KAI KF-21 Boramae (KFX Korean Indigenous Fighter program)

"Tech Transfer Hobbles South Korea's Fighter Program"
By Jung Sung-ki 2 p.m. EDT September 27, 2015

Source:
http://www.defensenews.com/story/defense/air-space/strike/2015/09/27/tech-transfer-hobbles-south-koreas-fighter-program/72808800/

SEOUL, South Korea — Only six months after Seoul’s KF-X fighter program took off, the state project valued at about US $15 billion faces a crash over tech transfer issues with the US government.

The KF-X is linked to Lockheed Martin’s sale of 40 F-35As to South Korea under the F-X III program through the Foreign Military sales process. Lockheed Martin offered to provide 21 technologies required to build the KF-X fighter jet as part of F-X III offset deals. The US aircraft giant was also selected as the main partner to build KF-X with Korea Aerospace Industries.

At the request of Seoul’s Defense Acquisition Program Administration (DAPA), Lockheed Martin also agreed to consult with the US government over the transfer of four more technologies related to the active electronically scanned radar (AESA), electro-optical targeting pod, infrared search-and-rescue systems, and radio frequency jammer.

In April, however, the DAPA received notice of the refusal to transfer the four technologies, according DAPA officials.

“We were trying to secure the US export license of those technologies but failed to get them,” Lt. Gen. Park Shin-kyu, head of DAPA’s procurement business bureau, said. “Instead, we’ll seek ways of obtaining those technologies from other countries or through local developments.”

The general hinted the KF-X timeline could be further delayed from 2025 without the US tech transfer.

The AESA, in particular, is a key specification of the KF-X, which is expected to be a twin-engine F-16-plus fighter jet with high-end sensor systems. Some 120 jets are to be built to replace the aging fleet of F-4s and F-5s.

Lockheed Martin says it is still consulting with the US government over transfer of the technologies.

“There is no F-35 customer nation receiving the AESA radar technology,” a Lockheed official said. “We made it clear that the tech transfer is only possible with the approval of the US government. We tried but failed.”

Cheong Wa Dae (executive office), however, said DAPA was trying to cover up the tech transfer failure.

The presidential office launched a probe into the suspicion that DAPA had selected Lockheed as the final bidder for both the F-X III and KF-X programs in an unfair manner.

“The presidential office will look into all paperwork and documents related to the KF-X program,” a Cheong Wa Dae spokesman said, adding that the fighter development plan could be halted if it is judged nonviable economically and technically.

Some pundits anticipate the KF-X fiasco could affect the F-X III contract in a worst-case scenario.

“I think it’s not possible that the South Korean government cancels the F-X III contract with Lockheed Martin at this moment,” Kim Dae-young, secretary-general of the Korea Defense & Security Forum, a Seoul-based defense think-tank, said. “But the controversy over the F-X III competition process could become an issue again.”

In 2013, Lockheed Martin was dropped in the middle of the F-X III competition, but the DAPA reversed its decision later, upsetting then-preferred bidder Boeing.

The DAPA argued the country’s Air Force needs to operate fighters fitted with the so-called fifth-generation radar-evading capability in response to North Korea’s threats. As a result, Lockheed Martin won the $7 billion deal, beating Boeing and Eurofighter though the latter promised to provide more lenient tech transfers than did the former.

“The easiest way is to buy US radar and sensors to fit them into the KF-X jet, but that may provoke a backlash,” Kim said, referring to the alleged trauma of the T-50 development.

Lockheed Martin helped South Korea develop the T-50 Golden Eagle supersonic trainer jet. But key technologies are protected by the US, so integrating European radars into the plane or exporting the jet to other countries requires US approval.

“This case should serve as an opportunity for us to look back on the way we treat the US,” said Rep. Kim Jung-hoon of the ruling Saenuri Party. “We decided to buy fifth-generation fighter jets with the expectation of sophisticated tech transfer, but now we have nothing.”
 
"Skeptical Politicians Keep Embattled KF-X Alive"
Oct 29, 2015 Bradley Perrett | Aerospace Daily & Defense Report


Source:
http://aviationweek.com/defense/skeptical-politicians-keep-embattled-kf-x-alive

SEOUL — South Korea’s KF-X fighter program survived encounters with the national executive and legislature this week, but remains more politically controversial than ever. The project’s funding for 2016, and therefore its future, is not yet assured.

