Prediction of Water Use Using Backpropagation Neural Network Method and Particle Swarm Optimization

Abstract

Clean water production has not been well considered between the balance of water use by the community and the production of clean water that is in accordance with the needs of the community. Prediction of water use in meeting the daily needs of the community is very necessary in order to be able to produce efficient water. This research can help PDAM Kota in Kalimantan to be able to produce clean water in accordance with the use of clean water by the community. The Backpropagation Neural Network method focuses on the recapitulation of water use by the community. For better prediction results, optimization is done with Particle Swarm Optimization (PSO). It is expected that the results in this study can predict community water use in daily activities. The test results showed that the Prediction results had RMSE of 0.040 with parameters for training cycle 600 values, learning rate 0.1 and momentum 0.2, and neuron size was 3 and in particle swarm optimization population size 8, max.of gene 100, inertia weight value 0.3, the value of local best weight 1.0 and global value of best weight 1.0

Hand image reading approach method to Indonesian Language Signing System (SIBI) using neural network and multi layer perseptron

“…Similarly, the multi-layer perceptron parameters, which include a shorter training cycle of 10, 10 generations, and 4 ensembles, point to a more iterative and ensemble-based approach, potentially aimed at capturing multiple perspectives within the dataset. The interaction of these parameters emphasizes the complex tactics used to maximize each model's performance in the defined experimental contexts [31]- [34]. Specifications of Backpropagation as show in Table 2.…”