Internet of Things (IoT) is receiving a great attention due to its potential strength and ability to be integrated into any complex systems and it is becoming a great tool to acquire data from particular environment to the cloud. Data that are acquired from Wireless Sensor Nodes(WSN) could be predicted using Artificial Neural Network(ANN) models. One of the use case fields of IoT is smart agriculture and there are still issues on developing low cost and power efficient WSN using advanced radio technologies for short and long-range applications and implementation of prediction tools. This is the reason why the target of this thesis is to develop a low cost and power efficient WSN and IoT based control system and analyze the best predictive model for such systems. With this purpose, we developed BLESensor node for short-range IoT applications and Internet of Plant(IoP) for long distance smart agriculture applications. A non-linear prediction model is developed in order to forecast acquired data from sensor nodes. BLESensor node Experimental test results reveal that newly developed BLESensor node has a good impact on the improved lifetime and applications could possibly make this emerging technological area more useful. The Android software has been tested on Samsung Galaxy SM-T311, running Android 4.4.2 and it works without any issues and it is supposed to work on all other Android devices equipped with BLE. The working temperature range of the BLESensor node is supposed to work goes from -20 °C to 70 °C due to battery temperature limits. The system has been tested in the climatic chamber (Challenge 250 from Angelantoni) present at the Neuronica Lab, which allowed the sensor to be software calibrated. Several measurements have been proven that each node offers an uncertainty of 1.2 °C for temperature. These values are acceptable for the type of application for which they are intended. The power consumption has been measured directly from scope analysis and simulating the code step by step and calculations resulted that the lifetime of the node lasts for a month. Considering a normal use of these sensors with a reasonable sampling time the lifetime could be increased. IoP node IoP node is a prototype device that works with WiFi protocol and collects temperature, humidity and soil moisture data of plants to the cloud. For IoP node, we have implemented a firmware, tested a prototype device and designed the PCB in OrCAD software and generated a Gerber file and developed an android application. Prediction model Comparisonofthreenon-linearmodelswithOakdatasetresultedinbetterperformance of NNARX model and we used NNARX model to predict 10 days step ahead maximum and minimum temperature and described the results of performances. The performance given by trained models in terms of Mean Square Error (MSE) for maximum temperature prediction provided an error of 0.8826 on unseen data for the month of September. Similarly, the performance of model predicting minimum temperature was tested and it resulted in an error value of 0.944. In conclusion, this work must be intended only as a proof-of-concept, although, the developed BLESensor system, IoP prototype device and predictive models showed expected optimum results, both in terms of functionalities and usability.

Internet of Things Applications and Artificial Neural Networks in Smart Agriculture / Aliev, Khurshid. - (2018 Jan 09).

Internet of Things Applications and Artificial Neural Networks in Smart Agriculture

ALIEV, KHURSHID
2018

Abstract

Internet of Things (IoT) is receiving a great attention due to its potential strength and ability to be integrated into any complex systems and it is becoming a great tool to acquire data from particular environment to the cloud. Data that are acquired from Wireless Sensor Nodes(WSN) could be predicted using Artificial Neural Network(ANN) models. One of the use case fields of IoT is smart agriculture and there are still issues on developing low cost and power efficient WSN using advanced radio technologies for short and long-range applications and implementation of prediction tools. This is the reason why the target of this thesis is to develop a low cost and power efficient WSN and IoT based control system and analyze the best predictive model for such systems. With this purpose, we developed BLESensor node for short-range IoT applications and Internet of Plant(IoP) for long distance smart agriculture applications. A non-linear prediction model is developed in order to forecast acquired data from sensor nodes. BLESensor node Experimental test results reveal that newly developed BLESensor node has a good impact on the improved lifetime and applications could possibly make this emerging technological area more useful. The Android software has been tested on Samsung Galaxy SM-T311, running Android 4.4.2 and it works without any issues and it is supposed to work on all other Android devices equipped with BLE. The working temperature range of the BLESensor node is supposed to work goes from -20 °C to 70 °C due to battery temperature limits. The system has been tested in the climatic chamber (Challenge 250 from Angelantoni) present at the Neuronica Lab, which allowed the sensor to be software calibrated. Several measurements have been proven that each node offers an uncertainty of 1.2 °C for temperature. These values are acceptable for the type of application for which they are intended. The power consumption has been measured directly from scope analysis and simulating the code step by step and calculations resulted that the lifetime of the node lasts for a month. Considering a normal use of these sensors with a reasonable sampling time the lifetime could be increased. IoP node IoP node is a prototype device that works with WiFi protocol and collects temperature, humidity and soil moisture data of plants to the cloud. For IoP node, we have implemented a firmware, tested a prototype device and designed the PCB in OrCAD software and generated a Gerber file and developed an android application. Prediction model Comparisonofthreenon-linearmodelswithOakdatasetresultedinbetterperformance of NNARX model and we used NNARX model to predict 10 days step ahead maximum and minimum temperature and described the results of performances. The performance given by trained models in terms of Mean Square Error (MSE) for maximum temperature prediction provided an error of 0.8826 on unseen data for the month of September. Similarly, the performance of model predicting minimum temperature was tested and it resulted in an error value of 0.944. In conclusion, this work must be intended only as a proof-of-concept, although, the developed BLESensor system, IoP prototype device and predictive models showed expected optimum results, both in terms of functionalities and usability.
9-gen-2018
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2697287
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo