Unicamp - Press Room

Industrial chemist and food engineer Alexandre Consul Mendes decided to innovate in the animal feed production process. He used the neural network system – intelligent systems – to control the food extrusion process. The results were positive. His research contributed to improving product quality, considering possible variations in process conditions and raw materials. This is an important fact for the feed industry, which is one of the fastest growing in the country.

The idea of developing control systems came from the observation that 90% of companies use the classic process called PID (Proportional, Integral and Derivative). This process, explains the supervisor of the master's thesis “Analysis of the extrusion process in the 'pet food' industry for the application of advanced control techniques”, professor Vivaldo Silveira Junior, would not be the most appropriate in the use of non-linear systems , that is, with sensitivities not proportional to the disorders. Hence the need to use intelligent controls.

Silveira, who coordinates a line of research in the area of controls with fuzzy logic and neural networks at the Faculty of Food Engineering, explains that the work managed to prove that there are better processes with a very good degree of efficiency, which can be alternatives to conventional use . In the study developed by Mendes, the animal feed extrusion machine maintains a uniform production standard throughout the entire scale of the process, without taking into account, for example, whether the flour used has undergone any changes during processing.

In the complex PID mechanism, Silveira explains that it is common for changes in variables such as temperature and pressure to occur during the food journey, which cause errors in standardization. “The example of flour, regarding gelatinization, is one of the most trivial”. With the adoption of the neural network system, the machine adjusts automatically, without the need to re-process because the raw material does not comply. “This increases the quality of the product and increases productivity”, says Silveira.

To develop the innovation, Mendes applied the control parameters at the company Effem, where he works, in the United States. “He adopted the experience at the production plant that produces tons of animal feed.”
Another important aspect of the research is that the methodology used by Mendes can be adopted in any machine that has the extrusion process, such as the one used to produce pasta.

Controls and processes – Silveira teaches the discipline of Instrumentation and Process Control in the Food Engineering course. Therefore, he understands the issue of food quality well. “The lack of standardization of the process or standardization of the product can lead the company to lose the customer”, he argues. He exemplifies by citing the case of traditional brands. “If the formulas change, it causes surprise to the consumer who may not buy the next time.” Therefore, he believes that intelligent controls, carried out in a well-instrumented plant, are the systems capable of maintaining the product's characteristics.

In this aspect, the researcher thinks that the development of new control systems in processes can favor this standardization work. According to him, human manipulation of certain products can alter the quality of the process and consequently the quality of the product, as there are countless situations that can occur. “It is unlikely that a product made by humans will be uniform, as can happen with intelligent systems.”