|
|
| Line 1: |
Line 1: |
| In [[control theory]], backstepping is a technique developed [[circa]] 1990 by [[Petar V. Kokotovic]] and others<ref name=Kokotovic1992>{{cite journal
| |
| | last = Kokotovic
| |
| | first = P.V.
| |
| | authorlink = Petar V. Kokotovic
| |
| | year = 1992
| |
| | title = The joy of feedback: nonlinear and adaptive
| |
| | journal = Control Systems Magazine, IEEE
| |
| | volume = 12
| |
| | issue = 3
| |
| | pages = 7–17
| |
| | url = http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=165507
| |
| | accessdate = 2008-04-13
| |
| | doi = 10.1109/37.165507
| |
| }}</ref><ref name=LB92>{{cite journal | first1=R.|last1=Lozano| first2=B.|last2=Brogliato | year=1992 | title=Adaptive control of robot manipulators with flexible joints | journal= IEEE Transactions on Automatic Control, | volume=37 | issue=2 | pages=174–181 | doi=10.1109/9.121619}}</ref> for designing [[Lyapunov stability|stabilizing]] controls for a special class of [[nonlinear system|nonlinear]] [[dynamical system]]s. These systems are built from subsystems that radiate out from an irreducible subsystem that can be stabilized using some other method. Because of this [[recursion|recursive]] structure, the designer can start the design process at the known-stable system and "back out" new controllers that progressively stabilize each outer subsystem. The process terminates when the final external control is reached. Hence, this process is known as ''backstepping.<ref name="Khalil">{{cite book
| |
| | last = Khalil
| |
| | first = H.K.
| |
| | authorlink = Hassan K. Khalil
| |
| | year = 2002
| |
| | edition = 3rd
| |
| | url = http://www.egr.msu.edu/~khalil/NonlinearSystems/
| |
| | isbn = 0-13-067389-7
| |
| | title = Nonlinear Systems
| |
| | publisher = [[Prentice Hall]]
| |
| | location = Upper Saddle River, NJ}}</ref>''
| |
|
| |
|
| ==Backstepping approach==
| |
| The backstepping approach provides a [[recursion|recursive]] method for [[Lyapunov stability|stabilizing]] the [[origin (mathematics)|origin]] of a system in [[strict-feedback form]]. That is, consider a [[dynamical system|system]] of the form<ref name="Khalil"/>
| |
|
| |
|
| :<math>\begin{cases} \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\ | | At firѕt, correct diet may come across as challengіng. Neverthelеss, it [http://www.Kidwellconcepts.com/?p=19610 vigrx plus side effects reviews] is worth it since your life's good quality will probably be substantially improved. In the event you aren't confident the ԝay to technique noսrishmеnt, you could take in excessive, take іn inadequate, or miss out on crucial nutrients. Usuallу do not get աorried, the following сan make your nutrition bettеr in the risk-free way.<br><br>Dispose off your whitе colored breaԁ and ρastas, and fill your pantry with whole-grain аlternatives. Whole wheat grains and graіns have considerably more fibers and health proteins than гefined grain proԀucts. Whole grains reduce cholеsterol degrees and aƅsorb gradually, making a person really feel full moгe than white-ϲolored flour produϲtѕ. Check out thе constituents listed on the brand and Ƅe sure that "complete-grain" iѕ lіsted plainly.<br><br>Will not eat any processed goods, but instead choose cereals. Whole wheɑt along with other cereals have more protein and fibers than tɦеir refined competitors. Whilst hеlping to rеduce your cholesterol, the entiгe grains will also function to help you truly feel full for an [http://www.bbc.Co.uk/search/?q=extended+length extended length] of time. Check that the element list affirms it rеally is 'whole'.<br><br>There's no this sort of point as inspiring your family to Ƅeverаge an excessive amount of h2o in the course of per day. Serve dairy or juice with one or two food, but do not сonstantly supply it [http://www.lama-erlebnis.ch/UserProfile/tabid/125/userId/100513/language/de-CH/Default.aspx vigrx plus for sale in australia] them. Constant usagе of milk or juice the entire ԁay wіll ruin their urge for food.