In the industrial world, mixing operations are very widely used to process raw materials into products such as petroleum, chemicals, and various other types, so increasing the effectiveness of mixing operations is necessary. Usually, in industrial-scale plants, the conventional PID controller is used in the main controllers that are relied upon, but they often underperform, especially when dealing with non-linear systems. In this research, a neural network (NN)-based controller is proposed to overcome that problem. The plant model used for the simulation in this research is a mixing water plant, where the temperature and water level in the mixture tank will be controlled. This plant consists of 2 water flow inputs, namely cold water and hot water, which flows into the mixed tank where the temperature and water level of the mixture will be maintained and controlled according to the desired set points by regulating the flow of cold water and hot water. There are two types of NN-based controllers in this simulation study. The first is NN-based controller which has inputs in the form of the set point (SP), the current step (n) process variable (PV), and the previous step (n-1) process variable, while the second is a NN-based controller with set point (SP), error, and change of error for the inputs. Both of these NN-based controllers were developed using feed-forward neural networks, and the simulation was conducted by using MATLAB/SIMULINK. The simulation results show that the proposed NN-based controllers provide better performances when compared to conventional PID controllers. The best performance is obtained using the NN-based controller that has inputs in the form of set point, error, and change of error, with settling time of 472.7 s and rise time of 242.0 s in controlling the temperature, and settling time of 984.4 s, rise time about 447.6 s in controlling the level, that is faster than the PID controller which has 613.7 s and 393.9 s for settling time and rise time respectively in controlling the temperature, while gives settling time of 3216 s and rise time 412.8 s in controlling the level. Moreover, the NN-based controller produces a system response that has no overshoot at all.