Water quality analysis system using labeling of water quality data and learning of artificial neural networks

摘要

The present invention relates to a water quality analysis system using labeling of water quality data that can accurately analyze water quality by minimizing the error between labeling of water quality data and an output layer of an artificial neural network and learning of an artificial neural network, ) is installed in a plurality of the stored water based on the water depth of the stored water (1) inside the stored storage tank (10), the sensor unit 100 for measuring the water quality of the stored water (1) according to the water depth with a preset value; A plurality of matrices are generated including a graph of the numerical value of the stored water 1 measured from the sensor unit 100 and the water depth of the stored water 1, and based on the shape of the graph, the stored water 1 ) labeling unit 200 for labeling 210 as good or bad water quality; A convolution layer 310, a pooling layer 320, and an output layer 330 are derived, respectively, and a bias is added to the sumproduct function-based weight to the matrix generated from the labeling unit 200, and a sigmoid ( sigmoid) function to derive the water quality of the stored water 1 to the output layer 330, and an artificial neural network unit 300 for determining the matrix as good or bad according to the output layer 330; and the weight and bias of the convolution layer 310 to derive the error between the labeling 210 and the output layer 330, and the weight and bias given to the pooling layer 320 to minimize the sum of the error. It may include; a comparison operation unit 400 derived by the gradient descent method or the back propagation method.