Final Report Summary - AMI4BLISK (Automated Geometrical Measurment and Visual Inspection for Blisks)
Executive Summary:
The demand for BLISKs (bladed disks) is strongly increasing, presumably by a factor 4 to 6 in the next 7 years. This growth will also be mirrored in the operating expense for measuring and inspection for BLISKs. Also the increasing complexity of aero engine parts, especially those coming from modern 3D aerodynamics and small tolerances, demand an adequate response in the measuring technique for such parts. At the same time, we also expect a deficiency of well trained inspectors for this time period, thus automated procedures for measurement and inspection is mandatory. Since nowadays, aero engine parts are produced at different places and countries, the inspection and measuring of such parts have to be done in an unified way to ensure homogeneity in the evaluation of the parts. The best way to assure this, it is to automatise such processes. Up to now, aero engine parts measurement and inspection procedures are done separately because of the insufficient and/or costly techniques available. First the simple accessible geometric data were acquired, second more specific geometric feature are measured and third a visual and tactile inspection took place for scratches, dip etc. All these steps should be comprised automatically in a measurement – inspection system. [Topic description of JTI-CS-2011-01-SAGE-04-010]
List of requirements:
- CMM (coordinate measuring machine) is the base technology
- concept of automated handling system
- partial automated generation of measuring program
- automated visual inspection
- capability of design and building of complex measurement systems
- time and space constraints
- measurement system compliant to international specifications
AMI4BLISK (Automated geometrical measurement and visual inspection for Blisks) was a project consisting of the three participants Hexagon Metrology GmbH, Hexagon Technology Center GmbH, and Fraunhofer Institute for Industrial Mathematics ITWM. The key innovations of AMI4BLISK is a unique and new Approach for automated BILISK measurement and inspection; an innovative and new image processing algorithms and visual inspection sensor setup and a new sensor for defect characterisation and measurement.
In a first step the geometrical measurement with a coordinate-measuring-machine (CMM) will provide the information of the shape and surface of the BLISK in combination with the CAD model that provides Basic information of the geometry.
In a second step all possible defects need to be found with the use of an intelligent image-processing system. The location and type of the questionable defects will then be qualified and stored automatically.
In a third step the defects will be classified. With the known position of the defects a program to measure all the defects with another CMM is generated automatically. These measurements give a quantitative description of the possible defects. In some cases where a defect may be questionable the inspection by a person of skill will still be necessary.
Project Context and Objectives:
The project is an integrated approach to the measurement of the geometry, automatic identification and characterization of surface defects of a Blisk. The approach consist three steps. Please see below the requirements as well as their relationship and responsibilities:
First step:
The geometrical measurement of the Blisk will be executed automatically by using a CMM. For this purpose adequate measuring software programs are available which provide in combination with the CAD model the basic information of the geometry. For the characterization of the surface property, roughness, a tactile sensor with a wrist will be used. The installed tool changer rack on the machine allows the quick exchange of the clusters needed for the geometric measurement as well as the roughness sensors. The infrastructure for the data flow in the controller firmware as well as in the host software is kept universal. This increases the flexibility faced to the adaption of new roughness sensors and sensor types.
The Blisk, after the geometric measurement has to be transported to the second CMM with the aid of a handling system. This is a robotic system on a guide rail fixed on the top of the frame that loads, unloads the CMMs and turns over the Blisks. The modular structure of it allows the very easy extension of the “measurement and inspection cell” with further CMMs. A superior software package controls the interaction between the CMMs and the handling system.
