文獻(xiàn)翻譯.pdf

Fusion Engineering and Design 86 (2011) 19581962Contents lists available at ScienceDirectFusion Engineering and Designjournal homepage: /locate/fusengdesA method for enabling real-time structuralcontrol results and 3D model morphingSauli KrassiVTTarticleArticleAvailableKeywords:ITERRemoteDivertorVirtualFlexibilityDeformationMorphingVirtual realityfacility, DTP2 (Divertor Test Platform 2) has been established in Finland forthe Remote Handling (RH) equipment designs for ITER. The first prototypethe Cassette Multifunctional Mover (CMM) equipped with Second Cassette Endto DTP2 in October 2008. The purpose is to prove that CMM/SCEE prototype can2nd cassette RH operations. At the end of F4E grant “DTP2 test facility oper-the RH operations of the 2nd cassette were successfully demonstratedFusion For Energy (F4E).CMM/SCEE robot has relatively large mechanical flexibilities when the robot2nd Cassette on the 3.6-m long lever. This leads into a poor absolutewhere the 3D model, which is used in the control system, does notreflect the actual deformed state of the CMM/SCEE robot. To improve the accuracy, the new method hasbeen developed in order to handle the flexibilities within the control systems virtual environment. Theeffect of the load on the CMM/SCEE has been measured and minimized in the load compensation model,which is implemented in the control system software. The proposed method accounts for the structuraldeformations of the robot in the control system through the 3D model morphing by utilizing the finite1.beenthesystem.deformationsing,morph2Dloadradialnelandpositionof0920-3796/$doi:element method (FEM) analysis for morph targets. This resulted in a considerable improvement of theCMM/SCEE absolute accuracy and the adequacy of the 3D model, which is crucially important in the RHapplications, where the visual information of the controlled device in the surrounding environment islimited. 2010 Elsevier B.V. All rights reserved.IntroductionThe paper presents how the load compensation functions haveimplemented in the control system software to improveabsolute accuracy and visualization accuracy of DTP2 controlIt also proposes a new method for accounting structuralin DTP2 control system through 3D model morph-utilizing the finite element method (FEM) analysis results fortargets. In addition to the actual component morphing, thetexture morphing will be utilized for representing the structuralper each component for better operator evaluation.During the 2nd cassette installation process, CMM travels intodirection, towards the reactor, on top the maintenance tun-rails with the aid of an electric motor drive, Fig. 1. The liftingtilting motions in the vertical plane are used for controlling theand orientation of the cassette according to uphill profilethe maintenance tunnel. The SCEE, which consists of the can-Corresponding author. Tel.: +358 504118792.E-mail address: sauli.kivirantavtt.fi (S. Kiviranta).Fig. 1. CMM and SCEE structural representation.tilever (CRO) and the hook-plate (HRO) rotational joints, is devotedto change the position and orientation of the cassette during thetoroidal motion towards the place of the 2nd cassette. see front matter 2010 Elsevier B.V. All rights reserved.10.1016/j.fusengdes.2010.11.015system by utilizing offline simulationKiviranta, Hannu Saarinen, Harri Mkinen, BorisTechnical Research Centre of Finland, P.O. Box 1300, FI-33101 Tampere, Finlandinfohistory:online 30 December 2010handlingTest Platform 2engineeringabstractA full scale physical testdemonstrating and refiningRH equipment at DTP2 isEffector (SCEE) deliveredbe used successfully for theation and upgrade preparation”,to the representatives ofDue to its design, thecarries the nine-ton-weightingaccuracy and into the situationdeformation in remote handlingS. Kiviranta et al. / Fusion Engineering and Design 86 (2011) 19581962 19592. Content of DTP2 deflection studies2.1. CMM/SCEE trialsAfter delivery of CMM/SCEE to DTP2, the system integrationphase was started in order to prepare the system for the actual test-ing. This testing was done in two phases starting from the factoryfloor and ending to the RH operated tests from the control room1.The initial motion programs for the cassette operations weredone by teaching, while having a continuous visual contact withthe cassette. Good repetition accuracy (3 mm) of the CMM/SCEEguaranteed successful repetition of the motion programs. However,static 3D model could not support operations properly, becauseof poor absolute accuracy of CMM/SCEE. On the grounds of 3Dmodel, it seemed that the cassette was colliding with DivertorRegion Mock-up (DRM) structure although in practice everythingwas fine. It was very clear, that absolute accuracy of the sys-tem should be improved before the remote operations could bestarted.2.2. Load compensationThe effect of load to the CMM-SCEE kinematic chain (body,wheels, links and