A senior politician has raised the possibility of replacing KF-X with an effort to upgrade the Korea Aerospace Industries (KAI) T-50 family of training and light attack aircraft, an idea that appeals to many engineering managers but not the air force.

KF-X, Turkey’s TF-X and India’s Advanced Medium Combat Aircraft are the key new opportunities for suppliers of combat aircraft engines and systems, with the South Korean project looking closest to becoming a realty. But it may be losing that lead because of repeated failures to move beyond preliminary development.

The project may not live into 2016, despite a decision of the defense committee of parliament on Oct. 29 to recommend the 67 billion won budget for KF-X proposed by the finance ministry, enough for more preliminary development.

The day before making its recommendation, the committee had warned the Defense Acquisition Program Administration (DAPA), that the money would stop—and so the program would be scrapped—if that organization, the defense ministry’s purchasing agency, failed to present a convincing and comprehensive report on KF-X.

DAPA has evidently been convincing enough, for the moment at least.

But the committee’s expressed skepticism means that parliament may not rubber-stamp the spending recommendation, as it ordinarily would. The committee’s chairperson, Chung Doo-Un, is calling for the Board of Audit and Inspection to review the program. And the committee has told DAPA to present another report, this time explaining to parliament how the country can develop the aircraft following Washington’s refusal to share technologies for integration of key systems.

The U.S. decision, based on protecting military secrets, threatens to cut off the KF-X from U.S. weapons, which the air force wants. European or Israeli companies could supply and help integrate equivalent systems, such as advanced radars, but the U.S. might then refuse permission to integrate its weapons.

The U.S. refusal to share sensitive technology, predictable but somehow unexpected by South Korean politicians and the public, has plunged KF-X into controversy.

Chung says the country should instead develop the long-proposed F-50, a fighter version of the T-50. That idea appeals to many people in the industry as far less risky and more achievable than building an all-new twin-engine fighter as big as the Eurofighter Typhoon. In 2014 KAI proposed a single-engine KF-X of about the size of the Lockheed Martin F-16.

The air force wants a large aircraft in part so it can carry heavy air-to-surface missiles, industry officials say.

Before the committee meeting, DAPA also had to report to President Park Geun-hye. She also could stop the program, by executive order. But instead she told DAPA on Oct. 27 to “make it succeed within the planned project period.” It not clear what that period is now.

Park has a reputation for populism.

“With KF-X she will probably be like a Roman emperor at a gladiator tournament,” says a South Korean engineer who has been close to the program. “If the crowd screams ‘life,’ she will let it live. If the project becomes more unpopular and controversial, and the crowd screams ‘death,’ then she will turn her thumb down and kill it.”

[Editor's Note: This story was amended to say that the KF-X program may not live into 2016, despite a funding recommendation by the defense committee of parliament.
 
Mach42 said:
That's the Philippines your talking about.
http://www.flightglobal.com/news/articles/kai-philippines-to-enter-final-negotiations-for-12-fa-50s-381688/

Regarding the FA-50: http://news.yahoo.com/philippines-gets-1st-fighter-jets-decade-amid-sea-062635863.html
 
https://www.flightglobal.com/news/articles/kai-ptdi-sign-kfx-development-contract-420629/
 
http://www.janes.com/article/59718/hanwha-thales-selected-to-develop-aesa-radar-for-korean-fighter-aircraft

http://www.janes.com/article/60743/south-korea-selects-ge-as-kfx-engine-provider
 
https://www.flightglobal.com/news/articles/seoul-sets-timeline-for-indigenous-kfx-aesa-develop-428366/
 
KF-X Development continues. The agreed upon strategy of 4 Classifications of development item procurement (determined by analysing Technology levels and vendor interest) continues.
[list type=decimal]
[*]Complete Korean Development and Production
[*]Korean Development (Korean IP) with Foreign Technical Assistance
[*]Korean Licensed Production (Non-Korean IP) of Foreign Systems
[*]Direct Import of Foreign Products
[/list]
http://cesa.aero/en/index.php?option=com_content&view=article&id=199:cesa-signs-a-contracrt-with-emk&catid=28:noticias&Itemid=31
http://cesa.aero/en/index.php?option=com_content&view=article&id=200:kfx-arrestor-hook&catid=28:noticias&Itemid=31
http://ir.triumphgroup.com/phoenix.zhtml?c=61870&p=irol-newsArticle&ID=2250831
http://texstars.com/texstars-to-develop-canopy-and-windshield-transparencies-for-the-kf-x-fighter