Second step:
The visual inspection system was installed on the second CMM and is built up of several camera and illumination units. These are responsible for acquirement of the images. The hardware set up depends on the blisk type and size. Image processing algorithms detect the questionable defects and localize them in 3D coordinates. This information is a pre-condition for the automatic generation of the measuring program taken in the third stage for characterization of the defects. For better understanding a catalogue with the expected range of defects was created. Several Use Cases helped to have a clear understanding about the image acquisition platform and processing algorithms. Pre-processing steps calibrate the image system to correct the optical distortion of the lenses. With support of blisk manufacturer it was agreed to produce a Blisk with artificial defects that allowed to develop the right hardware set up of the cameras and illumination units as well as to train the geometry and roughness measurements visual inspection and defect characterization handling system algorithms for a higher probability for detection. An easy extension of the algorithms is given by machine learning concept. The camera image delivers already 2D coordinates of the defect. To estimate the depth of every pixel in the 2D camera image is needed to have the blisk triangle mesh derived from the original CAD model and the position and orientation of the camera. The solution for the estimation of the z coordinate is the raytracing method: tracking of rays from the camera back to the 3D scene. Using the CAD data of the Blisk the visual inspection system extracts the 3D coordinates of the corners of the bounding box enclosing each defect. These coordinates are transmitted to the application software and stored. The algorithm determines the shape of the defect as well and colors the edges of the bounding box in red or blue/green.
Third step:
Within these coordinates of the bounding box the application software generates the scanning path for the optical sensor. With this approach a “real” characterization of every potential defect took place. First of all this information will be used to remove so called false-positive defects from the list of real defects. Falsepositives are false indications of the visual algorithms (the image processing module).
The final output of the defect measurement contains several values, which identifies the defect. Those values are the maximum depth, the maximum length of the defect as well as the location within the blisk. For those values, the standard reporting methods are used to create a report.
Project Results:
The following results have been obtained:
Geometry and roughness measurements:
The automatic measurement of the serial parts were implemented and integrated with success into a CMM (Coordinate Measuring Machine) platform. The measurements include all relevant features incl. Special geometries of the blisk. The measurement reports are generated with regards to the established standards.First measurements with the tactile roughness sensor in the CMM provided good results on a standardized roughness gauge as well as on some Blisk segments. Due to the housing dimension of the standard sensor not all locations on the Blisk are accessible. Because of this an alternative option was chosen by using a completely different technology: to use an optical sensor. Due to the small dimension of it was possible to solve the accessibility constrain of the existing sensor and to reach all relevant reas on the blisk surface. Measurements were comparable with the results of the tactile one executed on several sample blisks.
Concept of handling System:
The aim of this work package was the development of an automated handling system concept with regards to BLISK application using standard components:
- First step: draft alternative handling concepts
- Second step: elaborate one concept in detail (up to “ready to order”)
Zero damage of the BLISK during handling process had to be considered in each step of elaboration. The Blisk application includes not only the load and unload process but also the turn over operation of it. Several layouts of the intended CMMs in the existing inspection room were elaborated. The favored layout was finalized concerning to costs and type of the handling system which was agreed.
Image Analysis Algorithm:
Within this work package was developed:
- the algorithm for pre-processing which includes the calibration and denoising of the camera responsible for taking images of the blisk,
- the algorithm for detecting of the possible defect candidates by usage of adaptive filters.
Afterwards the candidates are classified according to various geometric and statistical features as well as preprocessing information into false positives and possible defects. The algorithm delivers for each defect candidate in the such reduced set of overall found candidates the
coordinates of its bounding box in 2D (x,y) in the image - the method of the coordinate transformation which maps the defect position in the image (the U and V pixel position) to the real defect position in CMM coordinates (X,Y,Z) on the blisk.
System software development:
Because the platform where the visual inspection algorithms run is a separate system a communication protocol have been developed between VIS (visual inspection system) controller and the host computer. Over this protocol are transmitted not only the set-up of the camera, illumination needed for image recording but also the extracted coordinates of the potential defects to the application software.
Path planning, measurement program:
The detection and measuring path was defined as follow:
- the camera system records all positions of the blisk surface
- intelligent image processing algorithms analyze the images during/after the recording phase
- location and supposed types of questionable defects are extracted and transmitted to the measuring software
- on the basis of the transmitted coordinates ( the edges of the bounding box) generates the measuring software the scanning path for the characterization of the detected defects
- the optical probe follows the path and scans the possible defects
- evaluation of the point clouds gives a precise description of the defect length, depth and orientation.
Test measurements for special geometry and defect analysis on real blisk:
The sensor used for these measurements is an enhanced optical sensor which is based on frequencymodulated, interferometric optical distance measurement. It exerts no physical impact on a part, delivering force-free measurement without a loss of accuracy. Adequate programs were generated to measure characteristic features on the blisk – special geometries. Within the coordinates which the potential defect is enclosed the application software generates the scanning path for the optical sensor. With this approach a “real” characterization of every potential defect takes place.