joints) was measured during the motion pro-grams. And it was realized that the positioning error at the tip ofthe cassette was in the worst case close to 80 mm. The measure-ment data was utilized for creating load compensation functions toimprove the absolute accuracy. The solution is general for the RHmaintenance tunnel operations, but for the toroidal operations theload compensation model is a look-up table based on the valuesof the CRO joint. The compensation approach is simple and wellreasoned because of generality for CMM to support future CMMoperations with other end effectors. The specific look-up tablesare used only when operating with SCEE for performing a spe-cific toroidal trajectory in and out. The compensation functionshelped to improve the performance of the equipment considerably.Thus, the positioning error at the furthest point of the cassette wasreduced from almost 80 mm to about 5 mm 1.Theimplementationofloadcompensationintothecartesianref-erence values can be seen in Fig. 2. The solution is divided into twophases depending on whether the cassette is loaded to the HRO ornot. Cartesian reference for ideal equipment (Fig. 2) is expressingthe location of a coordinate system (Fig. 1) which coincides withthe axis of the HRO joint. Thereby, HRO joint can be used only forchanging the orientation of the cassette around the vertical axis andonly the CRO joint can be used to reach a y-coordinate value of thereference data. For this reason, the load compensation functionsduring the toroidal motions depend on the CRO joint.If there is no load in HRO joint, an inverse kinematics solutioncan be used directly to solve corresponding values for the joint ref-erences. The solution is calculated using the DenavitHartenbergparameters, which include corrections based on the signature cal-ibration of the CMM/SCEE 1.When the load is attached to the HRO joint, the inverse kine-matics solution cannot be used directly because in this casethe cartesian reference includes also the components that rep-resent the deflections of CMM/SCEE. When the effect of loadis known, the correct value for CRO joint can be found eitheriteratively utilizing the inverse of the load compensation or bydefining the least-square polynomial fit between the measured y-coordinate values and the corresponding CRO joint values. Bothiterative solution and 7th order polynomial fit are working wellin practice. After the CRO joint value is defined, the positioncompensation in the x-, y- and z-directions and the orientationcompensation in the Roll- (R) and Pitch- (P) directions can bemade with respect to the cartesian reference. The compensationmovement in the Yaw (W) direction cannot be done becauseof lacking the ability to move in the yaw-direction with theCMM/SCEE.Fig. 2. Left: load compensation in cartesian space. Right: implementation of load effect to the joint data of the real device.1960 S. Kiviranta et al. / Fusion Engineering and Design 86 (2011) 195819622.3.anThisBecausewhichdirection.tactingoverweightto2.4.Mock-upCassetteCADend-effectorswarewereresultsThenandthatThedinateFig. 3. Ansys FEM result (DCM lifted from RH interface).Fig. 4. CATIA FEM result (DCM lifted from RH interface).Improving teleoperator visualization accuracyTilting the cassette in the yaw-direction can be visualized whenadditional joint is added to the 3D model of CMM/SCEE, (Fig. 2).joint has been placed between the hook plate and cassette.of that, the operator can see the effect of the cassette tilting,is 10 mm at the end of toroidal movement in the verticalTo increase the visualization accuracy, when the cassette is con-the inner and outer rails of DRM, the pressure differencethe Lift cylinder provides estimation about the load, as theof the cassette is gradually transferred from the hook platethe DRM rails or the other way around, (Fig. 2).Calculation of the deflections of the Divertor CassetteIn the real operation environment, the shape of the DivertorMock-up (DCM) is never equally represented by the 3D-model. The DCM deflects, when it is handled with the CMMand when it rests on the toroidal rails (Figs. 35).The deformations of the DCM were calculated using Ansys soft-and CATIA FEM-tools. The results of these two calculationscompared. In conclusion, both FEM tools provide similarif the restraints are specified correctly.In the next phase, the FEM results were divided to components.horizontal and vertical deflections of the hook plate handlingresting on the toroidal rails were compared to measurementshad been done for real DCM in the DTP2 laboratory (Table 1).measurement device is Sokkia NET05 high precision 3D coor-measuring system (theodolite).Fig. 5. Vertical and horizontal deflections in respect to cassette structures.Comparison between the FEM results and the Sokkia measure-ments showed that the real DCM behaves as it was analyzed.2.5. Visualization of the deflections of the Divertor