The latest iteration of the KF-X is the C107 design which has lots of little aerodynamic improvements and has gone to F-22 style vertical tails over the previous F-35 style ones. (The nose IRST is also a big change)

Further images @ this blog:
  • http://blog.naver.com/jhst3103/221026547190
  • http://blog.naver.com/jhst3103/221026834942
 

Attachments

  • 1.jpg
    1.jpg
    124.3 KB · Views: 619
  • 1496929983.jpg
    1496929983.jpg
    54.3 KB · Views: 220
  • 7.jpg
    7.jpg
    409.5 KB · Views: 112
  • 6.jpg
    6.jpg
    406.7 KB · Views: 85
  • 5.jpg
    5.jpg
    357.8 KB · Views: 534
  • 4.jpg
    4.jpg
    400.6 KB · Views: 566
  • 3.jpg
    3.jpg
    265.5 KB · Views: 570
  • 2.jpg
    2.jpg
    239.6 KB · Views: 578
Some left out slides due to attachment limit.

Also, to not make a competently content-less double post, here is the projected size/volume of the KF-X's IWBs. (Last Attachment)

The GBU-30 is a typo, we're talking about SDB-1s. I'm guessing (More hoping) that 6 purpose made BVRAAMs with folding fins may fit inside, but I'm not that enthusiastic about the chances.
 

Attachments

  • ??? (3).png
    ??? (3).png
    817.3 KB · Views: 198
  • 15.jpg
    15.jpg
    219.5 KB · Views: 192
  • 14.jpg
    14.jpg
    302 KB · Views: 104
  • 13.jpg
    13.jpg
    351.5 KB · Views: 88
  • 12.jpg
    12.jpg
    325.2 KB · Views: 92
  • 11.jpg
    11.jpg
    371.2 KB · Views: 97
  • 10.jpg
    10.jpg
    312.6 KB · Views: 90
  • 9.jpg
    9.jpg
    337.4 KB · Views: 116
  • 8.jpg
    8.jpg
    347.2 KB · Views: 142
I'm not an expert, so this is what i undestand and if someone more knowledgeable can correct me it is good:

The basic scheme for a control law follows the linear equations of motion that have this basic form:

acceleration = AircraftCharacteristics*AircraftState + ControlSurfaceCharacteristics*ControlSurfaceState

In this equation, you want to compute AircraftState (It is an Ordinary Differential Equation) . All ~Characteristics matrices are known. Acceleration is measured (by sensors on the plane). ControlSurfaceState is computed too.

And thus the control laws try to resolve this:

Desired performance = SomeCoefficient*AircraftState <---- (You computed AircraftState above. The SomeCoefficient matrix is known)

The desired performance can be a state (an AOA, a sideslip angle etc..) or an acceleration (pitch,roll rates for example).

In classical non adaptive control laws, ControlSurfaceState, which is the position of the control surface, is a lookup table, that is you have hard coded how much you must move your control surfaces to match a performance.
There're two problems with that:

1/ There're non linear aero and mechanical processes that can't be predicted with enough certainty. For example asymmetric vortex burst, or aeroelastic deformations of the structure. Thus, classical non adaptive models can't exploit the performance of the aircraft to the max because the lookup table value must be a safe estimation so necessarily under performing.

2/ You need a fairly specialized ControlSurfaceState matrix to account for your aircraft dynamics, which means if you want to transfer your control law to another type of aircraft..say a VTOL version, you need to rewrite the ControlSurfaceState matrix.

In NDI control law, the ControlSurfaceState matrix in computed in real time (by inverting the ControlSurfaceCharacteristics matrix, hence the name Dynamic Inversion. That means you are trying to find the ControlSurfaceState (unknown) matrix that gives the ControlSurfaceCharacteristics matrix (known), of course accounting for your AircraftCharacteristics matrix(also known)). Thus you can adapt the control surface deflections to non linear processes.

Since the surface motion is computed in real time according to you Aircraft and ControlSurface Characteristics matrices, if those latter incorporate non linear quantites like aerolestatic, or cross coupling (usage of the same surface for different functions) then the ControlSurfaceState computation will be even better adapted to non linearities e.g destruction of some part of the control surfaces, or high AOA dynamics etc..
Future fighters will likely use aeroelastic active wings or be tailless. In those case NDI like control laws allow for much better adaptation.

And if you want to transfer the data to another aircraft, all you have to do is feed in new ~Characterstics matrices and the inversion will find the required control surfaces motions.

