Document
NASA Technical Memorandum 107651
ON-LINE ANALYSIS CAPABILITIES
DEVELOPED TO SUPPORT THE
AFW WIND-TUNNEL TESTS
Carol D. Wieseman, Sherwood T. Hoadley,
and Sandra M. McGraw
N92-30732
July 1992
(,fASA-T',-IOTc£ t ) :jN-L [N_ ANALYSIS CAP&LilLITILS r;CVEL'-'hPL'_' T3 SUPPO_,I TI'-'E A_-, "_ _,T ''_''-T:Jt'iN'cL TFST3 (NASA) Oncl3s l I i) G3/05 0115674
N/kSA
National Aeronautics and Space Administration Langley Research Center Hampton, Virginia 23665-5225 ON-LINE ANALYSIS CAPABILITIES DEVELOPED TO SUPPORT THE AFW WIND-TUNNEL TESTS Carol D. Wieseman*, Sherwood T. Hoadley** NASA Langley Research Center Hampton, VA 23665-5225 Sandra M. McGraw Lockheed Engineering and Sciences Company Hampton, VA 23666 Controller Performance Evaluation Abswact CPE Closed Loop A variety of on-line analysis tools were developed to CL support two Active Flexible Wing wind-tunnel tests. DCS Digital Controller System Fast Fourier Transform These tools were developed to verify control law FFT Flutter Suppression System execution, to satisfy analysis requirements of the control FS S Periodic Pseudo Noise law designers, to provide measures of system stability in a PPN Rolling Maneuver Load Alleviation real-time environment, and to provide project managers RMLA with a quantitative measure of controller performance. RMS Root Mean Square Roll Rate Tracking System Descriptions and purposes of capabilities which were RRTS developed are presented in this paper along with examples. RTS Roll Trim System Procedures for saving and transferring data for near real- OL Open Loop time analysis, and descriptions of the corresponding data Introduction interface programs are also presented. The on-line analysis tools worked well before, during, and after the The cooperative NASA/Rockwell International Active wind-tunnel tests and proved to be a vital and important Flexible Wing (AFW) program 1 included wind-tunnel part of the entire test effort.
testing of an actively controlled aeroelastic wind-tunnel model that could be configured to roll. An important goal Nomenclature of the program was to test flutter suppression control laws G open-loop plant wansfer matrix and roiling maneuver control laws, first, independently, H open-loop controller transfer matrix and then simultaneously above the open-loop flutter I identity matrix boundary. A Digital Controller System (DCS) 2 was excitation u developed to implement these various control law x controller output functions while accommodating various types and X controller output transfer matrix combinations of control law implementation. The DCS Y plant output (sensors and swain gages) receives sensor outputs from the model, processes them Y plant output transfer matrix through the control laws, sums the various control law 7,.
eigenvalues actuator commands, and then sends these back to the model.
O" singular values, o = "_/-_(A*A), for a given In order to verify the execution of each control law matrix A; o are always non-negative real.
during various stages of development of the DCS and to maximum singular value evaluate controller performance during the tests, it was necessary to generate time-history responses to excitations. These excitations could be added to either the Subscripts refers to control law elements control law inputs or outputs at various points in the refers to elements external to control law execution loop and to perform analysis of individual control law performance. The DCS engineers needed these Notaaon capabilities to debug the internal implementations and execution of the various control laws. The control law det(-) determinant designers and the project managers all needed guarantees (-)* complex conjugate transpose that control laws were being implemented properly both (.)T matrix transpose prior to and during wind-tunnel testing in order to protect the wind tunnel and model from damage.
