For RNA-Seq, batch effect can be removed using Combat-Seq: https://github.com/zhangyuqing/ComBat-seq

For Microarray data you can use the limma package:

https://bioconductor.org/packages/release/bioc/html/limma.html

The following function should help you : limma::removeBatchEffect()

The so called 'batch problem' is not a problem but an opportunity. If it does exist an 'average ideal profile of expression' shared by all the samples coming from the same materila (cell kind, tissue..) this is a consequence of the fact each tissue or cell kind is an 'attractor' in the phase space of gene eexpression. This provokes the emrging of a first principal component that is a 'size' component, it is sufficient you start to look for differential expression starting from the second component onward. The mutual indpendence of components will immediately provoke an autoscaling of the data that makes 'batch effect' not to bias the results. On the other hand if you do not find such 'size' (see attached file) first component this casts doubt about the homgeneity of the material.

Amirhossein Hajialiasgary Najafabadi

https://bioconductor.org/packages/release/bioc/html/sva.html

You may check R SVA package.