So, from there, you understand why it was selected for F-35, and also T-50, which is a training aircraft with the functionality of simulating different aircrafts flight controls (hence you just have to put different ~Characteristics matrices)

However NDI is not perfect. Since the inversion is dependent on the ~characterstics matrices, that is the modeling of the aircraft, and that no model is perfect, there're errors and the inversion can't fully compute the necessary control surface motions needed to match a performance. Be it classical or NDI systems, you have a loop that compares the actual performance with the desired one. The FCS will continue to move the control surface until the performance is matched. But the FCS needs to know how to move it, so that brings us back to the ControlSurfaceState inversion. We need something to improve it in those cases.
To this effect, they are usually complemented with neural networks that statistically approximate the fully non linear equations relating the desired performance to the control surface motions. Neural Network have the great advantage of learning the relationship between performance and control surface motions in real time so they are great tools for that.



Just a vocabulary remark: While NDI treats with non linear processes, the actual equations used are linear.
 
IR Stealth: <a href="https://www.nature.com/articles/s41598-017-06749-0">Selective dual-band metamaterial perfect absorber for infrared stealth technology</a>

We propose a dual-band metamaterial perfect absorber with a metal–insulator–metal structure (MIM) for use in infrared (IR) stealth technology. We designed the MIM structure to have surface plasmon polariton (SPP) and magnetic polariton (MP) resonance peaks at 1.54 μm and 6.2 μm, respectively. One peak suppresses the scattering signals used by laser-guided missiles, and the other matches the atmospheric absorption band, thereby enabling the suppression of long-wavelength IR (LWIR) and mid-wavelength IR (MWIR) signals from objects as they propagate through the air. We analysed the spectral properties of the resonance peaks by comparing the wavelength of the MP peak calculated using the finite-difference time-domain method with that obtained by utilizing an inductor–capacitor circuit model. We evaluated the dependence of the performance of the dual-band metamaterial perfect absorber on the incident angle of light at the surface. The proposed absorber was able to reduce the scattering of 1.54 μm IR laser light by more than 90% and suppress the MWIR and LWIR signatures by more than 92%, as well as maintain MWIR and LWIR signal reduction rates greater than 90% across a wide temperature range from room temperature to 500 °C.

This work was partially supported by the Low Observable Technology Research Center Program of the Defense Acquisition Program Administration and the Agency for Defense Development, Korea.
 
Control laws aren't really based on the classic Newtonian equations of motion, at least in the way most people conceive them, they're more 'how do I do this' than 'my new speed will be'. For instance the pitch control laws for the 777 are based on an algorithm that's dominated by C*U, where C* is an expression of pitch rate, and U is the desired speed. Too slow, lower the nose, too high, raise the nose. There's no 'my speed is X, therefore I must accelerate by Y for Z seconds' - it's goal seeking, rather than pre-calculating changes.

More often than not it's the rates of change that are interesting, or even rates of change of rates of change, with commanded control surface deflections calculated to smooth out perturbations and drive the aircraft towards the desired state. Nor are the algorithms purely linear, they're frequently discontinuous across multiple flight regimes, for instance the rudder will be driven by one algorithm on the ground, another in the air, and might be locked out entirely above a certain airspeed, and it's entirely possible to smooth the transition across the discontinuities.

The idea of tidy matrices of control surface states and characteristics doesn't really match any system I've worked on, though it's possible it exists at the control law design stage. By the time it gets to us for implementation, it's more a series of algorithms for the major data items, C*U being an example. And of course that's only one part of the software, there's a whole separate set of functionality categorised as 'the autopilot', plus a hugely complex set of data comparisons going on before the flight data ever gets to the control law algorithms to decide if the data is reliable, and what to do with it if it isn't.

None of the systems I've worked on have include neural networks, that's much more a feature of experimental systems, not currently operational ones (with the F-35 as a possible exception). Incidentally there's a pretty good article on the F-35 control laws here that talks about them from the pilot's perspective:
http://www.codeonemagazine.com/article.html?item_id=187 I think the author does a good job of explaining that goal seeking behaviour and how that lets us build and fly aircraft that are dynamically unstable.
 
Hello DWG,

DWG said:
Control laws aren't really based on the classic Newtonian equations of motion, at least in the way most people conceive them, they're more 'how do I do this' than 'my new speed will be'. For instance the pitch control laws for the 777 are based on an algorithm that's dominated by C*U, where C* is an expression of pitch rate, and U is the desired speed. Too slow, lower the nose, too high, raise the nose. There's no 'my speed is X, therefore I must accelerate by Y for Z seconds' - it's goal seeking, rather than pre-calculating changes.