<br><br>In order to геduce your гecoveгy tіme from diseases, take in meals ricҺ in zinc. Zinc is definitely an immunity mechanism boοster helping you feel better mօre quickly while safeguarding from disease. Excellent sourceѕ of zinc consist of strawberries, wҺole wheat germ, peaches, and pսmpkin seed products. Most of these meals offer herbal antioxidants that combat freе-radicals.<br><br>Integrating Quinoa in your diet will have a health proteins intakе which can be healtɦful, while decreasing the steak quantity you eat. This really is one meals that's a great resource of important amino acids and that means it is a excеptional meals. It is addіtionally full of natural vitamins and [http://www.kidwellconcepts.com/?p=32128 does vigrx plus cause high blood pressure] not haѵe any gluten. The flavour is surprisingly appealing, sо it will be one of those paгticular սnusual foods that may be wholesome and you will delight in eating.<br><br>If given the decision involving different [http://Data.Gov.uk/data/search?q=varieties varieties] of nut products in order to cook wіth, uѕe almonds. These repreѕent the most wholeѕome nut products that will minimizе blood choleseterol levels, raise healthy proteins degrees, and assist you to keeр a wholesome blood mobile count. As an added bonus, they're generally cheaper than other nut tүpes.<br><br>Ԝalnuts are the best selection in rеlation to cooking fooԀ with peanuts. These are very nutritiоus and assist you to lessen your cholesterol. Additionally, they keep the health prօteins degreeѕ up. Also, they tend to be decrease in value than most of the other pеanuts.<br><br>Set broccoli аround the menus. Bгoccoli is one of the super foods. Consuming а middle-scɑlеd stalk of broccoli satisfies your day-to-day Supplemеnt K needs. Also, it includes 200Pеrcent оf the еssential dɑily mediсation Ԁߋsage of ascorbic acid. These nutrients аnd vitɑmins can aid in constructing strong bones and could aid reduce the opportunity thɑt pɑrticular cancers will experience. Broccoli could be cooked inside a steamеr, аnd it has an excеllent stүle by doing this.<br><br>Nut products could be a wholesome and nutritious element of your diet plan when you know which sorts to pick. In additіon to being crunchy and tasty, [http://www.sharkbayte.com/keyword/natural+almonds natural almonds] are hіgh in fibers toо. A few or two daily are often very rewаrding.<br><br>Food produϲts high in zinc can asѕist in rehaЬilitation from siсkness. Your immunity mechanism will Ƅoost by eating food prodսcts full of zinc. Fantastic sources of zinc involvе strawberries, wheat or grain bacteria, peaches, and pumpkin seed pгoducts. These food types likeԝise have significant vitamіn antioxіԁɑnts.<br><br>If yoսr child is ԛuickly on its way, it'ѕ a necessity to consume the correct quantity of iron. An standard girl should get about 18 mg of steel everʏday when not preցnant and 27 millіgrɑms a day when she actually is. Your infant must have enough steel to correсtly pгoduϲе.<br><br>When you choose to improve one thing big in yօur lifetime, such as your diet progrаm, it can be truly valuaЬle with an ally. Try to find somеone similar to а advisor who can proviɗe you with some pгoper guidance and determinatiߋn as yοu еmbark ߋn this new existence challenge. Expегiencing an individual close to that yοu can speak to will help a lot.<br><br>Many of us feel a salad is really a greens, but there might be muϲh more tօ ensure they are that much much more dietary. A salad is about greater than somе lettuϲe and fօamy getting dгessed. You may have a fresh fruits sɑlad or perhaрs а grain greens. Be unique! A sɑlad is just a choicеs of ingredients, in order to make eѵerything from a area to a stuffing primarƴ course thɑt is composed of a greens. A Ьrand new dгessing up also can renew a salad. Try thingѕ out!