Acronyms Various analysis packages and computer systems were AFW Active Flexible Wing explored for their capabilities. Most of these could not meet the requirements of the AFW program, either * Member, AIAA because of the unavailability of hardware, software, ** Associate Fellow, AIAA networking capabilities, programming support, or simply lack of computation speed. Since all the signals required for analysis were already available within the DCS and digitized at the sampling frequency of the DCS, and since a second DCS system was available as a backup to the manufactured by SKY Computers, Inc. This board performed all the Fast Fourier Transforms (FFT's) required primary system, it was decided that the most expedient solution was to develop the required analysis tools on the to compute transfer functions within a time frame which backup DCS. This second DCS, which would be used as would allow for near real-time processing. Figure 1 depicts the SUN workstation (SUN-l) which was used for a backup only upon failure of the central processing unit in the primary DCS, could be hooked to the primary DCS the primary DCS and the second SUN workstation (SUN- via an Ethernet line for data transfer. It was considered a 2) which was used as an on-line digital signal analyzer small investment that more cautious wind-tunnel runs where data translation and near real-time analyses were might have to be accommodated in order to perform on- performed. It also depicts the signals passed between the model and SUN-1 as well as the Ethernet connection line analysis before each critical step in the testing.
To satisfy the analysis requirements of the AFW between the two computer systems. Selected data was program, an extensive package of analysis capabilities saved automatically in binary form on SUN-1 and transferred as a binary data tiM, via the Ethernet line, from was developed. Since the signals used were those digitized SUN-1 to SUN-2. It was recognized that if the SUN-2 by the DCS and the analysis could be performed while the DCS was operating, the analysis capabilities are referred system had to be used as a backup DCS, data would have to be analyzed between test runs, requiring more cautious to herein as on-line capabilities. This package included data interface programs which converted integer data testing and fewer test accomplishments while the SUN-1 system was being repaired. Since the SUN-2 would be representing voltages to scaled signal data of selected signals. It included plotting routines which could provide required as a backup DCS only if the SUN-I central processing unit itself crashed, it was decided that this was time histories of all internally saved, digitized data from a small risk.
the DCS and Fourier analysis tools which calculated transfer functions of any combination of output/input On-Line Analysis Requirements pairs of signals from any control law could be computed and plotted. In addition to these basic analysis tools, a Different types of active control wind-tunnel tests Conlroller Performance Evaluation (CPE) code 3 was also were performed in the AFW program. These included developed. The CPE code proc_sed the matrix of transfer testing flutter suppression systems (FSS) and roll control functions for the FSS and RMLA control laws to laws. Several roll control laws were developed; a roll trim determine 1)closed-loop stability from open-loop system (RTS), a roll rate tracking system OLRTS) 4, and a measurements, 2)measures of stability for a closed-loop rolling maneuver load alleviation system (RMLA) 5. In system, and 3)open-loop plant stability from closed-loop addition to operating each of these control laws measurements.
individually, an FSS control law 6-9 could also be operated Some capabilities were considered essential to safe in combination with a rolling control law. Each type of testing of the model, while others were, simply, nice-to- testing had specific on-line analysis requirements have and provided additional analysis information from the associated with it. Table 1 is a summary of the types of wind-tunnel test. These two classifications of on-line analysis requirements for each type of testing to be capabilities, critical and supporting, are described in this performed in the wind tunnel.
paper with emphasis on those capabilities which were Execution of both types of control law, FSS and considered critical. Details of data saving and data Iransfer Roll, had to be verified in the DCS, f'LrStin a wind-off and a description of the Fourier analysis program are also environment with just the DCS, and then in the wind-on presented in this paper.
environment with the model included. This had to be done while each control law executed independently and in Hardware conjunction with other control laws. Evaluating total The primary and backup DCS were comprised of conlzoller performance, both with feedback off (open loop SUN 3/160 workstations configured with similar hardware (OL)) and feedback on (closed loop (CL)), was required boards. One of these boards was a fast array processor, while testing the model with the DCS in the loop. For Table I: TYPE OF TESTING Wind-off Wind-on REQUIREMENTS _DCS only) (DCS + model) Roll FSS FSS+Roll Roll FSS FSS+Roll No Control OL CL OL CL OL CL Law q q 4 Control Law Verification Time-Domain Con_olier ,J q q q _ 4 Performance Evaluation q q _ Frequency-Domain Controller Performance Evaluation Plant Determination ,J ,J q ,J q ,J 4 q 4 Flutter Boundary Prediction requirement to obtain plant transfer functions both with the rolling control laws (RTS, RMLA, and RRTS), time- and without a control law operating.
history plots were needed for the control law designers to evaluate the commands sent to the model and to evaluate On-Line Analysis Capabilities the performance of the control laws. Although external Fourteen on-line analysis capabilities were developed signals could be seen on strip charts, the internal signals in conjunction with the AFW program in order to meet used by the control law as inputs and outputs could not.
the five major analysis requirements listed in Table 1.
