Real-Time Machine Learning- Model for Smart Agriculture

                           Prof.Dr. R. Udayakumar

                         Dean (Computer Science and Information Technology)

          The problem of voltage regulation in smart agriculture units is well-studied. Several approaches exist to support voltage regulation in the agriculture sector. However, the methods consider only the number of motors that share the voltage as the key to regulating the artificial voltage of the agriculture unit. The methods suffer to achieve higher performance in intelligent agriculture. An efficient machine Learning-based Voltage regulation model (MLVRM) is presented in this paper to solve this issue. The method maintains the agriculture trace and computes mean voltage utilization (MVU) at various duty cycles. With information like the of smart motors connected, average voltage utilization of motors, and other features, the method computes MVU value. The method trains the neural network with the features extracted. The network is designed with a number of intermediate layers, and each layer neuron computes the value of MVU according to the features available. The output layer neurons produce a number of MVU values. Based on the MVU values obtained, the method computes the Optimal Regulation Voltage (ORV) for the current input voltage according to the required voltage for the smart motor connected. The proposed model, by providing a more efficient method for voltage regulation in smart agriculture, has the potential to significantly improve the performance of smart agriculture systems, leading to increased productivity and reduced energy consumption. This model is a significant step towards the advancement of smart agriculture, as it addresses a crucial issue in the field and offers a practical solution.

          The economy of any country is purely dependent on the development of agriculture. The growth of agriculture not only supports the country’s economy but also supports the provision of commodities for its citizens. Various products are cultivated yearly, and farmers work on them yearly. The changing lifestyle and culture moved the farmers out of agriculture, and only limited people were involved. Conversely, there is enormous scarcity and demand for farmers when agriculture is performed on huge acres of land. Moving the people throughout the land and monitoring the conditions of the crops and plants are impossible when involving limited people.

          Smart agriculture is the kind of modern agriculture that involves different technologies supporting agriculture. For example, fertilization is performed with the support of drones, which spread the fertilizer throughout the agricultural land for the support of farmers. Similarly, water regulation is the process that must be performed at a specific time and with a particular volume. Several methods are used in literature to accomplish this. For example, the electric motor intended for the water supply would be controlled with intelligent agriculture using sensors attached. The sensors allow the motors to be switched on and off whenever required.  In this process, electricity plays a vital role in running the engine.

The electric motor attached would have high horsepower and require specific high voltage. But in reality, the electricity supplied by the power distribution system would fluctuate and would not support the exact functioning of the motor. This voltage fluctuation would affect the motor’s functioning, and the water regulation would spoil. If such irregular regulation persists, it affects the plant’s growth, crop and yield. This must be considered, and the electrical devices should be provided or regulated with the exact voltage required. To regulate the needed voltage, the voltage received at the motor or the unit must be monitored, and based on that, the additional voltage must be supplied to the device.

          There are many approaches available to perform voltage regulation. Some of the methods consider the incoming voltage and regulate the required voltage. This would introduce poor regulation, and there will be voltage loss when higher voltage is received. A machine learning-based model has been presented in this article to solve this problem. The machine learning-based model monitors the incoming voltage and regulates the exact voltage required by the electrical motor. Doing so reduces the voltage loss, and the over-voltage state is avoided.  To support the problem, a neural network-based model is defined here, which computes the mean voltage utilization (MVU) by various electrical devices in the agriculture unit. Based on that, the method would calculate the optimal regulation voltage the unit must regulate. A detailed approach is sketched out in this section.

          This article presented a novel machine learning-based voltage regulation model that preprocesses the available agriculture trace, removes the noise from the data set, and extracts the features. Extracted features are converted into feature vectors and used to train the neural network. The method considers the number of connected intelligent motors, the average voltage utilization, the number of sensors available, the average voltage use of sensors, and so on. The network has been trained with many intermediate layers; the input layer takes the sample, and the output layer neurons produce mean voltage utilization (MVU). Based on the MUV value obtained, the method computes the Optimal Regulation Voltage (ORV) for the current input voltage according to the required voltage of various devices. The proposed method improves the voltage regulation performance and reduces the voltage loss.