Visual inspection

Visual inspection
Primary topic
Related topics
Methods and techniques

Visual inspection was the first method used for predictive maintenance.Until the Industrial Revolution there was not really any methods used for identifying potential risks or failures, which later evolved to a broad subject of a ‘predictive maintenance’. The first one, that was introduced, was visual inspection and it consisted of a maintenance technicians regular routine. This basically involved daily “walkdowns” around manufacturing systems and it served its purpose to find problems that could affect the product quality or production costs. Even though visual inspection is not a very modern way of monitoring around the workplace, it is still used and should be included in the maintenance management programs[1].

Categories of visual inspection[edit]

Almost all the goods produced in the industry are visually inspected with some intensity. Previously, this was the duty of a highly trained personnel or regular employees, especially designated for the task, however since 1970s, they have been slowly replaced by the automated visual inspection systems. Those have escalated to be more and more popular as well as efficient in the industry and other sectors, and have been rapidly growing since. To visually inspect is simply to inspect with the use of sight. It should play at least two roles – firstly in ensuring the quality factors of final or semi-final products, secondly it should provide detection and analyzation of the variations in the production process. Depending on different approaches, visual inspection can be divided into categories as follows[2]:

  1. Product completion: knowledge to recognize the objects, patterns:
    1. assembly site: are there all the components present?
    2. during the charging: is there any space left in the package?
    3. assembly of circuit boards: are all the components placed in position?
  2. Accuracy of both positions and orientation:
    1. assembly site: is the place and orientation correct for all the components?
  3. Constancy of all the factors: shapes, angles, dimensions of the components and tools
    1. are all the measurements determined?
    2. are all the essential specifications and an error margins agreements fulfilled?
  4. Condition of the surfaces, textures
    1. is the surface condition as modeled
    2. is the surface indiscrete and is it within the specific tolerance limits?
  5. Additional factors
    1. color
    2. surface reflection behavior
  6. Identification of materials
    1. automated sorting of different synthetic materials
    2. detacting and removing impurities from bulk goods streams
  7. Defects: Both aesthetic and technical defects:
    1. detection of varnish runs on varnished Surface
    2. detection of scrotchrs on optical components

Even though the visual system of a live person is more effective than the competitive performance of a machine, it still can be flawed with some impactful drawbacks. Some factors which can describe manual visual inspection from the negative perspective are[3]:

  • monotony
  • subjectivity
  • cost
  • lack of good reproductability
  • slowness
  • high cost of keeping a detailed documentation
  • physical labor

Classification of visual inspection techniques[edit]

Visual inspection can be used most efficiently in a rapid, noncontact survey around specific sections of the workplace and presents the most accurate extent of visual damage when followed by a sophisticated methods. It is very important and simple at the same time to identify gross defects, discoloration and external corrosion. Its efficiency can be enhanced by the use of some, specifically fit for the role, tools such as a magnifying glass, light source and a borescope[4]. It is possible to classify visual inspections techniques in two categories. The first one is about the comparison of the actual images of a brand-new, defect-free patterns and the actual, given ones. The latter is relating to theoretical decisions, which were made when verifying a set of characteristics of a given product and those of an ideal one.The mentioned categories are [5]:

  • Comparison with a pattern

The first technique mentioned is all about a very detailed comparison between the two images. One of them being the inspected product and the other one the ideal one. It is also essential to possess the visual model of the second one in order to conduct the study. When the multilevel images are taken into account, it is important to carefully develop the patterns in order to assure the detection of every small defects and that they are not connected with the inseparable variability of the visual factors of the product nor with the measuring operations.

  • Theoretic inspection

These types of methods evolve around the adequate factors of the ideal product. This set of factors consists of a many image characteristics. Among those are areas, shapes, moments, textures, histograms, perimeters. It is vital to come up with a classifier in order to decide on the range of defect and general quality of the inspected product.

Footnotes[edit]

  1. Mobley K.R. (2002)., An Introduction to Predictive Maintenance, Elsevier, p. 111
  2. Beyerer J. Puente F.L. Frese Ch. (2015)., Machine Vision: Automated Visual Inspection: Theory, Practice and Applications, Springer, p. 3-4
  3. Beyerer J. Puente F.L. Frese Ch. (2015)., Machine Vision: Automated Visual Inspection: Theory, Practice and Applications, Springer, p. 3-4
  4. Brown R. (1999)., Handbook of Polymer Testing: Physical Methods, CRC Press, p.781
  5. Denket M.A. (2006)., Frontiers in Robotics Research, Nova Publishers, p.3

References[edit]

Author: Katarzyna Maziarka