Integrated ANN and Bidirectional Improved PSO for optimization of fertilizer dose on Palawija plants

Abstract

With the rapid advance of Science and Technology, especially in the field of agriculture. One of the most important aspects that are critical in agriculture is fertilizer. Within the application of fertilizer itself, there are many types of fertilizers and a combination of different doses. Whereas palawija is a plant for crop rotation, that is planted after the rice cultivating season. Palawija is also grown in the highlands where rice cannot grow. Fertilizer application can give different impacts for Palawija. This paper will explain that with an Integrated Artificial Neural Network (ANN) and Bidirectional Improved Particle Swarm Optimization (BIPSO) can optimize the fertilizer dose on Palawija plants. The ANN method can be used to determine the effect on the plants arising from fertilizer application. After this, the user can input two of the effects on crops selected for optimization doses of fertilizer using BIPSO. The ANN method proved to be very good at predicting the value using training data and BIPSO is able to optimize the more than one vector thus fastening the process of the system. The smallest MSE value 8.6023E-03 is obtained from the test using 90% training data and 10% test data, iterating 100 times, with the number of hidden neuron at 10, learning rate of 0.6 and momentum of 0.6. The parameter values of BIPSO use standard parameters on Particle Swarm Optimization (PSO). The proposed method give the recommendation that to get the plant dry weight 4.4964 ton/ha and yield 6.99985 ton/ha is needed Urea 0.191 ton/ha or 191 kg/ha, SP36 0.201 ton/ha or 201 kg/ha, KCL 0.288 ton/ha or 288 kg/ha and Biochar 48.3 ton/ha.