CassetteMock-upDesign of the DCM has been made according to applicable designrules and standards of the machine design. As a result, the stressesare always below the proportional limit of the construction mate-rial and the behavior of material is linearly elastic. The initial testsin this study were conducted under the Hookes law assumptionfor linear deformations.Hence the results of the FEM analysis can be utilized for thevisualization of the DCM deformations. Problem of loaded deviceshape not being reflected to the teleoperator view makes accuratecontrolofthesystemnearimpossible.Teleoperatorvisualizationbyaccounting deformations can be carried out in two different ways.The traditional method is to divide a body into pieces and to createthe link mechanisms between the pieces 4. This approach requiresa lot of analysis work. The position of the joints and the maximumjoint values are the result of these analyses. Based on the analyses,it was recommended to divide DCM into three links, which wereconnected with two rotational joints, Fig. 6.Method proposed by this paper is to use the 3D morphing theprocess of gradual transformation between 3D bodies to describethe deformations of the body based on the FEM analysis results. Themetamorphosis or the (3D) morphing of the 3D graphical objects,also known as shape interpolation, is the process of transform-ing one shape into another 2. This technique allows utilizationof the FEM analysis results directly without laborious link-jointapproximations. In addition, this method enables the use of sep-Table 1FEM results compared to Sokkia measurements.DCM deflectionsCriteria Horizontal Vertical UnitTotal deformation based on FEMbetween hook plate handling andresting on toroidal rails7.2 9.2 mmSokkia measurements between hookplate handling and resting on thetoroidal rails6.6 7.3 mmS. Kiviranta et al. / Fusion Engineering and Design 86 (2011) 19581962 1961Fig.rotationalFig.(forces).aratetarget(accountingmultipletiondeformedstraintionSystemsbuiltthe6. The body of the cassette is divided into three rigid links connected with twojoints to approximate mechanical flexibilities.7. Simplified example of 3 links deformed by 9 individual morph targetsdeformation results by utilizing FEM results for each morphper force applicable to the component for a given scenarioFig. 7). This provides a high degree of adaptation capabilities fora large variety of flexibilities in complex systems wheresources of forces can affect each part of the system.For morphing the model, we have used the linear interpola-between the vertices in the non-deformed 3D model and theFEM model. A more advanced morphing algorithm is thefield interpolation 3 if the accuracy of the linear interpola-is not sufficient for the given application.For visualizing the deformations to the teleoperator, DassaultVirtools 5.0 was used (Fig. 8). The virtual environment isby directly utilizing ITER CATIA models in conjunction withFEM models that are utilized for creating morph targets.The benefits of the proposed method are the following:Applicationofthefullyflexiblemeshmorphingbetweenthecom-ponents unloaded neutral states and the deformed states for agiven maximum force per morph target thus directly utilizingthe FEM results, Fig. 4.Easier reuse of the existing deformation data obtained by theoffline and online analysis of real systems.More accurate representation of each section of the complex sys-tem components and full control over the continuum possibledeformation points, instead of rough estimations gained troughjoint-link approximation.Possibility to combine multiple deformations separated by indi-vidual forces in complex system.Fig. 8. Example of DCM morph targets within virtual environment.2.6. Controlling of the 3D model deformationsControlling the 3D model deformations means that the defor-mations in the virtual environment have to follow the actualdeformations. The deformation information can be determinedbased on the previously measured deformations for a givenoperation state or by using the hydraulic system pressuresto estimate the force, hence utilizing existing sensor informa-tion.In the case of robot operations, a more accurate solution can beachieved by employing strain gauges to measure the actual defor-mations of the robot links. In the laboratory tests, the strain gaugeswill be installed to the DCM.The advantage of the strain gauges includes:Complementarity to the morph target method, where for eachforce one can have an individual morph target controlled by adedicated strain gauge.Ability to measure the exact deformations immediately, notrelying on the previously measured static deformation data orhydraulic pressures that may not be available for all force direc-tions.3. Future workThe continuation regarding the fully flexible 3D virtual proto-typing of the DTP2 robot components will be as follows. Initially,the DCM flexibility studies will focus on s