<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Devadharshini on Medium]]></title>
        <description><![CDATA[Stories by Devadharshini on Medium]]></description>
        <link>https://medium.com/@devadharshini0722?source=rss-dc01e8039511------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*yoWxV10RqX0Emjoz</url>
            <title>Stories by Devadharshini on Medium</title>
            <link>https://medium.com/@devadharshini0722?source=rss-dc01e8039511------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 22:13:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@devadharshini0722/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Building an AI Powered Conveyor Belt Monitoring System]]></title>
            <link>https://medium.com/@devadharshini0722/building-an-ai-powered-conveyor-belt-monitoring-system-d5b01e68a023?source=rss-dc01e8039511------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5b01e68a023</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[belt-conveyor]]></category>
            <dc:creator><![CDATA[Devadharshini]]></dc:creator>
            <pubDate>Tue, 29 Sep 2026 17:16:29 GMT</pubDate>
            <atom:updated>2026-09-29T17:16:29.330Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>How was my idea born?</strong></h3><p>Conveyor belts have been the lifeline of mining operations. They keep moving iron ore, 24/7. But what happens when something goes wrong?</p><p>A faulty belt joint, an<strong> increasing temperature reading</strong>, a <strong>vibrating belt</strong> or a <strong>misaligned belt </strong>can quietly lead to a catastrophe. But what if these problems could be detected before they become serious?</p><p>While working on this problem, i posed a simple question: can we monitor the conveyor in such a way that we can get a warning before a catastrophe?</p><p>This became the foundation of my project, BELTGUARD.</p><h3><strong>What is BELTGUARD?</strong></h3><p>BELTGUARD is our prototype of an <strong>AI powered conveyor belt health</strong> <strong>monitoring and predictive maintenance system</strong>. The main idea is to gather data from various sensors, process it in an ESP32, analyse it using AI/ML and display results on a web dashboard.</p><p><strong>In short, the idea is: Sensors → ESP32 → Wi-Fi → Backend → ML Analysis → Dashboard → Alert</strong></p><p>As opposed to waiting for the conveyor belt to fail, system monitors the conveyor belt constantly, so that we can be alerted to a deviation from the norm, as soon as it happens.</p><h3><strong>What we are monitoring?</strong></h3><p>There are multiple parameters that give us information about a conveyor belt’s health. For our project, we monitored parameters like:</p><p><strong>Temperature</strong></p><p><strong>Vibration</strong></p><p><strong>Load</strong></p><p><strong>Speed</strong></p><p><strong>RPM</strong></p><p><strong>Belt position</strong></p><p><strong>Motor current</strong></p><p>By gathering multiple parameters, we can understand the health of our conveyor better.</p><p>For example, a vibration spike, with changes in temperature, could be an indicator of something going wrong.</p><h3><strong>How the system work?</strong></h3><ol><li><strong>Gathering data</strong></li></ol><p>The first step in the process is to gather data from our sensors. The sensor suite gives us data on the operating condition of the conveyor. Essentially, think of these sensors as giving the conveyor a set of senses, so that we can know what the conveyor is experiencing, as it operates. The data is then processed into the <strong>ESP32.</strong></p><p><strong>2. Using an ESP32</strong></p><p>An<strong> ESP32 </strong>is used as a controller. It is good at what it does for our prototype, as it can communicate over <strong>Wi-Fi </strong>to multiple other systems. Once the sensor data has been processed into the <strong>ESP32</strong>, it can be sent over <strong>Wi-Fi</strong> into our application.</p><p><strong>3. Processing the data</strong></p><p>Once the data has been received by the backend (<strong>built using Python and</strong> <strong>Flask),</strong> it is processed for the dashboard.</p><p><strong>4. Where does AI/ML come in?</strong></p><p>My system detects an <strong>abnormal operation of the conveyor</strong> <strong>belt</strong>, so that operators can take preventative measures. It goes beyond simple , this is the part that made us excited about my project. Just displaying sensor data is good, but we wondered — could ML algorithms detect a deviation from the norm, based on multiple parameters?</p><p>A simple <strong>ML mode</strong>l, trained using parameter combinations like <strong>Temperature + Vibration + Load + Speed + RPM,</strong> could be used to predict the likelihood that a given set of parameters belongs to a <strong>‘normal</strong>’ set of parameters.</p><p>Essentially, if an operator inputs <strong>temperature, vibration, load, speed</strong> and <strong>RPM</strong>, the <strong>ML</strong> model could predict if the operation of the conveyor is abnormal.</p><p>This way, I can go from a <strong>descriptive statistics view</strong> <strong>(temperature is too</strong> <strong>high), </strong>to a predictive one (given this set of parameters, the conveyor has an X% chance of failing).