Mostly by reading the literature & noticing needs for further investigation. Sometimes ideas come from 'real life' as well.

Near the beginning of your PhD program, you should be reading studies and papers across your field of study in a more general sense, often at the direction of your professors and advisors.

As you read those papers, there should be specific details or areas that stand out to you or interest you. These specific areas or references that you come across should each present a new line of research for you to pursue, a different "rabbit hole", so to speak. When you begin looking into these sources that interest you, they will then lead to others and then others still. Some you might find fruitless because they no longer interest you once you've looked into them further or perhaps most of the questions in that line have already been answered and it's no longer a topic that is being actively investigated or of interest in the field.

You must also always keep in mind that whatever path you go down, you must be able to realistically research it in your educational environment. Students might come up with very "big", important questions, but if you have no realistic way to gather data and produce a study that yields valid results to answer them, then it's not much use pursuing it at the moment, until you actually have the resources to test the assumptions.

15 years after the start of my PhD, this is still how I find the topics I ultimately publish on- I read research studies related to my field on topics that particularly interest me, and then I trace the sources back to find previous questions, measurement instruments, where the research now stands, etc., and then I figure out how to extend the conversation and apply the information within a context that is feasible to me.

I am a master's student and I lack the experience, but from what I have gathered last 3,5 years running a series of studies, it is good to stick to one, really well defined and concrete domain of research. This helps you get in the loop, provides a clear framework in which you can think, and allows you to structure your research programm on a step by step basis.

I dont know the requirements of a phd, but novelty should be a a small part of what is being done. Follow the work of 3-4 established researchers and slowly add to what they do. Replicating part of what has been done already is never bad, it is necessary. Also, when you study, stick to 1 thing and drive it to its limits.

This has worked for me pretty well :D hope it does for you too!