Water, the most precious resource on mother Earth, overflowing 70% of Earth’s area in approximate. It can be in the phase of solid, liquid, or gas. Water playing an important role as the most primary source for all living organisms, as its quality directly affects the circumstance of every creature existing worldwide. Therefore, water quality is compulsory to be maintained non-stop. Simply speaking, the quality of water is often measured via several approaches, mainly in chemical, physical or biological parameters, as they are the main elements behind all living creatures on the planet.
Developers spare no effort to create or ameliorate existing technologies, to maximize the efficiency of water quality’s measurement. However, there are still some shortcomings existing in the commonly used technique. For example, the most conventional and accurate method to gauge the quality of water belongs to a laboratory test. However, it is costly, time-consuming, and skill dependent, such that there is a ton of affiliated progress to be carried out before the test, including the collection and delivery of the test sample. Moreover, carrying such complex activities suffering high risk to corrupt the samples, resulting in an untrustworthy test result. The left portion of Figure 1 illustrating the real-time image of water quality acquired during a laboratory test. As visible from the image, it is difficult to differentiate the color pH value by the naked eye. Consequently, there is room for improvement to enhance the efficiency of water quality testing.
To overcome the shortcoming brought from conventional techniques of water quality testing, a water quality classification device is proposed, as visible from the right portion of Figure 1. The device contains three (3) main components, namely NIR spectrum LED, IoT database, and a machine learning model. Each component having its unique function in sequence to classify the water quality efficiently.
First and the foremost, pH value and concentration of ammonia in the water sample are detected using NIR (Shortwave Rear Infrared) reflectance spectroscopy as a core technique. A NIR reflectance spectroscopy can be defined as an emitter ejecting near infrared region of the electromagnetic spectrum, ranged between 700 to 2500 nanometers. The emitted ray travel across the water sample towards a light-detecting sensor. With measuring the change of light intensity scattered off and through the sample, NIR reflectance spectra manage to classify the quality of the collected water sample within a short time.
Figure 2 illustrates the rough flow diagram of Internet of Things (IoT) system for this device. First, data is collected using NIR spectroscopy and further transmitted into a database online. Then the machine learning will take place to classify the water quality. The result is then displayed on the panel for the user’s information. All the data received will be stored into the database for future used, along with the classified result.
The battery installed in this device is Li-Po rechargeable battery. The portable and rechargeable characteristic of Li-Po battery enabling the device higher sustainability. Besides, the LED used in this device providing function to switch the light intensity by manipulating the variable resistance manually. With such convenient function, the user is free to control the intensity for a specific purpose. The OLED display panel is installed to display the classification result regarding water quality. Lastly, the most user-friendly feature throughout the whole device belongs to a one-click button. The whole classification process of water quality is executed with just a single click.
For deeper information, the transmission is applied to classify the water quality. Transmission is applicable with materials or liquids which are transparent, or in another term, can be penetrated by light. The wavelength after passing through the samples are measured by a light-detecting sensor.
The classification model of the water quality used in the can be called an Artificial Neural Network machine learning. Artificial Neural Network extracting ideas from neuron cell existing in the brain, which playing an important role in learning and analyzing data by itself. A framework is better to describe the behavior of a neural network rather than an algorithm, such that it manages to handle and process complex input data. The device learns from instance to instance with huge data updates from the cloud system.
In addition, the IoT system acts a well-built center for data storage, cloud machine learning and data processing, with shareable properties. It carries a brilliant feature, such that all the data transferred into the cloud and can be monitor and accessed by multi-user. Users with permission are free to share and collect the data with each other.
In conclusion, there are various advantages found behind this device. As compared with conventional techniques of water quality test, this device is able to obtain a more refined result with relatively shorter time and simpler procedure. Besides, it is small in size, portable, user-friendly and safe. All this unique and irreplaceable feature not only replacing the common shortcoming from the conventional method but also enhance the efficiency of water quality test towards a higher level.