For the RMLA control laws as well as the FSS control These capabilities generally can be divided into time- laws, frequency-domain CPE was also required.
For some control law designers, plant transfer domain and frequency-domain analyses. Table 2 is a functions were necessary for use in improving their summary of the requirements and the specific analysis capabilities which were developed to achieve these con_ol law designs 7. There was also a requirement to requirements.
predict the open-loop flutter boundary while operating The data used for the analyses was digitized by the closed loop. The plant transfer functions were also DCS. In all the DCS modes of operation which involved necessary in order to meet this need. Since not all signals wind-on testing, different blocks of time-history integer could be saved while operating a control law, there was a Table 2: On-line Anal REQUIREMENTS Plant Flutter Control Time Frequency Law Domain Domain Determinatior Boundary CAPABILITIES Prediction Verification IPerformance Peaformance Evaluation Evaluation X X Plot Time Histories Time Calculate RMS Values Plot RMS Values Domain X X X X Calculate Transfer Functions X X X Generate Overall Transfer Ma_ix X X X Extract Plant Transfer Matrix X Extract Controller Tran_qferMatrix X Plot Transfer Functions X Frequenc) Calculate Singular Values and Determinants of Retum Difference Domain Mamces Plot Singular Values and Determinants of Return Difference Mau'ices Calculate Inverse Maximum Singular Values of Plant Plot Inverse Maximum Singular Values of Plant Calculate peak-Hold Data Plot Peak-Hold Data an Ethemet data line. Two data interface programs were data representing signal voltages could be saved on a written to convert the data into different formats. One binary file depending upon the mode of operation 2. The program converted the time-history data into Madab 10 length of each block was determined by the length of the format for use in plotting routines implemented in excitation, or specified by the DCS operator. The exact Maflab. The other converted the time-history data into a data which was saved was a subset, selected by the format required by a program written to calculate the control law designers, of the set of total possible signals.
transfer functions using the array processor. If the transfer The first binary record of the data file contained a header functions were for FSS analysis, the interface program for which included the tunnel tab number, and other transfer function data symmetrized or antisymmetrized the parameters including Mach number, dynamic pressure, time-history data dependent on whether the excitation was mode of operation, type of excitation, and whether the a symmetric or antisymmetric excitation. The interface excitation was symmetric or antisymmetric.
programs and analysis programs used the header Figure 2 shows a flowchart of the on-line information to determine the types of conversions and capabilities. The capabilities are enclosed within scaling required.
rectangular boxes. Requirements are indicated by bold In order to generate transfer functions for frequency- lettering. Arrows depict the flow of capabilities necessary domain analyses, a transfer function analysis program was to obtain data to satisfy each requirement. In each case, developed. This program could perform overlapped binary data files were shipped to the SUN-2 computer via
averaging ofall signals saved by theDCS, window the
Digitized response data was saved and sent to the data with one of several selectable windowing functions, SUN-2 where transfer functions were calculated using the and generate FFTs using the array processor. The array transfer function analysis program. Designer-supplied processor was capable of calculating an FFT of 4K data analytical frequency responses were also loaded and plots blocks in 0.007 seconds. Transfer functions were of the analytical transfer functions were superimposed to generated for any pair of signals. This entire program directly compare the digitized control law as generated by took less than half a minute to calculate all the transfer the DCS with the designed control law. This was repeated functions required for each excitation. Postprocessors of for all control law inputs. This capability was used to this data were then developed to either plot the wansfer verify both the FSS control laws and the RMLA control laws.
functions, perform state-space analyses, generate the plant transfer matrix, or extract the open-loop control law The next step in frequency-domain control law wansfer functions from a closed-loop system.
