ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Smart hydroponic system growth analysis

Peter Dolenc (2017) Smart hydroponic system growth analysis. MSc thesis.

Download (8Mb)


    Hydroponic approach to growing plants has many benefits over conventional approach, such as: plants grow faster on smaller area and require less water. Hydroponic system usage in industry is quite well established, however, these systems are usually very specialized and complex. Home users can therefore mostly choose between oversimplified suboptimal or overly demanding hydroponic systems. As a solution to this problem a hydroponic system was designed which has all the capabilities of the complex systems but is simple for managing since it comes with integrated logic that takes over most of the managing tasks. The designed system is capable of fully automated regulation of system parameters and is able to track plant growth by taking bird’s-eye view pictures of each plant separately. Once a day the system will take pictures of all plants and extract leaf area from them. After that it will proceed to find optimal growth settings for the hydroponic system. In order to do so, a special adaptive software was developed that uses two-level machine learning mechanism. On the first level, it uses neural network and gathered data to create model of plant growth based on hydroponic system settings. On the second level the hydroponic settings space is searched using genetic algorithm. Settings that promise the most growth for the plants currently in the system are applied. Plant growth under the developed adaptive regulation software was compared to growth achieved through expert settings in a series of experiments. Comparison of best experiments revealed that adaptive regulation resulted in the final leaf area increase of 13 \% and plant total growth percentage increase by a factor of 2.

    Item Type: Thesis (MSc thesis)
    Keywords: hydroponics, hydroponic system, garden, adaptive system, neural network, genetic algorithm, machine learning, aeroponics, lettuce, embedded system
    Number of Pages: 67
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    izr. prof. dr. Uroš LotričMentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1537359555)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 3778
    Date Deposited: 09 Feb 2017 09:09
    Last Modified: 23 Feb 2017 11:20
    URI: http://eprints.fri.uni-lj.si/id/eprint/3778

    Actions (login required)

    View Item