Manufacturing

Sandheep had worked extensively in manufacturing to develop Digital Transformation products and have completed digital transformation projects. 

images.jpeg
Improving productivity and accuracy of reports using Computer Vision

Problem : A process in manufacturing supply chain had a long conveyer, through which the processed products came out. Due to the random arrangement of the products on the conveyer, a person had to manually count and enter this data into a database. This costed time and accuracy. Multiple automated approaches were earlier used to count the parts using sensors and other methods, but none of them worked. 

Solution : A computer vision algorithm was developed and implemented on the machine to detected and monitor the parts coming out of the conveyer. The software also updated the counts into SQL databases every hour and updated the productivity reports using REST API's

Impact: The product helped in reducing the human effort needed in the process, thereby bringing savings. It also made the process of counting and reporting accurate with real-time reports. 


Tools used : Python, OpenCV, CustomVision.ai, PyQt5, SQL, Custom Dashboards

Skills : Product Management, Stakeholder management, Computer Vision, Deep Learning models, UI/UX, Databases, Multi platform integration

Invoices
Real time production visibility by automated OEE calculation

Problem : Production reports were usually collected for each shift and the reports were generated on a daily basis. This means that if there are any problems with the process at any production station, it used to be very late when they were identified. 

Solution: The product helped in real time visibility by monitoring the OEE of bottleneck machines. This included real time monitoring of Quality, Performance and Availability of the machine by using real time signals from the machine. This resulted in a real time dashboard which was updated every second.

 

Impact :  Real time dashboard helped in taking immediate response to problems with the machine and real time visibility of production

 
Tools used : Python, SCADA, Siemens PLC, Sensors, SQL, 

Skills : Product Management, UI/UX, Databases, Multi platform integration

Influence of external weather on manufacturing

Problem : It was observed that certain machines in the manufacturing shopfloor behaved differently under varying weather conditions. But the root cause was not identified and quantified. 

Solution: Data logging mechanisms had been implemented in certain machines. The data from these machines were collected for a period of 4 years. The data of various external influencers were scrapped from open source platforms. Later Analytics experiments were conducted and multiple dashboards were created. Reports and Dashboards were created out of this data to derive insights

Impact :  The impact of external weather on production was identified and quantified, on further discussion of the dashboard with different process experts and stakeholders. The findings were used as references for further designs

 
Tools used : Python, Power BI, SCADA, SQL, Tableau

Skills : Product Analytics, Data Mining, Data Analytics, Data Scrapping, Data preprocessing

Reduced downtime with condition based monitoring

Problem : Machines in production has to undergo maintenance after it is used for a specific number of cycles. In absence of accurate counts, maintenance was carried out after a specified period of time ( Weeks or months ) . This usually happens when no production is happening, thereby does not interrupt production. However, if the machine had been heavily used, it will result in machine breakdown during production time hence incurring huge costs. 

Solution: Sensors collected information about machine usage and these were integrated into a SCADA system which helped to collect real time information. This data was further connected to a maintenance platform using REST API's which generated service requests and allotted them to maintenance team whenever the machine has operated for a fixed number of cycles

Impact : It helped to prevent machine breakdown due to improper maintenance thereby saving breakdown costs


Tools used : Python, SCADA, SQL, REST API

Skills : Design thinking

Reduced downtime with condition based monitoring

Problem : Machines in production has to undergo maintenance after it is used for a specific number of cycles. In absence of accurate counts, maintenance was carried out after a specified period of time ( Weeks or months ) . This usually happens when no production is happening, thereby does not interrupt production. However, if the machine had been heavily used, it will result in machine breakdown during production time hence incurring huge costs. 

Solution: Sensors collected information about machine usage and these were integrated into a SCADA system which helped to collect real time information. This data was further connected to a maintenance platform using REST API's which generated service requests and allotted them to maintenance team whenever the machine has operated for a fixed number of cycles

Impact : It helped to prevent machine breakdown due to improper maintenance thereby saving breakdown costs


Tools used : Python, SCADA, SQL, REST API

Skills : Design thinking