I hate AFCIs and have studied the issues with them pretty thoroughly in the field. The vast majority of cases, it is a false trip. Sometimes there is an over-driven staple. Sometimes a ground wire is touching a neutral in a device box. But do you know what is almost never, ever, ever the problem? And actual arcing fault. One instance in 26 years of these things, just one, have I believed it to be an arc fault due to a loose wire but.
The latest thing I dealt with, literally yesterday, was a toaster oven tripping Square D AFCIs. I detailed this in another post. The circuit is good, verified by a Megger. I went so far as to install a brand new circuit solely for the toaster over, to no avail. Other brands of AFCI that I temporarily plugged in don't trip. The Square D algorithm simply doesn't like that toaster. It is a nice new oven with digital controls. I believe that the AFCI is seeing the PWM that the oven is doing to regulate the temperature as an arc fault.
My solution, unpopular with people who take the code as gospel, was to put in a regular breaker after the final inspection. Inspection was Thursday. Yesterday I put in a regular breaker. The solution is to get rid of AFCIs and use GFCI breakers if you want to protect the branch circuit itself. I believe that an arcing fault will develop a ground fault pretty quickly and GFCIs are tried and true at this point.