How would you resolve latent space drift when applying VAEs for anomaly detection in sequential data

Question

With the help of proper Python programming, can you tell me How you would resolve latent space drift when applying VAEs for anomaly detection in sequential data?