PDF(Pubmed)
Abstract：
Eutrophication is the excessive growth of algae in water bodies that causes biodiversity loss, reducing water quality and attractiveness to people. This is an important problem in water bodies. In this paper, we propose a low-cost sensor to monitor eutrophication in concentrations between 0 to 200 mg/L and in different mixtures of sediment and algae (0, 20, 40, 60, 80, and 100% algae, the rest are sediment). We use two light sources (infrared and RGB LED) and two photoreceptors at 90° and 180° of the light sources. The system has a microcontroller (M5stacks) that powers the light sources and obtains the signal received by the photoreceptors. In addition, the microcontroller is responsible for sending information and generating alerts. Our results show that the use of infrared light at 90° can determine the turbidity with an error of 7.45% in NTU readings higher than 2.73 NTUs, and the use of infrared light at 180° can measure the solid concentration with an error of 11.40%. According to the determination of the % of algae, the use of a neural network has a precision of 89.3% in the classification, and the determination of the mg/L of algae in water has an error of 17.95%.
摘要：
富营养化是水体中藻类的过度生长，导致生物多样性丧失，降低水质和对人的吸引力。这是水体中的一个重要问题。在本文中,我们提出了一种低成本传感器，用于监测浓度在0至200mg/L之间以及沉积物和藻类的不同混合物（0、20、40、60、80和100％藻类，其余为沉积物)。我们使用两个光源（红外和RGBLED）和两个光源90°和180°的光感受器。该系统具有微控制器(M5stacks),该微控制器为光源供电并获得由光感受器接收的信号。此外,微控制器负责发送信息和生成警报。我们的结果表明，使用90°红外光可以确定浊度，NTU读数高于2.73NTU的误差为7.45％，用180°红外光可以测量固体浓度，误差为11.40%。根据藻类百分比的测定，使用神经网络的分类精度为89.3%，水中藻类mg/L的测定误差为17.95%。
