Rolling piston compressors are small capacity volumetric machines used mainly in household
refrigerator and heat-pump units. One of the main characteristics of the compressor
is the way how the pressure builds up within the cylinder. This information can be
used in more complex models for further investigations, e.g. to analyse the mechanical
and thermal stresses of the compressor components. Modelling methods to resemble the
pressurisation process of the rolling piston compressor with non-linear oscillators
is presented in this paper. The mathematical description of the non-linear oscillators
can be used to produce a continuous signal resembling the pressure fluctuations within
the cylinder. For the definition of the necessary functions, the main characteristics
of the typical rolling piston pressurisation diagram were identified. At first the
pressurisation process was approximated by using the mathematical formula of the common
Van der Pol oscillator. The oscillator was modified and extended with a linear second
order differential equation to improve the resemblance between the target and model
functions. To achieve better agreement with the target diagrams, Morphed non-linear
oscillators were established. Polynomial and constructive incremental learning method
was used to achieve adequate data fitting and the results of both methods were compared.
The results show acceptable conformity and proved to be adequate to provide input
data for further investigations i.e. for parametric studies of the compressor design.
Besides modelling rolling piston compressors, this method can be applied for other
volumetric compressors with cyclic pressurization sequence.