verification involved exwacting the control law transfer functions from a system which included the plant in one Control Law Verification of five configurations. They were: Control law verification was required to assure that 1) extracting the control law transfer functions from the control law was loaded properly into the DCS and was an open-loop system in which the excitations were the same as the designed control law. Time-domain and added to the control law outputs frequency-domain capabilities were developed and used to 2) extracting the control law transfer functions from a verify the correctness of control law implementation.
closed-loop system in which the excitations were For time-domain analysis, time responses of the added to the control law outputs control law due to a specific input were plotted. For the 3) extracting the control law wansfex functions from FSS and RMLA control laws, the inputs were step an open-loop system in which the excitations were functions. For the RRTS and RTS control laws, the added to the final actuator commands input was a sine wave whose amplitude was large enough 4) exwacting the control law transfer functions from a to encompass the entire range of the control law. The closed-loop system in which the excitations were response time histories were compared directly with added to the final actuator commands similar responses provided by the control law designer, 5) extracting the control law mmsfer functions from a and discrepancies were accounted for by either correcting closed-loop system in which the excitations were the DCS, the scaling parameters, or the input data for the added to the sensor inputs.
control law.
An example of the transfer function plots resulting from Since time-history comparisons do not clearly show control law extraction is shown in figure 3. Both the discrepancies in frequency content and phasing, a conffol law which was extracted and the designed transfer frequency-domain method for verifying state-space control function match exactly, as they should.
laws was developed to supplement the time-domain analyses. This frequency-domain method included a series Time-Domain Controller Performance Evahiation of steps to determine the controller-only transfer functions Time-history plot capabilities were developed for use between various points in the DCS, providing a step-wise during rolling maneuvers to provide a means for the control law verification scheme.
designer to evaluate whether the control law was operating The first step in frequency-domain control law as expected, to evaluate whether the command input was verification involved computing transfer functions of all correct, and to assess the loads during the maneuver.
the outputs of the control law with respect to each input.
Separate plotting functions were written to plot the data To provide data for this step, excitations were input into saved in any one of the rolling modes, RTS, RMLA or each control law corresponding to each sensor input. A RRTS. The control law designer chose four of seventeen Matlab program for generating digital excitations was channels of saved data to be plotted during the test. The developed to provide excitations. These excitation signals plot routines were optimized to require a minimum of could be generated before testing and then loaded into intervention from the analyst providing the plots during memory at a specified time. The excitation options were wind-tunnel operation. Examples of two out of the four a linear sine-sweep, log sine-sweep, and a periodic pseudo time-history plots which were generated on-line for an noise (PPN). The PPN was a specially designed RRTS control law are shown in figure 4. They are the excitation which provided high signal to noise ratios with measured roll rate and the measured roll angle. Additional a specified frequency resolution subject to constraints on signals which were saved could also be plotted after a test control surface rates. It is not truly random and has a run to gain greater insight or to further evaluate controller specified frequency content, generated by picking a block performance. Plot routines were also written to plot any size which determines the frequency resolution.
of the seventeen channels of time-history data saved during Generation of all excitation types except the PPN was the FSS mode.
also possible by the DCS during execution. However, During the 1989 wind-tunnel test, calculation of the generating linear sine-sweeps, log sine-sweeps, and PPN's Root Mean Square (RMS) values of control surface required several minutes of execution time. These commands and rates was required to evaluate FSS excitations were, therefore, normally generated externally controller performance since high RMS values of control and saved on external files so desired excitations needed surface actuators would indicate saturation and impending only to be loaded (not generated) by the DCS. This closed-loop flutter. Consequently, the capability to process saved valuable test time.
calculate RMS values, mean values, and maximum values Figure 7 presents an actual output CPE for a point of any saved data, including control-surface commands and rates, accelerations, and loads, was developed. The RMS's above the open-loop flutter boundary. The upper plots in the figure are plots of the singular values of the return of symmetric and antisymmetric data were calculated for difference matrices. These provide measures of robusmess data saved during data acquisition for frequency-domain with respect to multiplicative uncertainty at the plant CPE in which excitations were either symmetric or input and plant output points, respectively. The plots antisymmetric, and those for fight and left wing data were calculated for data saved during peak-hold data acquisition. shown in figure 7 are for a single-input/single-output The capability was also developed to save the calculated system, so, in this case, both plots are identical. The plot in the lower left depicts a measure of robusmess with RMS data and plot them as a function of dynamic respect to an additive uncertainty. The determinant plot in pressure. Figure 5 is an example of the plots of RMS control surface deflections and rates versus dynamic the lower right provides a means of checking open-loop stability.
pressure.