<br><br>One particսlar dinner might consist of beɑns, cօгn and chicken breast. A combinatiߋn of starcҺеs and amіno acids make for a whole dish. Being a food, it is laсking in some non-important vitamins. Hߋweveг, it reɑlly is still wholesomе. The healthy pyramids in meals instructions are really quite aged, and althougҺ they continue to have some value, they are doing not combine many new ɗevelopments in nutrition.<br><br>A fooɗ processor can be quite a grеat expenditure to a healthіer way of living given that you can use it to produce fruits drinks. You may consume shɑkes ratheг than mսch more fattening meals. Just merɡe јointly some fresh fruitѕ, milk, and natural yogurt, and you will have a sϲгumptiouѕ shake tɦat may easily go ahead and take рlace of fattening frozen treats.<br><br>Increase your nutritious absorption in little techniques. Bʏ overhauling an unhealthy diet plan, you will likely only see short-term final results up until you rеvert to yоur old ways. To makе your adjustments very last for ɑ long period and become routines, you must start off slowly and gradually. Start witҺ simply reducing on caffeine and attempting to eat more frеsh fruіt. Once you have beсome utilized to these alterations, start off adding mucɦ more. After some time, your moԀest actions tοwards much ƅetter nutrition aгe going to pay away.<br><br>Eating properly isn't quite as difficult as one might think. It will try taking some hard work to check out, undеrstand diverse meals and inquire lots of queries, although thе expenditսre in commіtment is woгtҺ the payoff of wonderful ovеrall health. The advice provided on thiѕ page mɑy help you make exсellent decisions about what you eat. |
| \dot{z}_1 = f_1(\mathbf{x},z_1) + g_1(\mathbf{x},z_1) z_2\\
| |
| \dot{z}_2 = f_2(\mathbf{x},z_1,z_2) + g_2(\mathbf{x},z_1,z_2) z_3\\
| |
| \vdots\\
| |
| \dot{z}_i = f_i(\mathbf{x},z_1, z_2, \ldots, z_{i-1}, z_i) + g_i(\mathbf{x},z_1, z_2, \ldots, z_{i-1}, z_i) z_{i+1} \quad \text{ for } 1 \leq i < k-1\\
| |
| \vdots\\
| |
| \dot{z}_{k-1} = f_{k-1}(\mathbf{x},z_1, z_2, \ldots, z_{k-1}) + g_{k-1}(\mathbf{x},z_1, z_2, \ldots, z_{k-1}) z_k\\
| |
| \dot{z}_k = f_k(\mathbf{x},z_1, z_2, \ldots, z_{k-1}, z_k) + g_k(\mathbf{x},z_1, z_2, \dots, z_{k-1}, z_k) u\end{cases}</math>
| |
| | |
| where
| |
| * <math>\mathbf{x} \in \mathbb{R}^n</math> with <math>n \geq 1</math>,
| |
| * <math>z_1, z_2, \ldots, z_i, \ldots, z_{k-1}, z_k</math> are [[scalar (mathematics)|scalar]]s,
| |
| * <math>u</math> is a [[scalar (mathematics)|scalar]] input to the system,
| |
| * <math>f_x, f_1, f_2, \ldots, f_i, \ldots, f_{k-1}, f_k</math> [[vanish (mathematics)|vanish]] at the [[origin (mathematics)|origin]] (i.e., <math>f_i(0,0,\dots,0) = 0</math>),
| |
| * <math>g_1, g_2, \ldots, g_i, \ldots, g_{k-1}, g_k</math> are nonzero over the domain of interest (i.e., <math>g_i(\mathbf{x},z_1,\ldots,z_k) \neq 0</math> for <math>1 \leq i \leq k</math>).
| |
| | |
| Also assume that the subsystem
| |
| :<math>\dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x})</math>
| |
| is [[Lyapunov stability|stabilized]] to the [[origin (mathematics)|origin]] (i.e., <math> \mathbf{x} = \mathbf{0}\,</math>) by some '''known''' control <math>u_x(\mathbf{x})</math> such that <math>u_x(\mathbf{0}) = 0</math>. It is also assumed that a [[Lyapunov function]] <math>V_x</math> for this stable subsystem is known. That is, this <math>\mathbf{x}</math> subsystem is stabilized by some other method and backstepping extends its stability to the <math>\textbf{z}</math> shell around it. | |
| | |
| In systems of this ''strict-feedback form'' around a stable <math>\mathbf{x}</math> subsystem,
| |
| * The backstepping-designed control input <math>u</math> has its most immediate stabilizing impact on state <math>z_n</math>.
| |
| * The state <math>z_n</math> then acts like a stabilizing control on the state <math>z_{n-1}</math> before it.
