If the box boots to the kernel a drive in an "unclean" state will be detected and automatically fsck'd. If the drive doesn't get that far, the issue resides in upstream u-boot; which likely follows the Linux "stop in case you break something" approach rather than the vendor u-boot) "run
vendor code to fix things" approach. I'm not sure there's much to be done here, but the workaround would be to install HD vendor u-boot on the SD/eMMC and then boot LE using the "box" image instead of the "board" image. Then you have the benefits of very poorly written vendor code "fixing" your broken filesystem at the u-boot stage and it might continue booting.
I used to live in a house with frequent power cuts and no amount of software workarounds ever beat running essentials (things that you don't want to constantly fix or risk breaking) from a small UPS.
That never happened with CoreElec or LibreElec 9.0.2, just odd that LibreElec 11 builds are doing this. It's almost like write cache is open all the time and when a power cut happens, it borks something because caching was left open. LibreElec 9.0.2 build doesn't do this, only the LE 11 builds