The capabilities to plot the determinant plot, Since the model trip system worked so well in separately, in order to beUer identify encirclements, and to providing a measure of safety to the model and the frequency-domain controller performance capabilities generate a Nichols plot in order to view determinant data proved to be substantially accurate, the RMS calculating in a manner which not only showed encirclements but capability was used only as a secondary source for CPE also gave gain and phase information were also developed.
during the 1991 wind-tunnel test entry.
plant Determination Freouencv-Domaln Controller Performance Evaluation To determine the plant in the case when there is no Frequency-domain capabilities were developed as a control law operating, the plant transfer matrix can be primary source for evaluating controller performance. 3 A derived directly from the calculated transfer functions. In flowchart of the frequency-domain CPE capability is the case when there was a control law operating, the plant shown in figure 6. Transfer functions were fgst calculated had to be exwacted from the closed-loop system. In either and combined into a transfer matrix and the frequency case, the purpose of plant determination was two-fold.
range over which to execute the CPE code was selected.
The fn_t was to provide transfer function data to engineers The open-loop plant transfer matrix, G, and controller for their use in redesigning control laws and the second transfer matrix, H, as well as the open-loop system purpose was to use the open-loop plant to evaluate open- transfer matrices at the plant output and the plant input loop plant stability. Some elements of the plant uansfer points, HG and GH, respectively, were then calculated or matrix were extracted during CPE calculations; however, extracted, using equations presented in reference 3, for an additional capability was required to calculate the either an open-loop or a closed-loop system. Singular remaining elements of the plant transfer matrix.
values and/or determinant values of various return- Figure 8 shows a block diagram of the plant and difference matrices were then calculated. From these, controller. The "c" subscript refers to the control law maximums, minimums, and inverse maximum values elements. The "e" subscript refers to elements external to were calculated and plotted in order to evaluate the the control law tested. Table 3 outlines the equations performance of FSS and RMLA control laws.
needed in order to calculate all the elements of the plant One exception to the procedure outlined in figure 6 transfer matrix: was made for the FSS control law described in reference 7, having more sensor inputs than control law outputs. In order to reduce wind-tunnel testing time needed to extract the open-loop controller transfer matrix, H, from the Gee Gee closed-loop system as described in reference 3, H was analytically generated prior to the wind-tunnel test and loaded separately into the CPE code.
In the table, Xc and Xe are the transfer functions of the Table 3. Basic Plant Equations* Open -Loop
Cnosed-Loop
Gcc -- Ycc
Gcc = (tI- XcT]'tYccT)T
Gec = Yec Gec = ([I- xcT]-IYecT)T Gee ffi Yce Gce =Yce+GccXe G_ = Ye_ G_ =Y_+G¢_X_ * All matrices are functions of c0.
control law outputs, x, with respect to Uc (excitations of the transfer functions of the plant outputs, Ye, not used by control surfaces used by the control law) and ue (those the controller with respect to Uc and ue, respectively.
notused by the control law). Ycc and Yce are the transfer Flutter Boundary Prediction functions of the plant outputs, Yc, used by the controller with respect to Uc and u¢, respectively. Yec and Yee are One of the purposes of the on-line analysis was to Digital Control Systems, AIAA Journal of Guidance,
determine the open-loop plant stability from closed-loop
data. The inverse maximum singular values of the plant Control and Dynamics, pending publication.
were computed for many dynamic pressures. A plot of the inverse maximum singular values of the plant at one test 4. Moore, Douglas B.: Maneuver Load Control Using condition is shown in figure 9. The point at which the Optimized Feedforward Commands, AIAA Paper 92- inverse maximum singular values goes to zero is the 2100-CP, presented at the AIAA Dynamic Specialists Conference, Dallas, Texas, April 16-17, 1992.