Integration of components into the CMM platform:
The visual defect classification and characterization system was integrated into the CMM. The components of this system are listed below:
1. CMM:
a. rotary Table
b. wrist
c. tactile and optical Sensor
d. application software
2. Visual Inspection System:
a. One camera and 3 illumination units
b. SW packages with image processing algorithms for defect detection: for every illumination setup and defect type an algorithm is needed
Potential Impact:
In the framework of the present Project, Hexagon Metrology GmbH will generate a sellable system for automated BLISK measurement. This measurement system based on a coordinate measuring machine includes:
- the geometrical measurement of the complete blisk surface by using the tactile sensor
- the roughness determination of strategic surface areas on the blisk and
- the measurement of special geometries such as radii and edges by using the optical sensor developed and tested during the presented project
The automated visual inspection system needs further development to be marketable and ready for serial production. These activities are scheduled and will be traced at Hexagon Metrology GmbH. The outcome of it will be a system that delivers more reliable results than resently delivered by human inspectors. Therefore the system finally enables manufacturers to produce more reliable engines that finally contribute to more security in the equipped airplanes. Moreover, it will also contribute to a longer lifetime of the engines and therefore a more efficient use of natural resources. Finally the new engine generation will also show significantly lower noise emission.
List of Websites:
Fraunhofer Gesellschaft zur Förderung der angewandten Forschung e.V.
Fraunhofer-Platz 1
67663 Kaiserslautern
Germany
Tel.: +49 631 31600-0
Fax: +49 631 31600-109
Hexagon Metrology GmbH
Siegmund-Hiepe-Strasse 2-12
35578 Wetzlar
Deutschland
Tel.: +49 6441 207-207
Fax: +49 6441 207-122
Hexagon Technology Center GmbH
Heinrich-Wild-Strasse 201
9435 Heerbrugg
Switzerland
Tel.: + 41 71 7274728
Fax.: + 41 71 7274674
The demand for BLISKs (bladed disks) is strongly increasing, presumably by a factor 4 to 6 in the next 7 years. This growth will also be mirrored in the operating expense for measuring and inspection for BLISKs. Also the increasing complexity of aero engine parts, especially those coming from modern 3D aerodynamics and small tolerances, demand an adequate response in the measuring technique for such parts. At the same time, we also expect a deficiency of well trained inspectors for this time period, thus automated procedures for measurement and inspection is mandatory. Since nowadays, aero engine parts are produced at different places and countries, the inspection and measuring of such parts have to be done in an unified way to ensure homogeneity in the evaluation of the parts. The best way to assure this, it is to automatise such processes. Up to now, aero engine parts measurement and inspection procedures are done separately because of the insufficient and/or costly techniques available. First the simple accessible geometric data were acquired, second more specific geometric feature are measured and third a visual and tactile inspection took place for scratches, dip etc. All these steps should be comprised automatically in a measurement – inspection system. [Topic description of JTI-CS-2011-01-SAGE-04-010]
List of requirements:
- CMM (coordinate measuring machine) is the base technology
- concept of automated handling system
- partial automated generation of measuring program
- automated visual inspection
- capability of design and building of complex measurement systems
- time and space constraints
- measurement system compliant to international specifications
AMI4BLISK (Automated geometrical measurement and visual inspection for Blisks) was a project consisting of the three participants Hexagon Metrology GmbH, Hexagon Technology Center GmbH, and Fraunhofer Institute for Industrial Mathematics ITWM. The key innovations of AMI4BLISK is a unique and new Approach for automated BILISK measurement and inspection; an innovative and new image processing algorithms and visual inspection sensor setup and a new sensor for defect characterisation and measurement.
In a first step the geometrical measurement with a coordinate-measuring-machine (CMM) will provide the information of the shape and surface of the BLISK in combination with the CAD model that provides Basic information of the geometry.
In a second step all possible defects need to be found with the use of an intelligent image-processing system. The location and type of the questionable defects will then be qualified and stored automatically.
