​This month (August 2020) marks the second anniversary of our first AI, Automatic Defect Recognition (ADR) system being installed in a production environment.
​
To date we haven’t promoted our ADR system publicly, deciding to concentrate on working with select companies and individuals to fully develop it. However it has formed a natural addition to OverSeer’s framework. OverSeer collects inspection data and performs validation and calculations upon it, notifying stakeholders of results and storing data for further analysis. ADR is just another calculation.

​
This flagship installation at an aerospace metal production facility was installed in August 2018 and utilises many of JetSoft’s software features. We are gathering positive feedback from the site about the quality and productivity benefits it is delivering, including:
​

• Automatic defect and characteristic recognition
• Combining factory wide data with inspection information
• Driving process change
• Generating new insight and enabling predictive maintenance
• Reducing the requirements on downstream processes

In a series of upcoming posts, I will discuss these areas in more detail to explain how our processes deliver these benefits, starting with an insight into our ADR software.
 

A note on ADR
Firstly, this article uses the term ADR, which is an industrial term to mean the automatic detection of a problem within a component. ADR isn’t clearly defined yet within the industry, as explained in the previous post ADR - what’s expected? 

​The installation discussed in this article is an Automatic Indication Detection (AID) system, where inspections are still signed off by a qualified human operator.
 

How it works
JetSoft’s core mission is to provide systems that automatically aggregate and catalogue inspection data into a single database. We understand inspection data, and we gather it into a convenient format to take advantage of it. Our solutions combine quality control information with data from other manufacturing systems. These include details about the processing conditions that the part has gone through, details of the component structure and the customer specifications. Combining this information, and making it available for all, has the huge benefit of keeping quality at the forefront of everyone’s mind. I will talk about this in more detail in a future article.

This collated data is passed through an extensible list of models which are trained to look for particular faults or certain characteristics. Each model outputs a confidence level based on the likelihood that the inspection contains a positive match. Any disagreement between the system and inspector is highlighted to the responsible level 3 or automatically sent for a second opinion. This drastically minimises the potential for misdiagnosis. Additionally, the output of these models is stored in the system and is made available for analysis. 

​ 
Thresholding of output confidence enables the user to select the desired level of Probability of Detection (POD).

Model training is in depth and employs a hybrid approach using a combination of pre-processing, AI machine learning techniques using historic data and deterministic algorithms.

One of the key advantages of this system is its ability to subcategorise defects. For example, this system can differentiate between a lamination crack caused by rolling and a quench crack caused by heat treatment. Detecting this distinction is non-trivial for a human inspector, requiring further analysis

​Similar defects which are caused by different process may look very similar to inspectors, but it is important to differentiate them to target improvements towards the appropriate processes. For inspectors to make this distinction may require additional training and further analysis, whereas the systematic solution is automatic, quick and consistent.

​I mentioned earlier that the system can be configured to identify any characteristics, not just defects. This is important as there are many things we may want to observe and track that could indicate future problems which in time could lead to defects, or just a general reduction in quality. The time and cost investment of training inspectors to manually identify such trends or patterns would not be justified, however by recognising these warning signs early, there is time for mitigating action to be taken to reduce scrap rates in the long run.

 
Key benefits summary
An independent assessment – this system provides an independent assessment alongside the inspector. If there is any difference between the system and the inspector, the responsible level 3 can be alerted or it can be automatically sent for a second opinion. This improves quality and reduces the chance of mistakes or misdiagnosis, without increasing operating costs.

Sub categorisation of defects – this is the big commercial driver behind the software. It is the application of this feature that has seen scrap rates plummet. An inspector’s job is not necessarily to interpret the cause of a defect, but to highlight its existence. By training models to examine for defects with a known cause, we can target our time and efforts on improving process where we will see the most commercial benefit. This creates a feedback loop of high-quality, useful, data from inspection to all other departments.
 
Next time
Quality should be everyone’s responsibility, but this is only achievable if everyone has a clear vision of quality, and their impact upon it. JetSoft is providing the tools that deliver real time statistics which demonstrate and communicate quality to all.