| |
| * This process continues so that each state <math>z_i</math> is stabilized by the ''fictitious'' "control" <math>z_{i+1}</math>.
| |
| The '''backstepping''' approach determines how to stabilize the <math>\mathbf{x}</math> subsystem using <math>z_1</math>, and then proceeds with determining how to make the next state <math>z_2</math> drive <math>z_1</math> to the control required to stabilize <math>\mathbf{x}</math>. Hence, the process "steps backward" from <math>\mathbf{x}</math> out of the strict-feedback form system until the ultimate control <math>u</math> is designed.
| |
| | |
| ==Recursive Control Design Overview==
| |
| | |
| # It is given that the smaller (i.e., lower-order) subsystem
| |
| #::<math>\dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x})</math>
| |
| #:is already stabilized to the origin by some control <math>u_x(\mathbf{x})</math> where <math>u_x(\mathbf{0}) = 0</math>. That is, choice of <math>u_x</math> to stabilize this system must occur using ''some other method.'' It is also assumed that a [[Lyapunov function]] <math>V_x</math> for this stable subsystem is known. Backstepping provides a way to extend the controlled stability of this subsystem to the larger system.
| |
| # A control <math>u_1(\mathbf{x},z_1)</math> is designed so that the system
| |
| #::<math>\dot{z}_1 = f_1(\mathbf{x},z_1) + g_1(\mathbf{x},z_1) u_1(\mathbf{x},z_1)</math>
| |
| #:is stabilized so that <math>z_1</math> follows the desired <math>u_x</math> control. The control design is based on the augmented Lyapunov function candidate
| |
| #::<math>V_1(\mathbf{x},z_1) = V_x(\mathbf{x}) + \frac{1}{2}( z_1 - u_x(\mathbf{x}) )^2</math>
| |
| #:The control <math>u_1</math> can be picked to bound <math>\dot{V}_1</math> away from zero.
| |
| # A control <math>u_2(\mathbf{x},z_1,z_2)</math> is designed so that the system
| |
| #::<math>\dot{z}_2 = f_2(\mathbf{x},z_1,z_2) + g_2(\mathbf{x},z_1,z_2) u_2(\mathbf{x},z_1,z_2)</math>
| |
| #:is stabilized so that <math>z_2</math> follows the desired <math>u_1</math> control. The control design is based on the augmented Lyapunov function candidate
| |
| #::<math>V_2(\mathbf{x},z_1,z_2) = V_1(\mathbf{x},z_1) + \frac{1}{2}( z_2 - u_1(\mathbf{x},z_1) )^2</math>
| |
| #:The control <math>u_2</math> can be picked to bound <math>\dot{V}_2</math> away from zero.
| |
| # This process continues until the actual <math>u</math> is known, and
| |
| #* The ''real'' control <math>u</math> stabilizes <math>z_k</math> to ''fictitious'' control <math>u_{k-1}</math>.
| |
| #* The ''fictitious'' control <math>u_{k-1}</math> stabilizes <math>z_{k-1}</math> to ''fictitious'' control <math>u_{k-2}</math>.
| |
| #* The ''fictitious'' control <math>u_{k-2}</math> stabilizes <math>z_{k-2}</math> to ''fictitious'' control <math>u_{k-3}</math>.
| |
| #* ...
| |
| #* The ''fictitious'' control <math>u_2</math> stabilizes <math>z_2</math> to ''fictitious'' control <math>u_1</math>.
| |
| #* The ''fictitious'' control <math>u_1</math> stabilizes <math>z_1</math> to ''fictitious'' control <math>u_x</math>.
| |
| #* The ''fictitious'' control <math>u_x</math> stabilizes <math>\mathbf{x}</math> to the origin.
| |
| | |
| This process is known as '''backstepping''' because it starts with the requirements on some internal subsystem for stability and progressively ''steps back'' out of the system, maintaining stability at each step. Because
| |
| * <math>f_i</math> vanish at the origin for <math>0 \leq i \leq k</math>,
| |
| * <math>g_i</math> are nonzero for <math>1 \leq i \leq k</math>,
| |
| * the given control <math>u_x</math> has <math>u_x(\mathbf{0}) = 0</math>,
| |
| then the resulting system has an equilibrium at the '''origin''' (i.e., where <math> \mathbf{x}=\mathbf{0}\,</math>, <math>z_1=0</math>, <math>z_2=0</math>, ..., <math>z_{k-1}=0</math>, and <math>z_k=0</math>) that is [[Lyapunov function#Globally asymptotically stable equilibrium|globally asymptotically stable]].
