Bringing AI to Shop Floor
Published on : Monday 30-11--0001
Artificial Intelligence has started entering our day-to-day lives in various forms. Be it a leading ecommerce company in India developing deeper customer insights using big data to transform online shopping experiences or a leading carrier in India analysing fuel efficiency on ground for cost optimisation analytics or the conversational humanoid chatbot of a leading bank in India to gain deeper understanding of behaviour patterns – AI is now real and happening around us.
IBM recently announced the release of the largest ever AI toolset tailor-made for nine industries and professions – a set of pre-packaged tools in bringing AI to transform customer service, supply chain, marketing, advertising, automotive, agriculture, human resources, etc.
However, the biggest impact that IBM is making is bringing AI to Shop Floor which can transform the way companies operate. There is a lot of buzz across the world around Industry 4.0 and according to an analysis by McKinsey, if Indian companies adopt Industry 4.0 across functions such as manufacturing, supply chain, logistics and procurement, they can enhance their operating profits by 40% at less than 10% of the planned capital expenditure as India ranks lowest in the wage to productivity comparison among its Asian peers.
At IBM, we are partnering with our clients to leverage IoT, advanced analytics and AI to reduce operational costs and increase uptime, bring better products to market faster, and find new business value across three main asset classes: industrial equipment, buildings and facilities.
IBM has brought powerful new AI and Analytics capabilities to Shop Floor including:
1. Production Optimisation
2. Production Quality Insights
3. Equipment Maintenance Assistant
IBM Production Optimisation solutions use machine learning and AI to help achieve production targets. This solution can help manufacturers and operators spot patterns in machine and process data. This helps eliminate availability, quality and performance constraints that contribute to OEE loss and result in lower throughput. IBM is engaging with countless customers across all types of manufacturing to bring our AI-powered approach to the factory floor. Successful use cases include aluminium smelting energy efficiency improvements, injection moulding downtime reductions, steel process quality improvements, automotive body shop predictive downtime, and cement mill optimisation between quality and energy consumption.
IBM Visual Inspection for Quality is a manufacturing quality monitoring and alerting solution that can take in images of in-process and finished products and assemblies and classify them into defect categories. VIQ can help organisations significantly reduce cost and time associated with manual quality inspection processes and achieve greater consistency in defect detection and recognition. VIQ incorporates Deep Learning models from IBM Research which can detect and classify defects into previously trained/learnt defect classes, identifying emerging quality defects quicker. VIQ is available as an enterprise-wide hybrid solution with the capability to score at the edge (on the plant floor) with connection to the analytics centre in the Cloud (hybrid software-as-a-service (SaaS) solution). We are now using Visual Insights for detecting defects in Solar Panel Manufacturing to Gears to Automotive Paint Shops. VIQ now enables quality control supervisors to easily monitor key KPIs of visual inspection processes – number and percentage of defects detected, number of detections requiring human expertise for confirmation, and average defect rate on a production line. Supervisors can also adjust the sampling rate (% of product images captured by camera that are analysed) depending upon the defect trend of the specific production line.
IBM Acoustic Insights is another manufacturing quality monitoring and alerting solution that can take in sound recordings of in-process and finished products and assemblies and detect anomalies in product quality. Sound based inspection is traditionally done by expert engineers who can easily detect a malfunction sound from a machine cycle run. This skill is grown over the length of their career and experience and is not easily transferrable to new inspectors or engineers. The AI solution can be trained on normal and defect cycle sounds and deployed to the Cloud to detect anomalies or defect cycles in incoming sound samples. AI can help organisations significantly reduce cost and time associated with manual quality inspection processes and achieve greater consistency in defect detection and recognition. Acoustic Insights incorporates Machine Learning and Deep Learning models from IBM Research which can detect and classify defects into previously trained/learnt defect types, identifying emerging quality defects quicker. Finished products on the manufacturing line, for e.g. consumer electronics need to be tested through all cycles of operation. It is critical to test these non-destructively, e.g., a rinse cycle in a washing machine needs to be tested when the machine is running. AI uses sound recording of the machine cycles sent to the Cloud where it is quickly classified as an anomalous or normal operation.
IBM IoT Equipment Maintenance Assistant applies cognitive methods to a wide range of unstructured data to identify entities and concepts such as equipment details (model, version, configuration, controller), equipment status/conditions, service technician notes, tests and test results, hypothesised failure, prescribed repair procedures, repair resolution, operational procedures, tooling, expertise, and evidence, and utilise this insight to provide probability-ranked guidance regarding diagnostic and resolution options or next best action recommendations to help pre-empt or resolve correlated failures.
IBM IoT Equipment Maintenance Assistant can isolate impending problems just as they begin to form, group comparable issues across similar assets, and help synthesise asset and repair history data in one place. By harvesting best practices and technical expertise maintenance personnel become more effective and efficient in preventing or remedying asset failure and degradation.
The next big thing in manufacturing will be all about digital transformation and the adoption of AI in manufacturing. Cognitive Manufacturing will digitise and optimise previously inaccessible areas of manufacturing processes and help manufacturers minimise downtime, optimise asset and equipment performance, improve quality and yield from design to support as well as reduce costs by driving efficiency in process, labour and energy.
IBM is uniquely positioned to help businesses in asset-intensive industries as they scale their focus on Industrial IoT and AI knowledge. IBM was recently named a leader in Forrester’s evaluation of the “most significant” Industrial Internet of Things (IoT) Software Platforms, Q3 2018.