Design of an intelligent control system for small-scale poultry farms

Abstract

The world has been experiencing rapid growth in the poultry sector and this growth is projected to continue in the foreseeable future. The per capita consumption of poultry products is increasing rapidly and it is putting stress on the poultry farmers especially in low-income countries. This is so because low-income countries depend on small-scale farmers for their poultry products but small-scale farmers fail to afford the inclusion of technology which would help them to improve their production the same way it improves production for medium to large scale farmers. With these small-scale farmers in mind, it is, therefore, a requirement that technology is developed to ensure improved production while maintaining low-cost input as a requirement for small-scale production. This study focused on using new methods in the development of an embedded control system for poultry farms. Firstly, temperature and humidity data were collected to be used to determine their distribution in a room. It was found that in a well-ventilated room, the temperature and humidity have an equal distribution such that one sensor is enough to monitor both these parameters. This result was then used to extend to rooms with one heater, another with more than one heater. The circle packing algorithm was then used to propose a way to arrange the heaters across a poultry house. Based on this arrangement, a sensor placement algorithm was then proposed which enables the monitoring of the poultry farm in zones such that a system malfunction in one zone can be pinpointed and not be let to affect the system performance of the whole farm. Having proposed the sensor placement model, the way actuators are controlled was also examined. A hypothesis was proposed to say that if the relationship between temperature and humidity was considered during their control, an algorithm would be designed to leverage the relationship and achieve control of the said parameters using less electrical energy. An algorithm was then designed and simulated. A comparison was done against a conventional algorithm that does not consider the temperature-humidity relationship. It showed that using the proposed algorithm, the actuators controlled the respective parameters in less time hence using less electrical energy. The actuators also operated in short bursts when using the conventional algorithm as compared to the proposed algorithm. The hypothesis was them proved to be true. After these two algorithms were developed, an embedded system was designed using the algorithms. It was designed to have multiple sensor and actuators for each of the monitored environmental parameters. It was also designed with multiple communication channels to be able to send data to the cloud even when one of the communication channels is broken. This embedded system was then coupled with the cloud to provide real-time monitoring of the system performance and data storage for later use. The cloud had an API, Database and a web data dashboard. The embedded system sent data to the cloud through this API and the API persisted the data to the database. Simultaneously, the data was also sent to the front-end dashboard through the same MQTT broker to which the dashboard was subscribed to. Users were provided with the ability to monitor the poultry farms in real-time. The users also use the same dashboard application to set configurations for the embedded system as well as to manually control some actuators as an override to the automatic operation of the system. The study achieved all its objectives by doing the above discussed but there is still much more potential for further research in the field of poultry farm control systems