point at which open-loop flutter is predicted to occur. A plot of these global minimum points is shown in figure 10. The curve is extrapolated to predict the open-loop 5. Woods-Vedeler, Jessica A. and Pototzky, Anthony S.: flutter boundary. The predicted flutter boundary using this Rolling Maneuver Load Alleviation Using Active Controls, AIAA Paper 92-2099-CP, presented at the technique compared well with a hard flutter point which AIAA Dynamic Specialists Conference, Dallas, was determined from open-loop testing at the end of the Texas, April 16-17, 1992.
wind-tunnel test entry.
In order to predict closed-loop flutter, the capability to perform peak-hold analysis was developed to determine the 6. Waszak, Martin R.: Flutter Suppression for the Active Flexible Wing: Control System Design and peak value at each frequency of the autospectra of a signal Experimental Validation, AIAA Paper 92-2097-CP, as it was calculated over a period of time using overlapped presented at the AIAA Dynamic Specialists processing. Data due to random turbulence was saved by the DCS, and thecapability of calculating and plotting the Conference, Dallas, Texas, April 16-17, 1992.
peak-holddataof multiplechannels both symmetrically Christhilf, David M. and Adams, William M.jr.: and antisymmetrically during the wind-tunneltest was 7.
Multifunction Tests of a Frequency Domain Based developed. Any of thesaved sensordata could be used to help determine the closed-loopflutter boundary during Flutter Suppression System, AIAA Paper 92-2096- CP, presented at the AIAA Dynamic Specialists closed-loop testing. First, the maximum peak-holddata pointwas determined for each test point and the inverse Conference, Dallas, Texas, April 16-17, 1992.
maximum points were then plotted as a function of dynamic pressure.This curve was then extrapolated to 8. Mukhopadhyay, Vivek: Flutter Suppression Digital zero to predictwhere closed-loop flutter would occur. Control Law Design and Testing for the AFW Wind- Results from the peak-hold capability compared wellwith Tunnel Model, AIAA Paper 92-2095-CP, presented at other sources.
the AIAA Dynamic Specialists Conference, Dallas, Texas, April 16-17, 1992.
Concluding R¢marks .
Klepl, Martin J.: A Flutter Suppression System On-line capabilities, implemented using the Digital Using Strain Gages Applied to Active Flexible Wing Controller System and its backup equipment, were Technology: Design and Test, AIAA Paper 92-2098- developed to support the AFW wind-tunnel test. The purposes of the on-line analyses were to verify that CP, presented at the AIAA Dynamic Specialists Conference, Dallas, Texas, April 16-17, 1992.
control laws executed properly on. the Digital Controller System, to provide control designers with a means to 10. PRO-MATLAB User's Guide, The MathWorks Inc.; evaluate overall controller performance, and to provide 21 Eliot Street; South Natick, MA 01760.
guidance to the wind-tunnel test manager in determining the progress of the wind-tunnel test. The capabilities worked extremely well before, during, and after the wind- tunnel test and proved to be a vital and important part of the test effort by providing on-line near real-time analysis capabilities.
References 1.
Perry, Boyd. III; Cole, S.; and Miller,G: A Summary of the Active Flexible Wing Program, AIAA Paper No. 92-2080-CP, presented at the AIAA Dynamics Specialists Conference, Dallas, Texas, April 16-17, 1992.
2. Hoadley, Sherwood T. and McGraw, Sandra M.: The Multiple-Function Multi-Input Multi-Output Digital Controller System for the AFW Wind-Tunnel Model, AIAA Paper 92-2083-CP, presented at the AIAA Dynamic Specialists Conference, Dallas, Texas, April 16-17, 1992.
3. Pototzky, Anthony S.; Wieseman, Carol D.; Hoadley, Sherwood T.; and Mukhopadhyay, Vivek: Figure 1.- Hardware involved in on-line analysis.