I think we are not talking about the same thing. what you are describing is the control command, that is the stability objective of the FCS. it is C*U for the 777, for most fighters it is either a pitch rate/AOA at low speeds, a G command at high speed.
To achieve this goal, an augmented system (an aircraft with a computerized FCS) will have input(pilots command)/output(achieved parameter). For the FCS to know how to move the control surfaces in relation to actual conditions, you need a model; and that model is based on the combination of the equations of motion for a rigid body and a state/output equation (the two equations that i described).
All my sources (research, graduate, engineering) clearly describe the FCS to be modeled after equations of motion(1,2). The F-35 CLAW is explicitly modeled after those equations too(3).
What changes between it and a linear quadratic system is the way the control surface movement needed is computed(1(chapter 4),3).



More often than not it's the rates of change that are interesting, or even rates of change of rates of change, with commanded control surface deflections calculated to smooth out perturbations and drive the aircraft towards the desired state. Nor are the algorithms purely linear, they're frequently discontinuous across multiple flight regimes, for instance the rudder will be driven by one algorithm on the ground, another in the air, and might be locked out entirely above a certain airspeed, and it's entirely possible to smooth the transition across the discontinuities.

Linearity here refers to its theoritical definition, that is, superposition and proportionality and its application to FCS is understood to model control surfaces motion within a given speed/altitude frame as being purely linear in their output. This is not the case for high AOA, when aeromecanic coupling, surface saturation etc... In those cases the actual control surface motion to output relationship is not predictable. NDI system, by computing the control surface motion needed, taking into account the non linear evolution of the control surface efficiency matrix handles those kinds of non linearity.
I think you may be talking about gains, which are modified filters after the control surface needed is computed.

The idea of tidy matrices of control surface states and characteristics doesn't really match any system I've worked on, though it's possible it exists at the control law design stage. By the time it gets to us for implementation, it's more a series of algorithms for the major data items, C*U being an example. And of course that's only one part of the software, there's a whole separate set of functionality categorised as 'the autopilot', plus a hugely complex set of data comparisons going on before the flight data ever gets to the control law algorithms to decide if the data is reliable, and what to do with it if it isn't.
If that is not too personal, i'd be interested in knowing what your job consists in. As i've said, this is not my line of work and i'm only discussing based on the documentation i find :)

None of the systems I've worked on have include neural networks, that's much more a feature of experimental systems, not currently operational ones (with the F-35 as a possible exception). Incidentally there's a pretty good article on the F-35 control laws here that talks about them from the pilot's perspective:
http://www.codeonemagazine.com/article.html?item_id=187 I think the author does a good job of explaining that goal seeking behaviour and how that lets us build and fly aircraft that are dynamically unstable.
I have probably mistaken a research paper developing an online NN based on the F-35 FCS for the operational one. my bad.


(1)Tian,L. 1999. A study of nonlinear flight control system designs. Iowa State University.
(2)Campbell S.F and Kaneshige J.T. 2009. A Nonlinear Dynamic Inversion Predictor-Based Model Reference Adaptive Controller for a Generic Transport Model. IEEE
(3)Nixon D.W 2004 Lockheed-MartinFlight Control Law Development for the F-35 Joint Strike Fighter. miadc presentation
 
I was a software engineer on the 757RT, 777 and Eurofighter flight control systems, though I did very little coding on Eurofighter, being primarily focused on change control and liaising with customer QA. And of course the F-14 and Nimrod MRA.4 teams were just along the corridor.

WRT who is talking about what, we may simply be having a philosophical difference through coming at the process from opposite ends. The control laws are a set of mathematical models that model how the aircraft should respond to pilot and air data inputs. Everything comes down to Newtonian mechanics, but in terms of describing how the system works I'm not sure that's a useful explanatory tool. We can explain, for instance, why AoA is significant in terms of the basic laws of motion, but people get a much clearer level of understanding if you increase the abstraction and talk about the competing effects of lift and drag, and how that can be channelled into speed and or altitude changes.

By the time the model gets to the implementation team, it's a set of written equations for us to turn into code - you might think of the difference between a structural engineer running stress equations to design a part, and the final notation on the design that just lists dimensions. We don't need to know the discrete modelling that's gone into stuff like eliminating flutter frequencies, but it's still there in the model. Incidentally, thinking of them as simply quadratics is probably misleading , there's an awful lot of logic (Weight on Wheels being the obvious one, and even that's much more complex than people imagine), ranges, smoothing, and so on. Sufficient that we broke at least one Ada compiler on 777 which couldn't cope with 130 nested or-else statements (it sticks in the mind as I got to write the error report for that one and chase the fix).

