Using JSON with Pyomo is a pain due to couple of reasons. With simple models, the translation layer provided by Pyomo called DataPortal (2) works fine. However, with more complex interactions, the system probably will not allow flexibility. Here are a few notes to remember when using JSON with Pyomo:

• Remember to stick to standard JSON schema. This will allow you to parse values conveniently.
• If your model is simple and is not too nested, you can easily pass a standard JSON object to DataPortal().load(filename=<name_of_file>) to create your input data for the model.
• Pyomo does not support numpy.float64 types. All types need to be standard Python types so remember to cast them to float. This can be checked using functions from numpy.
• For AbstractModel with IndexedParam in nested Blocks, it will be easier to write convenience functions to (a) parse the values correctly and (b) convert the indexed param to the correct dictionary format for Pyomo
• Use to_json() wherever possible to convert Pyomo objects to serializable and readable format

References

1. Dictionaries with Pyomo
2. DataPortals at Pyomo Documentation