In a third step the defects will be classified. With the known position of the defects a program to measure all the defects with another CMM is generated automatically. These measurements give a quantitative description of the possible defects. In some cases where a defect may be questionable the inspection by a person of skill will still be necessary.
Project Context and Objectives:
The project is an integrated approach to the measurement of the geometry, automatic identification and characterization of surface defects of a Blisk. The approach consist three steps. Please see below the requirements as well as their relationship and responsibilities:
First step:
The geometrical measurement of the Blisk will be executed automatically by using a CMM. For this purpose adequate measuring software programs are available which provide in combination with the CAD model the basic information of the geometry. For the characterization of the surface property, roughness, a tactile sensor with a wrist will be used. The installed tool changer rack on the machine allows the quick exchange of the clusters needed for the geometric measurement as well as the roughness sensors. The infrastructure for the data flow in the controller firmware as well as in the host software is kept universal. This increases the flexibility faced to the adaption of new roughness sensors and sensor types.
The Blisk, after the geometric measurement has to be transported to the second CMM with the aid of a handling system. This is a robotic system on a guide rail fixed on the top of the frame that loads, unloads the CMMs and turns over the Blisks. The modular structure of it allows the very easy extension of the “measurement and inspection cell” with further CMMs. A superior software package controls the interaction between the CMMs and the handling system.
Second step:
The visual inspection system was installed on the second CMM and is built up of several camera and illumination units. These are responsible for acquirement of the images. The hardware set up depends on the blisk type and size. Image processing algorithms detect the questionable defects and localize them in 3D coordinates. This information is a pre-condition for the automatic generation of the measuring program taken in the third stage for characterization of the defects. For better understanding a catalogue with the expected range of defects was created. Several Use Cases helped to have a clear understanding about the image acquisition platform and processing algorithms. Pre-processing steps calibrate the image system to correct the optical distortion of the lenses. With support of blisk manufacturer it was agreed to produce a Blisk with artificial defects that allowed to develop the right hardware set up of the cameras and illumination units as well as to train the geometry and roughness measurements visual inspection and defect characterization handling system algorithms for a higher probability for detection. An easy extension of the algorithms is given by machine learning concept. The camera image delivers already 2D coordinates of the defect. To estimate the depth of every pixel in the 2D camera image is needed to have the blisk triangle mesh derived from the original CAD model and the position and orientation of the camera. The solution for the estimation of the z coordinate is the raytracing method: tracking of rays from the camera back to the 3D scene. Using the CAD data of the Blisk the visual inspection system extracts the 3D coordinates of the corners of the bounding box enclosing each defect. These coordinates are transmitted to the application software and stored. The algorithm determines the shape of the defect as well and colors the edges of the bounding box in red or blue/green.
Third step:
Within these coordinates of the bounding box the application software generates the scanning path for the optical sensor. With this approach a “real” characterization of every potential defect took place. First of all this information will be used to remove so called false-positive defects from the list of real defects. Falsepositives are false indications of the visual algorithms (the image processing module).
The final output of the defect measurement contains several values, which identifies the defect. Those values are the maximum depth, the maximum length of the defect as well as the location within the blisk. For those values, the standard reporting methods are used to create a report.
Project Results:
The following results have been obtained:
Geometry and roughness measurements:
The automatic measurement of the serial parts were implemented and integrated with success into a CMM (Coordinate Measuring Machine) platform. The measurements include all relevant features incl. Special geometries of the blisk. The measurement reports are generated with regards to the established standards.First measurements with the tactile roughness sensor in the CMM provided good results on a standardized roughness gauge as well as on some Blisk segments. Due to the housing dimension of the standard sensor not all locations on the Blisk are accessible. Because of this an alternative option was chosen by using a completely different technology: to use an optical sensor. Due to the small dimension of it was possible to solve the accessibility constrain of the existing sensor and to reach all relevant reas on the blisk surface. Measurements were comparable with the results of the tactile one executed on several sample blisks.