The control laws are essentially what's described on pages 11 and 13 of the Nixon presentation, the big Simulink/stateflow model with hundreds of inputs and outputs and thousands of internal blocks, the difference (apart from the different control law philosophy) is that we wrote the code manually from a written description of the model, while they were automatically generating it direct from the model. While automatic code generation is clearly 'efficient', it does introduce another element that needs to be verified, the code generator, and you can see from some of Nixon's comments that that still had some rough edges to be worked out. I wish he'd included more on testing and formal verification as it's not clear to me that theirs is as robust as ours was.
 
Thanks for your reply.

Durham et al (2016) include a matlab model based on CLAW. Documentation page 242 describes the AERO_OBM block and one can see the equations again and they are used directly by the implementation.

So going back to the Nixon presentation, i guess i'm talking about the Control Laws block inside the general model. Since it is full of acronyms i can't totally understand everything (appart from the Controls Laws, Gain data loops).

If i understand correctly you were responsible with the actual programming of the algorithms. But you never had to implement such equations? I mean, they are in matrix form, so already adapted to most programming languages. At which step or over which block you would have coded the model?

Very interesting in any case.

Durham W, Bordignon K.A and Beck R. 2016. Aircraft Control Allocation. Wiley publishing.
 
It's just plain mathematical equations for each output, with associated logic to tell you what to do in which part of the flight envelope, often with associated filtering and smoothing for either input or output. With Weight on Wheels that ended up being about a dozen lines of logic to combine half a dozen inputs just to decide if we were actually on the ground or not. (That's technically the Mode Logic part of Dixon's central model, but IMO you can't usefully separate it from what he labels Control Laws or Gains because the operations on any individual data item are a combination of all three.) Similarly for every other piece of data we're processing.

I'm puzzled by your assertion that matrices are adapted to programming languages, combined with the apparent assertion that therefore doesn't require implementation. I'm an Ada specialist, but I've worked in most of the engineering languages at one time or another. Most have data structures that will handle matrices,to some extent or other, but you still have to define that data structure in code, still have to get the data into it, with associated filtering and smoothing, and then you have to code every mathematical operation that happens to that data (and if you're using a sensible language, ensure that every variable the data passes through is dimensionally and range consistent - remember Mars Climate Orbiter). And it's been a while, but I really don't remember any significant use of matrices, mostly just individual variables for stuff like ComputedAirspeed or PitchAngle or WeightOnLeftMainGear. I suspect you're being confused by the differences between a Matlab model, a requirements document, and actual operational flight code.
 
Hello DWG,

If i'm not mistaken, you say that the actual code doesn't contain the control allocation equation that you can find in Durham, Nixon and other sources. This what i don't really understand. How would you code the control allocation then (how much you need to move a control surface to reach the desired performance)?

As for the languages, i had in mind C,C++ and fortran that are pretty well adapted (and efficient) to matrix algebra. i said "most" because i guessed that you probably coded in ada and i don't know a single thing on that language :)
 
It absolutely does contain the equations to drive the control surfaces, that's what I've been saying all along, but they might be laid out something like:

LeftAileronDisplacement := CombineLAileronRollAndPitchDemand
(RollDemand : SmoothedRollDemand,
PitchDemand : SmoothedPitchDemand);

Where CombineLAileronRollAndPitchDemand is a function that takes RollDemand and PitchDemand as parameters from the variables SmoothedRollDemand and SmoothedPitchDemand,compares them with current Roll and Pitch, and current Roll and Pitch Rates, and works out the aileron offset based on combining all of those as the control law equation defines, together with logic to suppress stuff like pitch oscillation and control force fights, all while staying within the allowable maximum Roll and Pitch rates, and adjusting the offset so it the aircraft doesn't overshoot the demanded Pitch and Roll once it gets there. So within that function you might have stuff like:

AdjustedRollDemand := LimitRollRate(DemandedRoll : RollDemand,
ActualRoll : Current Roll,
RollRate : CurrentRollRate,
MaxRollRate : CurrentRollRateLimit);

and

CombineLAileronRollAndPitchDemands := CombineRollAndPitchDemands
(RollRate : AdjustedRollRate,
PitchRate : AdjustedPitchRate);


which are themselves function calls to code that works out the actual roll rate to aim for, and the aileron offset that corresponds to the adjusted pitch and roll rates. And a similar function will be doing the same for the right aileron, but there the roll rate will adjust the aileron demand in the opposite sense.