​Next time we’ll look in more detail at how we’re achieving this and how JetSoft’s is helping everyone to take ownership of quality.
 

​​Automatic Defect Recognition (ADR) is a term that is gaining traction within the NDT world. However, neither ADR or the functions such a system should deliver are clearly defined.

​At JetSoft we prefer to split ADR into two stages, using the terminology Automatic Indication Detection (AID) and Automatic Part Sentencing (APS).
 
The problem with ADR
The scope of ADR isn’t clear. It seems, depending upon your source, ADR can range from a system which highlights abnormalities to an inspector, to one which can perform full sentencing (go/no go), eliminating the need for an inspector at all. In practice, most end users anticipate a solution resembling the latter, but this may not be true of the products promoted by equipment vendors.
 
AID and APS
At JetSoft we don’t use the term ADR, due to the lack of clarity on what this means, instead we use the terms:

• Automatic Indication Detection (AID)
• Automatic Part Sentencing (APS)

Defect detection is a two step process. The first step is the discovery of an indication or discontinuity, the second step is evaluating that indication against a specification to determine whether it is a defect. ADR may (or may not – depending upon who you talk to) be considered to be the automation of this two-step process. However, in JetSoft’s experience each stage can be automated separately, indeed it would be difficult not to.

JetSoft refers to the automation of the first step as Automatic Indication Detection (AID) with the results highlighted to an inspector or used as an independent assessment and compared with the manual outcome.
​
The second step, qualifying the indications against defect specifications, JetSoft calls Automatic Part Sentencing (APS). Because once a system is confidently automatically qualifying defects it must logically be able to decide if the part passes or failed.

​Is ADR overhyped?
News about ADR (go/no go) using deep learning and machine vision is rife and in my view racing towards the peak of inflated expectations on the Hype Cycle. 

The Hype Cycle, developed by the technology firm Gartner, details the phases a new technology goes through to reach maturity and widespread adoption. 

The scale of the peak and trough of this graph is exacerbated by the lack of clear definition of ADR. It is not difficult to imagine a situation where customers and suppliers differ on the expected functionality, resulting in a negative view of ADR and its possibilities.

However, ADR will happen. There is a large commercial appetite to reduce the cost of inspection and new advances in computation technologies prove that it is possible technically. But it may take a while to get there and require more investment (time and money) by users than anticipated. The market should be prepared not to expect a single turnkey algorithm that performs ADR on any component for any defect, but, should rather, anticipate solutions that require dedicated configuration, development and training based on what the users is looking for and the component/material type.

We hear a lot about the benefits involved with ADR, but perhaps less so of the realism of implementing and delivering them.
 
What to expect
JetSoft is amongst several companies utilising these new technologies, with working systems, using deep learning and machine learning techniques in place in production environments.
​ 
It is JetSoft’s experience that performing accurate automatic detection of defects takes dedicated time, development, and extensive testing. Furthermore, for an AID system to be effective, additional information beyond inspection data is required. Including component details and upstream process data, both of which are critical.

​It is important to note that in every case the preceding step to achieving an APS must be a fully working, and empirically proven AID system.
 
Contact us
As always, we would love to hear your thoughts on this, and if you have any questions, or would like to know more about what we do and the systems we have developed please get in contact.

We’re offering free Cloud instances of OverSeer and EyeView to help people to work from home.

JetSoft has instigated a work from home policy following the latest UK government recommendations. This advice is designed to stem the spread of the Covid-19 virus to keep infection levels within the capacity of the heath service (to squash the sombrero [curve] as The UK Prime minister put it).
 
It is our collective duty to do what we can during these unprecedented and frankly bizarre times. Whilst it is easy for us to implement home working (we’re a software company, all we need is a laptop and access to caffeinated drinks) this is not true of most people, including in the NDT world. The added pressure of school closures is compounding the problem, and some individuals will be forced to stay home. This got us thinking, is there any way we can help?
 