</p><h3><strong>Our BELTGUARD dashboard</strong></h3><p>I wanted my dashboard to be intuitive, so i built an application that allows for the operator to see the complete picture of the health of the conveyor. This includes:</p><p><strong>Overall conveyor health</strong></p><p><em>Sensor values</em></p><p><em>Warning flags</em></p><p><em>Risk level</em></p><p><em>Trending values</em></p><p><em>Alerts</em></p><p><strong>Maintenance info</strong></p><p>I wanted to make sure that the operator didn’t have to sift through dozens of sensor readings, to find out that the conveyor is operating fine</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*21PFc_KvRXSrhPg3DQqnlg.png" /></figure><h3><strong>Building the prototype</strong></h3><p>One of the things that made this project exciting, was seeing the combination of the software and the hardware. We built a miniature conveyor setup, to showcase how our sensors and algorithm could work together. This allowed us to think about how the entire stack could work:</p><p><strong>Physical Conveyor → Sensors → ESP32 → Internet → Software → AI/ML analysis → Dashboard</strong></p><p>My conveyor prototype with the monitoring setup</p><p>The prototype is only a small part of what we are doing. It showcases how our proposed monitoring stack could work.</p><h3><strong>The tech behind BELTGUARD</strong></h3><p><strong>BELTGUARD </strong>is a combination of various technologies, rather than a singular technology. The technologies we used, include:</p><h4><strong>Hardware</strong></h4><p><em>ESP32</em></p><p><em>Temperature sensing</em></p><p><em>Vibration sensing</em></p><p><em>Load/current monitoring</em></p><p><em>Speed monitoring</em></p><p><em>Belt position monitoring</em></p><p><em>Thermal imaging</em></p><h4><strong>Software</strong></h4><p><em>Python</em></p><p><em>Flask</em></p><p><em>HTML</em></p><p><em>CSS</em></p><p><em>JavaScript</em></p><p><em>Machine Learning</em></p><p>By combining various components, able to create a software system that monitored a physical system.</p><h3><strong>What did I learn while building it?</strong></h3><p>While it seems simple on paper, there are many intricacies involved in building a system like this. There are considerations for how the sensors <strong>communicate with the ESP32</strong>, how the d<strong>ata gets to the backend</strong>, how this <strong>data is visualised</strong>, <strong>how ML algorithms can be used to predict possible</strong> <strong>failures </strong>and how this prototype could scale to an industrial level.</p><p><strong>Some of the challenges i faced, include:</strong></p><p>Getting consistent sensor readings</p><p>Connecting multiple hardware pieces</p><p>Sending ESP32 data to the backend</p><p>Keeping up with sensor data</p><p>Designing a good dashboard</p><p>Interpreting ML results</p><p>Understanding how ML works with sensor data</p><p>Thinking about how my prototype could scale</p><h3><strong>What makes i think this is a good approach?</strong></h3><p>When it comes to industries, it is common practice to do maintenance checks on equipment, based on certain triggers (a failure has happened) or on a schedule. With continuous monitoring, we are in a position to take advantage of the massive amount of data being generated by the system, to better monitor the system.</p><p>BELTGUARD is an exploration into how <strong>IoT + Real-time Monitoring + Data-analysis + ML</strong> can be combined, to create a system that does more than just show pretty graphs.</p><p>The ultimate purpose of our system is not to be a pretty dashboard, but rather to help operators understand what is happening in order to detect abnormal conditions sooner.</p><h3><strong>What can i improve in the future?</strong></h3><p>Since i am working with a prototype, there is a lot of room for improvement. In the future, i would like to focus on:</p><p>Gathering real-world data</p><p>Using industrial sensors</p><p>Predictive maintenance improvements</p><p>Mobile notifications</p><p>Thermal image analysis improvements</p><p>Maintenance suggestions</p><p>Cloud-based monitoring</p><p>Real-world testing</p><p>Industrial validation</p><p>I believe that real-world data and validation would be essential before deploying such a system in a real mining environment.</p><h3>Final thoughts</h3><p>BELTGUARD was born from a simple question: can i know that something is going wrong with the conveyor before it becomes a problem? While working on our project, we have discovered that trying to solve an industrial problem is not as simple as picking an ML algorithm, throwing a few sensors and calling it a day.</p><p>It is about <strong>combining hardware, software</strong>, data and people, to solve a problem. As such, our current prototype is just the start. I believe that with more real-world data and industrial level sensors, the world of predictive maintenance could take off. For us, BELTGUARD was more than just a project, it was a chance to see how <strong>IoT, AI/ML and Industry 4.0 concepts</strong> could be applied in the real world.</p><p><strong>Build. Monitor. Learn. Predict. — This is the BELTGUARD philosophy.</strong></p><h3>Technologies</h3><p><strong>ESP32 • IoT • Python • Flask • Machine Learning • HTML • CSS • JavaScript • Sensors • Thermal Imaging</strong></p><h3><strong>Tags</strong></h3><p><strong>Artificial Intelligence, Machine Learning, IoT, Predictive Maintenance, Mining, ESP32 Industry 4.0, Conveyor Belt.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5b01e68a023" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>