The actual roll calculation within AdjustedRollDemand might look something like :

If (DemandedRoll /= CurrentRoll) then
If ((DemandedRoll-CurrentRoll) > (CurrentRollRate)) then
CurrentRollRate := IncrementRollRateTowardsMax (RollRate : CurrentRollRate.
MaxRoll : CurrentMaxRollRate);
Else CurrentRoleRate := DecrementRollRateToPreventOvershoot (RollRate : CurrentRoleRate,
Current Roll : Current Roll,
DemandedRoll : DemandedRoll)
Endif;
Endif;

Though the whole thing would be somewhat more complex as you need the code to function whichever direction you're adjusting the roll demand and the nested if statement doesn't account for that, but is probably close enough to get the idea. You would also need to pay attention to data types to ensure they're dimensionally consistent. And that adjusted roll rate would then be used to determine the magnitude of aileron displacement, whether by a table look up, or proportionally.


It's all straightforward arithmetic/geometry with no need to involve matrices/arrays. You can do it with arrays of data, but there's no mathematical reason it has to be done that way, and, IMO, the legibility of the programme is much improved this way, making it far easier to spot errors.

As for Ada, see above. It's at least as powerfully as any of the others at matrix/array operations, but that doesn't mean they're necessarily the best way to lay out your data. Ada looks much like Pascal, with strong typing as opposed to C's weak typing, which IMO shouldn't be used on safety critical projects. In fact the MISRA C standard referred to in the Nixon presentation actually starts 'We really don't think you should use C for safety critical code, but if you must...' And that was for motor industry levels of safety (MISRA = Motor Industry Standards and Research Association if memory serves). C (and Matlab AIUI) will happily let you add a variable representing a value in feet to one in degrees, which is nonsensical, whereas properly written Ada will spot that at compile time. F-35 chose to use C, but that was solely because there are more C programmers available than Ada, but that choice actually made their development less safe, by introducing the potential for a whole class of errors it's impossible to make in Ada, and inherently by C being less readable and more cryptic than Ada.
 
Thanks for the detail dwg :)

Do not focus on the matrix form, i said that because i'm used to work with matlab C++ and fortran but i tend to forget that i'm not doing real time computations :)

Thanks again for the explanations.
 
A quick question for DWG; do you know if they're using neural networking in more FCS? I remember one of the hypersonic wave rider designs that was flight tested used that for low speed flight control, since it was reportedly such a difficult design to control in that part of the envelope. They felt it was better if it figured out the best way to fly itself, as opposed to telling it what it's control laws should be, at least explicitly, anyway.
 
Any news on the latest configuration ?
 
Deino said:
Any news on the latest configuration ?

More details being revealed at ADEX/Seoul Air Show would be a good bet. Apparently both the F-22 and F-35 have flight demos so there's no way I'm not going.
 
Sundog, I'm effectively retired and haven't worked on FCS development since the early 2000s, so I'm not completely up to date on stuff that doesn't make it into Flight or AvLeak, but within those limits it's plain there are a lot more FCS development projects due to the proliferation of UAVs, and also that that same proliferation allows much more experimental work. When you can throw an airframe together in a few months, and fly it without risking a test pilot, it becomes much easier to justify, and execute, experimental projects with radical configurations of both airframe and software. I'd be surprised if there aren't quite a few projects looking at neural-nets, but what would really interest me is whether anyone has managed to certify one, or even worked out how to approach that problem. A neural-net is a moving target WRT traditional testing, and if you can't demonstrate complete testing, how do you certify it?
 
Good overview and news on KFX.

http://aviationweek.com/defense/south-koreas-kf-x-grows-considerably-development
 
Latest iteration ... maybe !?
 

Attachments

  • Korea KFX - 201710.jpg
    Korea KFX - 201710.jpg
    83 KB · Views: 701
ROKAF Pilots are being sent to United States Naval Test Pilot School (not USAF for some reason?) to become certified Test Pilots and Flight Test Engineers for the KF-X Programme.

http://kookbang.dema.mil.kr/kookbangWeb/view.do?ntt_writ_date=20171120&parent_no=5&bbs_id=BBSMSTR_000000000006

Off topic: Why is the KF-X moved to the Postwar Subforum?
 
