional method regarding cycle quality and computational efficiency in many cases. The proposed method regards the synthesis of driving cycles for vehicles but could be adopted to create testing cycles for any other machine.
