What is statistical quality control? Give examples

statistical quality control

Verification of compliance with quality standards, especially for industrial products, using statistical methods. In principle, all statistical methods that appear suitable can be used. However, specific methods have been developed for the specific purposes of statistical quality control. For example, various types of control cards are used to monitor ongoing production. In order to check, for example, whether a prescribed mean target value is being adhered to, random samples of size n could be taken from ongoing production at fixed time intervals. If the target value is adhered to in production, the calculated sample mean values ​​will be within the control limits shown on the control card with a given probability. Statistical quality controls are also often carried out after production has been completed or when deliveries are accepted by customers. So-called acceptance test plans are used here, which enable a standardized application of random samples (statistical test procedures). A distinction is made, for example, with acceptance sampling plans for metrically scaled (quantitative) features and so-called attribute test plans for binary features (i.e. features with only two values, such as "good" or "bad"). The orderly compilation of such sampling plans for alternative parameter combinations is referred to as a sampling system. Some of these sampling systems have become industrial standards through national or international standards (e.g. DIN 40080, ISO 2859, MIL-ST 1050). Literature: Masing, W (ed.), Handbuch der Qualitätssicherung, Munich, Vienna 1980. Uhlmann, WZ, Statistical Quality Control, 2nd edition, Stuttgart 1982.

Statistical procedures for the quality assessment of products on the basis of random samples (partial control). With the help of statistical quality control, the correct setting of the production facilities is monitored (process control) and the quality of delivered products is assessed (acceptance control).

• For process control, a few copies are taken from ongoing production in order to determine possible deviations from certain default settings of the machines on the basis of this random sample compared to previous random samples (e.g. the filling weight of a coffee package). It may be necessary to intervene in the production process to readjust machines and systems.

• In the batchwise acceptance inspection, a random sample is used to determine the quality of an entire delivery. If you come across too many faulty or inferior copies, the entire delivery may be rejected. Wrong decisions are possible: - In the event of a randomly bad sample, the entire delivery may be rejected even though the actual extent of the defects would still be acceptable. - If the random sample is good, the extent of the defects will not be apparent and the delivery may be wrongly accepted. The test plan is therefore agreed between the producer and customer as part of the delivery contract.

Previous technical term: Statistical process control | Next technical term: statistical security



Report this article to the editors as incorrect and reserve it for editing