This paper shows simulation models for diurnal variation of sub-ionospheric Very Low Frequency (VLF) signals using machine learning approach. Recording of VLF transmitter signals using a ground-based radio receiver provides a beautiful and cost-effective way of monitoring the lower ionosphere (D/E regions) in the altitude range (60-90 km). VLF signals respond to the ionization variations due to the Sun and other terrestrial or extra-terrestrial sources. Consequently, it has many applications in remote sensing of the lower ionosphere. Therefore, predicting or simulating the diurnal variation of VLF transmitter signals using past data will help to understand the variability of the ionosphere. Here, the VLF signal from the Indian transmitter VTX (18.2 kHz) received at Kolkata is used for the training, validating, and testing purposes in the machine learning models. Two predictive models, multiple linear regression (MLR) and artificial neural network (ANN) have been built and Pearson correlation coefficients outside the training range are obtained as R=0.94 and R=0.93 respectively for the two models. Variation of the VLF transmitter signal is also calculated using the well-known Long Wave Propagation Capability (LWPC) code coupled with the International Reference Ionosphere (IRI-2016) model and the same is compared with the MLR and ANN model predictions. Both the MLR and ANN models are found to be performing better than the LWPC simulation.