modelling cyclic plasticity in abaqus

I have to model cyclic plasticity in ABAQUS. The method to use ( this is as per ASME Boiler and Pressure Vessel code) involves only a single loading ( rather the load range has o be applied) and using a stabilized cyclic stress strain data expresssed in Ramberg Osgood format. The parameters for Ramberg Osgood are specified in the equation to be used fpr stabilized cyclic stress strain curve.

 Implementing the same in ABAQUS, gives me the following possibilities :

1) Hardening type Isotropic-which I believe should not be used with cyclic loading.

2) Hardening type Kinematic -where only 2 data points can be given as input-thereby making the graph a bilinear one. The first data point is yield stress at zero plastic strain. I think I can use this model with the second data point computed using the Ramberg Osgood type equation provided in the code . However I feel that since this input is only for two points, it does not represent the data in adequate form.

3) Use combined hardening where kinematic and isotropic components can be specified independently.In this option, the no. of data points for kinematic hardening is no longer confied to 2 points , thereby making the graph a nonlinear one. I think I can use the Ramberg Osgood type relationship from the code to generate data for multiple points. I am not sure how the stabilized stress strain data can be used to specify the isotropic hardening part of the input. 

Kindly let me know if my above perception is correct or is there any better or more accurate way by which I can give apply the available data.

The basis for the single application of loading and using cycli stress strain data is somewhat like this: Many materials, after initial period of cyclic hardening/softening exhibit a stable behaviour when the hysteresis loops stablize.The cyclic stress strain curve is generated by using these stabilized loops.Thse stabilized loops have been considered to be representaive of the whole fatigue life and hence the applied loading in the form of range and using the cyclic stress strain data correctly simulates the cylic behaviour w/o having to do a cycle by cycle counting.

Regards