A major feature of JetSoft’s solutions is a centralized repository for all NDT information, providing access from any device and location. Our software is built using Cloud and web technologies and data is accessed and viewed from a web browser, therefore our tools can be used from any computer. While many NDT tasks require you to be close to the point of inspection, others such as analysis and report writing can be undertaken remotely.
 
We’re offering to setup free instances of our software hosted in the Cloud. Data can be uploaded from the factory and then will be available via a secure portal, for viewing and analysis from any location, plus you’d get access to all the other great tools we have.
 
Designed to be technique and format agnostic, OverSeer accepts and understands lots of different data including:
   Caldata
   .csv
   Matec
   Midas
   Mistras
   DICONDE
   .pdf
   ScanMaster
   Testia
   Tiff
   USL
   .xlsx
 
For a full list, please get in contact.
 
Note: Cloud technologies are not for everyone so we can also setup an on-premises instance, hosted on your company network, which means data doesn’t leave your environment and will be available from home if you’re able to VPN into your network.

What is a PACS?

Picture Archiving and Communication Systems (PACS) provide central storage and access to images from multiple sources across multiple techniques. The Picture in the PACS acronym is a bit of a misnomer as PACS don’t just deal in pictures. They also contain all the data around that picture, including when/by who/how that picture was taken. It contains that critical, contextual information that is written on the back of an old photograph.

 
Where are they used?

PACS are used extensively and ubiquitously in the medical world, enabling a plethora of medical practitioners to offer a smooth patient centric service. PACS act as a hub in the network of collection equipment and viewing terminals, allowing the collectors of data and the evaluators to work at arms length. Medical images can be transferred electronically and so instantaneously to wherever the patient requires them. Medical PACS revolve around the DICOM standard, and are very mature, having been first developed in the 1970s.
 
What are the benefits?

This is great for the medical industry, but why do we need them for industrial NDT? Well, we do but for different reasons, reasons that are becoming more critical and useful.

Unlike the medical industry manufacturing is production, not component centric. In the DICONDE format – the NDT equivalent of DICOM standard – component is analogous for patient, except the component is not normally a repeat customer and the suite of tests in manufacturing are generally prescribed and linear, and diagnosis is normally subject to quantifiable standards.

However, industry can benefit from PACS technology in other ways. As projects become more global (such the countries participating in the F35 project), and companies become more likely to have an off site level 3, stakeholders or critical NDT decision makers are not in the same location as the inspection. In this case a PACS helps to make the flow of data and communication more efficient, reducing the time needed to make a decision thus making manufacturing more efficient.

NDT image data quantity is expanding and becoming more complex, both in terms of volume and size, and quite often management of that data will incorrectly fall on to the NDT department. They have the impossible task of managing continuous data generation across many different proprietary formats from multiple locations and techniques. A good PACS will order this data automatically, make this data manageable, and provide quick and easy search and view functions, reducing the time demands on key NDT personnel allowing them to work more effectively.
 
Where can I get one?

JetSoft’s OverSeer is a PACS solution that specialises in NDT information. OverSeer can automatically read and understand many different formats from multiple suppliers and techniques, reducing resources absorbed with data management. OverSeer is DICONDE compliant and so can both input and output DICONDE information and can act as a DICONDE converter for your existing data.

OverSeer’s database with web interface makes it easy to search and find inspections by their contextual information, for example all the scans done on a particular piece of equipment, by a certain operator or with a particular defect.

OverSeer goes further than a standard PACS and offers Business Intelligence functionality, allowing performance of Statistical Process Control (SPC), trend analysis, and full health check of the NDT department.

By using OverSeer and having NDT information in a central database allows us to perform some new types of analysis, including data fusion, automatic defect recognition, or dataset analysis. An example of this was a defect heatmap that was recently performed for a JetSoft customer, which provided new insight and enlightening results

For more information or to arrange a demonstration please get in contact with JetSoft.

http://www.jetsoft.co.uk/contact.html