The USN Test Pilot School takes foreign students, I'm not clear if the USAF School does or not.
 
Mach42 said:
ROKAF Pilots are being sent to United States Naval Test Pilot School (not USAF for some reason?) to become certified Test Pilots and Flight Test Engineers for the KF-X Programme.

http://kookbang.dema.mil.kr/kookbangWeb/view.do?ntt_writ_date=20171120&parent_no=5&bbs_id=BBSMSTR_000000000006

Off topic: Why is the KF-X moved to the Postwar Subforum?

As a follow up: Aparently the ROKAF has been having exchanges with the

https://en.wikipedia.org/wiki/U.S._Air_Force_Test_Pilot_School

since 2014 as well: http://english.yonhapnews.co.kr/search1/2603000000.html?cid=AEN20180422002000315

Also KF-X Program news:

Structural test facility built for the future fatigue/durability testing of KF-X ground-test article.
http://www.yonhapnews.co.kr/bulletin/2018/05/31/0200000000AKR20180531120600003.HTML?from=search

KF-X flight software being tested on KAI-owned FA-50 test plane. Things like Auto-GCAS, Automatic way-point navigation, Automatic terrain following (Things that KAI hasn't developed for the FA-50 that the KF-X needs to be a front-line fighter)
KF-X 선진자동비행제어 기술실증 초도비행 성공
5월 11일 KF-X 선진자동비행제어 기술실증기의 초도 비행이 성공적으로 수행되었다. 본 사업은 항공기계통실 KF-X제어체계팀 주관으로 KF-X에 적용하기 위한 자동지상충돌회피, 자동자세회복, 자동경로항법 그리고 자동지형추적 기능과 같은 선진자동비행 기술을 독자 개발, FA-50 시제 2호기를 통해 사전에 기술 검증하는 사업이다. 자동지상충돌회피 기능은 조종사의식상실, 조종착각 등의 상황에서 항공기의 지상 충돌을 자동으로 회피하는 기술이며, 자동지형추적 기능은 항공기가 적의 레이더망에 탐지되지 않도록 초저고도로 지형을 추적하며 비행하는 기능이다. 또 자동자세회복 기능은 조종사의 버티고 (VERTIGO, 비행착각) 상황 등에서 항공기를 안정적인 자세로 회복하는 기능이고, 자동경로항법 기능은 자동으로 3차원 경로점을 비행하는 기능이다. KF-X 선진자동비행제어 기술실증 사업은 올해 7월까지 총 19쏘티(SORTIE, 단독 항공기 출격 횟수)의 비행시험이 계획되어 있고, 내년 4월까지 설계 성숙도를 높여 기술준비수준(TRL)을 향상시킬 예정이다.
http://kaiwebzine.com/webzine_2018_06/2476

AESA Radar development seems to be moving along
http://english.yonhapnews.co.kr/search1/2603000000.html?cid=AEN20180531002900315

And this doesn't seem very relevant but it's interesting
http://english.yonhapnews.co.kr/search1/2603000000.html?cid=AEN20180605004700315
 
Indonesian shares on KFX. Excerpted from MoD regulations on KF/IF-X development. parts written in Indonesian are not important.

Problem from Indonesian side is related to attitude in our govt as some consider the project is fruitless while we need to keep paying. ON the academic side, we wish for more control which kinda absurd. However it difficult to exit anyway as we already poured funds into manufacturing facility for the fighter. We also already purchased some manufacturing equipment like autoclave and others. The facility is ready and some part already used to help develop our local N-219 light aircraft. There is also concern related to the number we would acquire that some consider 40-50 aircrafts just won't make break even points.
 

Attachments

  • 34963414_472560366547246_4328075918984609792_n.jpg
    34963414_472560366547246_4328075918984609792_n.jpg
    54.3 KB · Views: 1,086
Design of the KFX :

https://www.defensenews.com/air/2018/06/29/south-korea-unveils-first-images-of-kf-x-design-with-european-missiles/
 
Would be interesting to see how much Power Aperture Product the KFX AESA would have. To make use of 160-180km range of Meteor the design must feature at least 10 kW (like ECR-90) or more if the aperture area is less.

Hypothetical AESA with 1000 modules, 10 W power for each module and 25% duty cycle allows 260km Detection range for 50% probability and lock on/ tracking range (R90%) of 160 km. Against 3 sqm target.
 

Similar threads

Back
Top Bottom