while applying perceptron models on the data shouldn’t we do data preparation by standardizing the data and then splitting it into test and train?
You build a model using your training data. So yes, before you build your perceptron model (or any other model), you will split your dataset into test and train, followed by using the training data to build the model. In case of building a perceptron model, it basically means identifying w’s and b’s such that loss can be kept minimum without overfitting.
Also whether to standardize your data or not is separate decision choice (data may already be good to go), but yes, if one chose to standardize the data, it will be done before it is fed to the model building process.