In this study, it is proposed, a Deep Learning Model (DLM) focused on obtaining a model based on neural networks to represent the behavior of the temperature of the exothermic process in manufacturing phenol-formaldehyde (PF) resin (resol type). This process was studied in a 3500kg batch reactor, polymerizing phenol with formaldehyde through an alkaline reaction with an average pH of 8.5, and a molar ratio of phenol/formaldehyde that equals 1.5. Unlike the models based on multiple chemical equations of the process that require measurements of concentration, pH, and temperature; in this study, the heat of the reaction is identified employing a method in which the variables are the exothermic energy (due to polymerization of phenol with the formaldehyde), the temperature changes of the reflux line of the reactor, the water flow and the temperature differences in the heat exchanger. This condition is achievable because the PF resin is always manufactured under chemical and thermal constant parameters. Therefore, the most significant achievement is that the neural network identified the reactor temperature with relevant precision (99.5%) during the PF resin manufacturing process. In such a way, the task of obtaining the DLM model was carried out by using the Google Colaboratory (GC) platform to minimize the use of computing resources. Also, the GC is easy to use, and computing processes is performed with vast resources in the cloud. The successesful adjustment of the non-linear dynamic of the process with the DLM model indicates that the proposal is a promising alternative to identify exothermic processes, with the same dosing of reagents per batch, using temperature, flow, and thermal energy sensors. Additionally, this work aims to provide models for the implementation of modern control system, to carry out processes that ensure the rheological properties, quality product and free of runaway chemical reactions that may affect the facilities and the working environment.