Detection of defects in fabric and quality control in modern textile

Project Title

Detection of defects in fabric and quality control in modern textile

Project Area of Specialization Electrical/Electronic EngineeringProject Summary

Pakistan is one of the finest exporters of the fabric in the world. Maximum type of quality checking is done manually. For an Employ, it is not possible to identify defect in large number of pieces.

As this is the era of automation in every phase of industry but in fabric industry, human does the final inspection of the product. The probability of error is too high. Our goal is to solve this problem using Modern technologies. We will design a machine that will do the following checks.

• Firstly, our machine will check Design Variation using Machine Learning on the basis of probability. It will check whether the design is same as per requirement or not.
• Secondly, our machine will go to check for Missing Loop. This would be done using Image processing. This test will check the presence of torn fabric, stained fabric, cluster of yarn on fabric, and missing threads on the fabric.
• Thirdly, we are using Sensors (TCS3200 & DHT11) to check the color of fabric and the humidity level in the fabric. This must be close to standard otherwise it’s not going to work properly and fabric would be considered as defected.

These techniques will be implemented in Raspberry Pi 4. The fabric will be moving on a conveyer belt. We will use a camera, which will capture the image for Design Variation and Missing loop test. If both tests are passed, then we will check the color variation and humidity of the fabric. Then if all the tests are clear. That means the fabric will be considered as fine product. If all these tests are not clear, then it will go to the defected product section. It will be isolated with the help of the robotic arm. This machine provides a lot of benefits to the textile industry.

• It will reduce labor rate
• Energy efficient because it works on 12V DC
• Ease to monitor

Textile contributes 8.5% of the GDP of Pakistan. With this machine our time is saved. It will enhance the quality. Its design is environmental friendly. It consumes less area. On the other hand, raspberry pi is advance microcontroller. It can work in the harsh environment of the industry. Furthermore,

Its accuracy is based on the machine learning algorithm. Our data set is consisting on the 1000 images. We can test its accuracy with evaluation matrix. In case of sensors, we have 100% accuracy as it is working on comparison model. If our color is according to base model than it will procedure otherwise it will declare it as defected fabric.

Project Objectives

• Reduce the number of employees in the Quality Control Process
• Advancement in Automation in Textile Industry
• Introducing an efficient solution for the textile industry
• Reduction of Power consumption and labor cost
• Automatic isolation system in Quality Control Process
• Modern solutions with modern techniques
• Reduction of expenditures and Fast the quality control process
• Robust solution for the changes in design fabric color

Project Implementation Method

Our Project is consisting on Two main parts. These parts are described in a detailed manner.

1. Software part
2.  Hardware part

• Software Side:

• Design Variation Test
  • The main purpose of this test is to check whether the design on the fabric is the same as per requirement or not. This process will be done by using Machine Learning. The main problem is to have a dataset. We created our own dataset by capturing images.
• Missing Loop Test
  • The main purpose of this test is to check four elements on the fabric. One is torn cloth, second is stain on the fabric, third one is missing of yarn and last one is cluster of yarn. This is done by morphological filter operator. If any element is present on the fabric. Its shape is detected by morphological filter
• Color and Humidity Test
  •  Color matching is one of the most amazing task. It will be done by the sensor TCS-3200. It is a color sensor that is easily interfaced with any type of Micro-controller. The Second test for this phase is the humidity test. It will also be carried out using the sensor DHT11. This test ensures the proper humidity in the fabric

• Hardware Part

The hardware part consists of designing of the following three main parts.

1. 1. The Robotic Arm

It will be made up of motors and moment arms that will help it to move in 3D.      The main role of Robotic arm is to isolate the defected fabric.

1. 1. The Conveyor Belt

This conveyor belt will be made of leather which is moving over the pulleys. The motion of these pulleys will be totally in control of the motors.

1. 1. The Main Machine Design

The machine design is a tough task but it will be purely made of aluminum, where the camera will be mounted over the conveyor belt and all other sensors will be attached on the top of the conveyor belt.

Benefits of the Project

The major benefit of our project would be complete automation without involving humans in the quality control process. The accuracy of our machine learning algorithm is 98% along with the sensor part which gives 100% accuracy. We have to save our time, manually, it consumes a lot of time. The nominal core frequency of Raspberry pi 4 is 1.5GHz. In this way, Our microcontroller can do these tasks in seconds.

It is the introduction of automation in the quality control process of the textile industry. Our robotic arm will isolate our defective piece. It means no more labor in the quality control process and it reduces the cash outflows. We are merging three different departments of a quality control process in a machine to save place and time. Color variation, design variation, missing loop, and humidity test is done by a single machine.

In this way, our labor rate decreases, our power consumption is decreases and our accuracy is increases. We can use this machine for fabric, socks, and garments also. Its assembly is short and simple. So, its maintenance is easy. One of the most important things is it is a Robust solution for the changes in design fabric color.

Technical Details of Final Deliverable

Our project is controlling a camera, robotic arm using Raspberry Pi and conveyer belt with Arduino. Our project is consisting on two parts.

1. Software work
2. Hardware work

Software work includes coding for machine learning and image processing. The software part also includes gathering the dataset for the design variation test. The machine learning algorithm we used is a deep neural network that has layers that will help us to classify and find out the design. We will use our dataset for the execution of the code. All programming is done in jupyter notebook of anaconda and googles colab for its better performance.  

Hardware work includes the making of the robotic arm and similarly making of the conveyor belt. The next part is to assembling our conveyer belt with the camera and robotic arm. The main 3D model will be made up of a 1.5 mm gauge aluminum rod that strengthens the whole structure. All components are placed according to the AutoCAD design. There will be an LCD that will be displaying all the work regarding machine working. The power will be consumed as shown.

• Input Voltage: 220V AC
• Operating Voltage:12V DC
• Max input current: 1 amp

Conveyor Belt

• Stepper Motor NEMA17
• Leather sheet 5 feet long
• Arduino Uno
• Conveyor Rollers
• Conveyor shafts
• Motor Controller L-298

Main Module

• Raspberry Pi (Controller)
• Robotic arm (for Isolation)
• U-shape frame
• Pi camera

Sensors

• DHT11 (for humidity)
• TCS-3200 (For color detection)

Miscellaneous

• Voltage Regulator (Power supply for sensor and microcontroller)
• 7 inch LED Display (To show results)

Final Deliverable of the Project Hardware SystemCore Industry ManufacturingOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Good Health and Well-Being for People, Affordable and Clean Energy, Decent Work and Economic Growth, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	42850
Raspberry pi 4 - 4GB	Equipment	1	15000	15000
Raspberry Pi Cam	Equipment	1	3800	3800
Flash Light	Equipment	2	100	200
DC Stepper Motor Nema 17	Equipment	2	750	1500
Aluminum lengths 25 feet	Equipment	25	170	4250
L-298 Motor driver	Equipment	2	200	400
5V voltage regulator	Equipment	2	20	40
DHT11	Equipment	2	180	360
TCS-3200	Equipment	2	550	1100
12V DC power supply	Equipment	1	600	600
Arduino UNO R3	Equipment	1	700	700
7 inch LED display	Equipment	1	10500	10500
Conveyor belt rollers	Equipment	2	700	1400
Conveyor belt shaft	Equipment	1	200	200
Conveyor belt leather 5 feet	Equipment	5	300	1500
Robotic Arm	Equipment	1	1000	1000
Veroboard	Equipment	2	50	100
Connecting Wires	Equipment	1	200	200