Egg Production Line. Software for Automation

Automation helps various areas of human knowledge to convert daily routines performed manually into automatic routines which, in general, are faster and safer (Bezerra, 2007).
In many companies the selection of the egg is still done manually and in very unsanitary conditions for those who perform it. In addition, at the sites involved in these processes the information is collected and stored in vague or informal manner, which does not allow adequate control of production.
That is why an automatic selection and processing of eggs is very useful because it avoids losses due to vaccinating unfertilized eggs, the cost of hiring workers, labor expenses associated with problems caused by repetitive stress and unhealthy situations, besides creating security and greater control on the process. Thus, this paper presents the specification of a management program not only for the classification of the eggs, but also the control of production through a historical record by lot, which will determine the number of eggs that were not exploited, allowing the supervisor to identify the most productive breeding houses at that time and record any anomalies in the execution of the incubation or the acquisition of a complete batch.
The program enables getting information on the lot from the moment it arrives until it is sent to the farm, allowing even better control of incubators and issuing alerts and reports on production statistics over time.

A preliminary analysis was performed to determine requirements and identify existing failures in the control of the production line of a typical incubation business. This stage was conducted through interviews and observation of the whole process, taking into consideration the theories resulting from the requirement engineering (Pressman, 2006). After determining the requirements and failures, an outline of the system is prepared using diagrams and it also outlines the tasks to be executed thereby. Figure 1 shows one of the diagrams generated. With this, the operations can be logically and rationally designed before being encoded. This stage is critical to the development of a software program, so much so that several authors emphasize its importance (Sommerville, 2007).

Figure 1. Class Diagram proposed

In order to facilitate its construction, the program creation process was divided into two parts. The first with regard to the classification of the eggs and the second relates to the control of the production line.

In the first part, designed by digital candling, the need for the program to be able of sorting the eggs to see if they are fertilized or not was determined, being it necessary therefore to have image acquisition. At this stage we used a Sony CyberShot digital camera to take pictures with settings of 7 mp. Visits to the premises of a company were conducted to obtain images of the production line. Like most companies, the eggs are reviewed and classified manually before the vaccination, as shown in Figure 2.

Figure 2. Image taken as the reference standard for the study and classification

 

Several photographs were taken to build an image bank. The eggs were on day 18 of incubation. Nevertheless, many of these photographs were discarded for not presenting the possibility of performing digital candling, since it presupposes the use of a light source beneath the eggs and the shot taken from certain angles producing shadows prevents classification. That was why several tests were performed to obtain a good enough angle for classification.

The section on the candling was performed using the Matlab environment that has features that facilitate processing of numerical matrix, since ultimately, a picture is nothing other than a matrix. First, the overlay of images is performed with a mask (Figure 3) to remove irrelevant parts of the image through logical operations between it and the mask. The program runs through the image using the mask to locate the eggs. Once the eggs are located, another module is run for the classification, storing the status of the egg and its position.

Figure 3. Image with the mask, leaving only fractions of the image to be studied

 

The method that carries out the classification of the egg is based on the use of a threshold applied to the red, green and blue levels (RGB). See Gonzalez and Woods (2008) for more information on using the threshold.

We propose that the method which evaluates the condition of the eggs after classifying, sends the information of the positions where the unfertilized eggs are located so that the machine only removes the good ones from the tray leaving the unfertilized ones to be discarded. The vaccinator machine also receives information about which eggs should be discarded and, thus, will not inject them, avoiding wastage of vaccine.

Besides the control program, there will be a database that will store the information of production houses, suppliers, destination of chickens, egg lots and information on batch processing (arrival and departure from the incubator, the percentage of fertilized eggs, quantity of eggs processed per line/shift, chickens usable), among other statistical and traceability data. Thus, the administrator of the production line will have more real-time information on the lot assessed. The information contained in the database may vary depending on the needs of each incubation company.

The software will also make a sortable list of more cost-effective providers. With the historical record, the system administrator can assess more accurately, identifying failures occurring in production. The second part of the project was implemented using the Java language and NetBeans environment. This selection was made because the pre-compiled and portable set of features, and since it is object-driven, allowing easy maintenance, portability, reuse, etc.

Implementing the system requires the use of some equipment: 3 computers (or a CPU and three satellite terminals to reduce costs), a vaccinator machine, imaging equipment in a controlled environment, a machine with suction pads to remove the eggs. Both the vaccinator machine and the egg removing machine must support the programing through the software. Figure 4 illustrates the interaction between these pieces of equipment.

Figure 4. Illustration of the order in which the components will interact and flow of information

According to direct observations in a company in the industry, it is estimated that a production line losses up to 20% of the eggs. Candling enables the vaccination of these eggs, which can be performed automatically or manually. Manual candling creates significant costs, for example, in a production line of 100,000 eggs, about 8 workers are needed to perform the task. Over the months, the expense for these workers is quite considerable. Machado (2009) proposes the digital candling taking into account the arrival of the egg, but based on what has been observed during the requirements analysis, this is not a suitable procedure in fertile eggs, as the large-scale application may not work, forcing individual manipulation of the eggs, which if too intense can compromise embryonic development , delay production, and finally does not minimize losses and expenses in the hatcheries, since candling performed at the end is critical given that it is at this point that one can say with certainty which eggs should be vaccinated and which should be discarded. It is also important to note that the eggs go through selection during collection in the houses (sheds or galleys) and, since the system will control production, it is difficult to see which suppliers are offering a quality product.

With regard to control, it is still in testing and adaptation, being so far that the unit tests are quite satisfactory, while in production environments, the tests are conditional to repetition, a factor that is also necessary to implement the system as a whole, so we are looking for companies that allow us to perform the test in order to complete this project.

With a computer program to correctly classify all the eggs in a batch, quickly and without the use of manual processes, the vaccination of infertile eggs is avoided, saving money and ensuring the safety of the process. It is estimated that in a one-year term, such a system would produce savings of around RXXnbsp;150,000.00 in a production line of 100,000 eggs/day due to the reduction of labor and savings in vaccines.

In addition, a system with dynamic and flexible features with the characteristics specified herein, will lead to an increase in the organization of the company, facilitating and accelerating the collection and recovery of data, generating important reports for faster and more accurate management of the production line.

E. Bezerra 2007. Princípios de Análise e Projeto e Sistemas com UML. Elsevier Rio de Janeiro.

Gonzalez RC & Woods RE. 2008. Digital Image Processing. Prentice Hall. New York.

Machado DS. 2009. Sistema de inspeção visual automática aplicado ao controle de qualidade de ovos em linhas de produção. Dissertação de Mestrado. Centro Federal de Educação Tecnológica de Minas Gerais. Belo Horizonte.

Pressman RS. 2006. Engenharia de software. McGrawHill. São Paulo.

Sommerville I. 2007. Engenharia de Software. Pearson Addison-Wesley. São Paulo.