Concept of handling System:
The aim of this work package was the development of an automated handling system concept with regards to BLISK application using standard components:
- First step: draft alternative handling concepts
- Second step: elaborate one concept in detail (up to “ready to order”)
Zero damage of the BLISK during handling process had to be considered in each step of elaboration. The Blisk application includes not only the load and unload process but also the turn over operation of it. Several layouts of the intended CMMs in the existing inspection room were elaborated. The favored layout was finalized concerning to costs and type of the handling system which was agreed.
Image Analysis Algorithm:
Within this work package was developed:
- the algorithm for pre-processing which includes the calibration and denoising of the camera responsible for taking images of the blisk,
- the algorithm for detecting of the possible defect candidates by usage of adaptive filters.
Afterwards the candidates are classified according to various geometric and statistical features as well as preprocessing information into false positives and possible defects. The algorithm delivers for each defect candidate in the such reduced set of overall found candidates the
coordinates of its bounding box in 2D (x,y) in the image - the method of the coordinate transformation which maps the defect position in the image (the U and V pixel position) to the real defect position in CMM coordinates (X,Y,Z) on the blisk.
System software development:
Because the platform where the visual inspection algorithms run is a separate system a communication protocol have been developed between VIS (visual inspection system) controller and the host computer. Over this protocol are transmitted not only the set-up of the camera, illumination needed for image recording but also the extracted coordinates of the potential defects to the application software.
Path planning, measurement program:
The detection and measuring path was defined as follow:
- the camera system records all positions of the blisk surface
- intelligent image processing algorithms analyze the images during/after the recording phase
- location and supposed types of questionable defects are extracted and transmitted to the measuring software
- on the basis of the transmitted coordinates ( the edges of the bounding box) generates the measuring software the scanning path for the characterization of the detected defects
- the optical probe follows the path and scans the possible defects
- evaluation of the point clouds gives a precise description of the defect length, depth and orientation.
Test measurements for special geometry and defect analysis on real blisk:
The sensor used for these measurements is an enhanced optical sensor which is based on frequencymodulated, interferometric optical distance measurement. It exerts no physical impact on a part, delivering force-free measurement without a loss of accuracy. Adequate programs were generated to measure characteristic features on the blisk – special geometries. Within the coordinates which the potential defect is enclosed the application software generates the scanning path for the optical sensor. With this approach a “real” characterization of every potential defect takes place.
Integration of components into the CMM platform:
The visual defect classification and characterization system was integrated into the CMM. The components of this system are listed below:
1. CMM:
a. rotary Table
b. wrist
c. tactile and optical Sensor
d. application software
2. Visual Inspection System:
a. One camera and 3 illumination units
b. SW packages with image processing algorithms for defect detection: for every illumination setup and defect type an algorithm is needed
Potential Impact:
In the framework of the present Project, Hexagon Metrology GmbH will generate a sellable system for automated BLISK measurement. This measurement system based on a coordinate measuring machine includes:
- the geometrical measurement of the complete blisk surface by using the tactile sensor
- the roughness determination of strategic surface areas on the blisk and
- the measurement of special geometries such as radii and edges by using the optical sensor developed and tested during the presented project
The automated visual inspection system needs further development to be marketable and ready for serial production. These activities are scheduled and will be traced at Hexagon Metrology GmbH. The outcome of it will be a system that delivers more reliable results than resently delivered by human inspectors. Therefore the system finally enables manufacturers to produce more reliable engines that finally contribute to more security in the equipped airplanes. Moreover, it will also contribute to a longer lifetime of the engines and therefore a more efficient use of natural resources. Finally the new engine generation will also show significantly lower noise emission.
List of Websites:
Fraunhofer Gesellschaft zur Förderung der angewandten Forschung e.V.
Fraunhofer-Platz 1
67663 Kaiserslautern
Germany
Tel.: +49 631 31600-0
Fax: +49 631 31600-109
Hexagon Metrology GmbH
Siegmund-Hiepe-Strasse 2-12
35578 Wetzlar
Deutschland
Tel.: +49 6441 207-207
Fax: +49 6441 207-122
Hexagon Technology Center GmbH
Heinrich-Wild-Strasse 201
9435 Heerbrugg
Switzerland
Tel.: + 41 71 7274728
Fax.: + 41 71 7274674