According to trial and error z nbsp nbsp seems

A reasonable range for all design variables AGI-6780 also determined from experience and are shown in Table 2.
Table 2.
Design variable ranges and required precision.Design variabled: Displacer stroke (m)ö: Phase angle (deg)f: Frequency (Hz)Minimum acceptable value0.026012Maximum acceptable value0.1212022Required precision±0.001±0.25±0.25Full-size tableTable optionsView in workspaceDownload as CSV
4. Neural network
The total optimization procedure consists of two parts; pattern recognition and interpolation.
First, consider the engine model nature as a static transfer function. The engine model gets the design variables and gives output power and efficiency. While the inputs change, each of the outputs may uniformly ascend, uniformly descend or consecutively ascend and descend. There is also a situation in which some of input variations negate the others; thus the outputs may remain almost constant. For example, convective heat transfer coefficient in a heat exchanger pipe increases whenever the piston swept volume increases. But microevolution may remain constant if engine working frequency decreases simultaneously.