On-Line Performance Evaluation of Multi-Loop SUN-1 SUN-2 Data Interface [ | Data Imerface ¢c') I .11-UAl_-_ fo_t J. / I cc_ Functions and Peak Hold _ _ CONTROLLER PERFORMANCE I Data and Corresponding I [histories of [Jr I EVALUATION __ I Calculate Transfer (FORTRAN) De_gner §uppt_ [ I control law I I • T Control Law I [ response [ I [ Calculate RMS Values ] ] Plot time [
t
Tnmsfer Functions [ [ (MATLAB) [ L._ Svmm_xic or [ [ histories of " " -"-7 X_'t_m,_'¢ IRightandL,,f, I I_°_d-l°°P Transfer Matrix
~MATLAB fon'nat 1
(O
PLAN_DETERMINATION I \/ i
c_at_ I i, J' _ J, Matrix Extraction Extraction Plant Transferl , y D.- , -._ T , and Code Mazrix i i Listing of [ Plot RMS Values I Calculate e.tATL_) eUA_> i [ RMSval,m [ venas I Difference Matrices (MATLAB) FLUTrER PREDICTION FREQUENCY.DOMAIN Calculate and Plot Peak-Hold Data Inverse Maximum -MATLAB fomaat CONTROLLER Open-Loop _ Closed-LOOPconvert [ PERFORMANCE Valuesof (C) EVALUATION Frequency (.)Denotes Programming Language Plot CPE Results r-_que,,cy I I Plot vmms l Plot Minimum
L_L CUATL_W> I
of Inverse Maximums versus Dynamic Pica MLums Pressure (MA'II.Ae) Figure 2.- Flowchart of on-line analysis capabilities.
HEADER INFORMATION " ....... 4 i i ' ! _ !
t i .-' i i i , _ .......... 2.1 _-100 4.8 4.9 5.1 5.2 5.3 5.4 5.5 5.6 $3 5.8 -- Extracted Control Law Designed Control Law 100 ....
100 , , , , , , , , 50 ....... ! ................ ! ................ _ ................. ................. _ .................. ! ................. _ ................. - .................
- 0| ! _ i -'x,, i ! ! !
i i i i i i
° "'" i i i'.-:.t- -
a. -I00 -50 ............ _............... _............. ' ......... : ................ :- ............. _.............. _............... _............ 4i ............
4 6 8 10 12 14 16 18 20 4,8 4.9 5 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 time(me) Freq, Hz Figure 3.- Comparison of control law transfer function Figure 4.- Time-history plots of data acquired during extracted from a closed-loop system with one rolling maneuver with RRTS operating.
supplied by control law designer.
HEADER INFORMATION 0.2 J _ FFFs and T 0.15 -_ - Combiae tnmsf_ ftmotim_ Save ia MA'I'LAB fenma I 0.1 MATLAB CI_ COMMJTATIONS - --_ __...', .............................................. i ....................
I I.,md opm_ m' dm_-Iaop Ira_fm- funclion ram/era, YU mdXu 0 t t T 150 200 250 300 q(psO OPEN-LOOP T _ED-LOOP 15 'IEO IC_w - I , IO_.O._..o I _swe i I0_..
c,.,p .- I l,,o--,_i "_ 5 ..................................................................
050 '
250 3O0 rain/max o(I+GH) I doffl+i'kJ) I q(psO ;/o_}qr.,_'b i Figure5.-Plots of RMS values of control surface Figure 6.- Flowchart of frequency-domain CPE deflections and rates.
procedures.
HEADER INFORMATION 8j mrni__6_.l 6 |1! i ] 81r_in'aX--0.6_ A6 _11 0 5 10 15 20 0 5 10 15 20 Freq, Hz Freq, Hz det(I. -HG) 2O max=22. I1
rain=0.29 r
[D + 6 ..... iit7.[ Hz ..........
................. . ................
u 2:: -20 0 5 10 15 20 -20 20 Freq, Hz Real Figure 7.- Closed-loop CPE results of a symmetric FSS control law for M--0.44 and q=249 psf.
Plant HEADER INFORMATION 2O Ye U@ • i i ' rain= 0.0_874 at 9.9_ Hz' 18 ............. _ .......... i .............. _ ................ ! .........................
G
Y¢ 14 ...... i ......................................
