Quality assurance is supposed to guarantee that quality requirements for products are met. In practice this happens by rigorous specifications which are however based on the subjective judgement of experts. This traditional form of quality assurance is currently being accompanied by the idea of predictive quality which refers to the prediction of future developments concerning product quality based on data. For example, data collected in vehicles can be used to foresee future defects due to component overload. However this approach only targets the symptoms, not the causes of defects originating in the design of a product itself.
In our opinion, quality assurance has to be preventive, predictive and prescriptive and the same time. Product quality has to be taken care of in advance but based on data which allows foresights. Instead of time consuming manual analyses which are expensive and subjective, we employ automated machine learning technologies which do not only point at potential problems but which offer direct assistance in resolving causes.
What is "Predictive Quality"?
Predictive Quality refers to the prediction of qantities relevant for quality based on data collected during the actual use of a product. For example data collected using telemetry in cars can offer valuable clues to if and when individual components have to be replaced before actual defects occur.
What is "Prescriptive Quality"?
Potential quality problems do not only have to be detected but also resolved. To achieve this artificial intelligence can assist the person responsible for a product by pointing out possible solutions based on data.
What is "Preventive Quality"?
An ideal quality assurance does not only avoid that flawed products reach the customers – it guarantees in advance that such products are not even produced in the first place by identifying bad decisions early on.
Automated Data Analysis
Structured and unstructured enterprise data are analyzed automatically to uncover relations between product features and failures.
Preventive Defect Detection
Product design and conception are continously monitored by checking the current development status in the background for potential problems. In case of high defect probabilities the person in charge receives a notification.
Assisted Defect Avoidance
To avoid actual defects an assistance system suggests alternative product configurations which are similar but bear fewer risks.
Increased Product Quality
Unfortunate decisions are prevented before they are made. Both customer and company benefit from the resulting increased product quality.
Icons by https://icons8.com
Stop relying on your good fortune and count on innovative technology instead. Predictive Quality will lower your defect probabilities and improve your odds.
Configure your own piece of clothing and observe how your design choices influence product quality in real time. What are the trade-offs entailed in a particular material mix? Have your products manufactured in India or China? Our tool will tell you immediately.
THESE INDUSTRIAL SECTORS
Our Predictive Quality concept targets companies manufacturing large quantities of products – independent from a particular industrial sector. Its input consists of structured information about the features of individual products on one hand and data from quality checks and returns on the other hand.
The concept is however particularly interesting for creative industries such as the furniture and fashion industries which cannot beneift from predictive quality to date. In such industries products have to be created from scratch every couple of months in order to react to changing trends and continously offer customers something new.
At the same time there is a vast range of product variants and raw materials. In the scope of its current research project Preventive Quality Assurance, the AWSi is demonstrating that the use of articial intelligence can offer predictive analyses even for these companies. More information can be found on the corresponding project homepage.