Re: One solution...
The only reason I can see for down voting (which I didn't do as I agree with the underlying principle) is that the vulnerable person doesn't see the middle account. It is highly likely that the alleged abuser already has the persons real bank details.
I would suggest you idea holds if the vulnerable person is given a new account which mirrors the existing account so any activity is matched on both saving the person having to update bank details with others unnecessarily as payments continue to go in and out the original account.
The original account would then have a flag on it for any inappropriate references which could allow money to go through but block the sender details which are then referred to the police who would possibly have a case file that could be used to then chase the abuser down.
2 problems - Would the bank want to take on the extra complexity and costs first of all as there isn't any real benefit to them beyond a customer service aspect (shouldn't be underrated)
The second problem is the referral to the police where experience suggests it will fall flat on it's face.
I have a friend who separated and then divorced her husband. He has then failed to pay any maintenance for their son but continues to drive around where she lives, sit in the car park where she works (out of town shop) hours before the store opens,send abuse from and his new partner etc. etc.
The police have taken all the evidence promising action would be taken, including an arrest, warnings and possible court action for harassment then decided they couldn't do anything despite all the evidence and it would need to be a civil action. When another incident occurred the police comment was they didn't know how no action had been taken but couldn't find the ex-husband, then got an address and didn't take action. The cycle still continues.
Child maintenance charge my friend for guaranteeing payment of child support, they have investigated the ex husband and worked out he owes over £5,000 back payments. In 2 years they recovered exactly zero but when he does make the odd small payment they take their cut for 'handling it'.
If the authorities took proper action then the banks might review it and decide there is a basis for putting in protections as there would be action at the end of it, otherwise why spend the money / investment in systems when nothing else will be done to stop the alleged perpetrator continuing what they are doing?