Control Law It_ 10_ ............. _ ........................................................
--8[['i [ i ............ ...... ............. ................. , ........
4 .............. i .............. : i ........... i ............ ........... i ......... ....
61 ............. i ............. _ ............. i ............... i ........ _........ i ............ i ............ i .............. i ............
Figure 8.- Controller-plant diagram depicting the control o2 .............. :i ............. _ ........................ .............. i ...................... i ..............
problem with negative fe_back.
0 2 4 6 8 I0 12 14 16 18 20 Freq,Hz Figure 9.-Plot of inverse maximum singularvalue of the open-loop plant transfer matrix, q=230 psf.
HEADER INFORMATION 0.8 0.7 .............. _ ................ i .............................. i ............................. _ .............................. i ............................
0.6 .............. i ............... i............................ i ............................. _ .............................. ..........................
0.5 [_0.4 0.3 0.2 .....
0 , 160 170 180 190 200 210 220 230 240 250 260 q(psf) Figure 10.- Open-loop flutter prediction using closed-loop CPE results.
Form Approved REPORT DOCUMENTATION PAGE OMB No 070_-018S Public re_tll_g burden for _hlscollection of informaL*on _s est_mate_ to average 1 hour per response, _ncluding the time for reviewing instructions, searching existing data sources, gathering and rnamtam#ng the data needed, and completing anO reviewing the coHechon of information Send commentsrec_arding this burden estimate or any other aspectof this collectionof mtormat_on, irlcludlng suggestlor'_s for reducing this 13urden _o Washington HeadQuarters Services.DirectoratetOT _nformatlon Operations and ReDocts,1215 Jefferson OavtsH=ghway, Sude 1204 Arhngton, VA 22202-4302, and to the Office of Management and Budget. PaperworkReductionPrOject (0704-018B),Washington, OC 20503, 3. REPORT TYPeANDDATES COVERED 1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE July 1992 Technical Memorandum 4. T'ITLE AND SUBTITLE 5. FUNDING NUMBERS On-Line Analysis Capabilities Developed to Support WU 505-63-50-15 the AFW Wind-Tunnel Tests 6. AUTHOR(S) Carol D. Wieseman, Sherwood T. Hoadley, and Sandra M. McGraw 8. PERFORMING ORGANIZATION 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) REPORT NUMBER NASA Langley Research Center Hampton, VA 23665-5225 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING / MONITORING AGENCY REPORT NUMBER National Aeronautics and Space Administration NASA TM-107651 Washington, DC 20546-0(X)1 11. SUPPLEMENTARY NOTES Presented at the AIAA Dynamic Specialists' Conference, Dallas, Texas, April 16-17, 1992.
Wieseman and Hoadley: NASA Langley Research Center, Hampton, VA; McGraw: Lockheed ngineering and Sciences, Co., Hampton, VA.
12a.DISTRIBUTION / AVAILABILITY STATEMENT 12b, DISTRIBUTION CODE Unclassified - Unlimited Subject Category 05 13. ABSTRACT (Maximum 200 words) A variety of on-line analysis tools were developed to support two Active Flexible Wing wind-tunnel tests. These tools were developed to verify control law execution, to satisfy analysis requirements of the control law designers, to provide measures of system stability in a real-time environment, and to provide project managers with a quantitative measure of controller performance. Descriptions and purposes of capabilities which were developed are presented in this paper along with examples.
Procedures for saving and transferring data for near real-time analysis and descriptions of the corresponding data interface programs are also presented. The on-line analysis tools worked well before, during, and after the wind-tunnel tests and proved to be a vital and important part of the entire test effort.
14. SUBJECT TERMS 15. NUMBER OF PAGES Analysis tools; controller l:_rl'ormance; controller verification; 16. PRICE CODE data reduction; near real time; plot routines; software; wind-tunnel tests A02 17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT OFREPORT OF THIS PAGE OFABSTRACT Uncla_ifie,¢l Uncla_._ified NSN rsa0-01-:s0-5500 Standard Form 298 (Rev 2-89) PrescrlDed by ANSI Std Z39-18 298-102