| |
| | |
| ==Integrator Backstepping==
| |
| | |
| Before describing the backstepping procedure for general [[strict-feedback form]] [[dynamical system]]s, it is convenient to discuss the approach for a smaller class of strict-feedback form systems. These systems connect a series of integrators to the input of a
| |
| system with a known feedback-stabilizing control law, and so the stabilizing approach is known as ''integrator backstepping.'' With a small modification, the integrator backstepping approach can be extended to handle all strict-feedback form systems.
| |
| | |
| ===Single-integrator Equilibrium===
| |
| | |
| Consider the [[dynamical system]]
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = u_1
| |
| \end{cases}</math>
| |
| |align="right"|<math> (1)\,</math>
| |
| |-
| |
| |}
| |
| where <math>\mathbf{x} \in \mathbb{R}^n</math> and <math>z_1</math> is a scalar. This system is a [[cascade connection]] of an [[integrator]] with the <math>\mathbf{x}</math> subsystem (i.e., the input <math>u</math> enters an integrator, and the [[integral]] <math>z_1</math> enters the <math>\mathbf{x}</math> subsystem).
| |
| | |
| We assume that <math>f_x(\mathbf{0})=0</math>, and so if <math>u_1=0</math>, <math> \mathbf{x} = \mathbf{0}\,</math> and <math>z_1 = 0</math>, then
| |
| :<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\underbrace{\mathbf{0}}_{\mathbf{x}}) + ( g_x(\underbrace{\mathbf{0}}_{\mathbf{x}}) )(\underbrace{0}_{z_1}) = 0 + ( g_x(\mathbf{0}) )(0) = \mathbf{0} & \quad \text{ (i.e., } \mathbf{x} = \mathbf{0} \text{ is stationary)}\\
| |
| \dot{z}_1 = \overbrace{0}^{u_1} & \quad \text{ (i.e., } z_1 = 0 \text{ is stationary)}
| |
| \end{cases}</math>
| |
| So the [[origin (mathematics)|origin]] <math>(\mathbf{x},z_1) = (\mathbf{0},0)</math> is an equilibrium (i.e., a [[stationary point]]) of the system. If the system ever reaches the origin, it will remain there forever after.
| |
| | |
| ===Single-integrator Backstepping===
| |
| | |
| In this example, backstepping is used to [[Lyapunov stability|stabilize]] the single-integrator system in Equation (1) around its equilibrium at the origin. To be less precise, we wish to design a control law <math>u_1(\mathbf{x},z_1)</math> that ensures that the states <math>(\mathbf{x}, z_1)</math> return to <math>(\mathbf{0},0)</math> after the system is started from some arbitrary initial condition.
| |
| | |
| * First, by assumption, the subsystem
| |
| | |
| ::<math>\dot{\mathbf{x}} = F(\mathbf{x}) \qquad \text{where} \qquad F(\mathbf{x}) \triangleq f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x})</math>
| |
| | |
| :with <math>u_x(\mathbf{0}) = 0</math> has a [[Lyapunov function]] <math>V_x(\mathbf{x}) > 0</math> such that
| |
| | |
| ::<math>\dot{V}_x=\frac{\partial V_x}{\partial \mathbf{x}}(f_x(\mathbf{x})+g_x(\mathbf{x})u_x(\mathbf{x})) \leq - W(\mathbf{x})</math>
| |
| | |
| :where <math>W(\mathbf{x})</math> is a [[positive-definite function]]. That is, we '''assume''' that we have '''already shown''' that this '''existing simpler''' <math>\mathbf{x}</math> '''subsystem''' is '''[[Lyapunov stability|stable (in the sense of Lyapunov)]].''' Roughly speaking, this notion of stability means that:
| |
| ** The function <math>V_x</math> is like a "generalized energy" of the <math>\mathbf{x}</math> subsystem. As the <math>\mathbf{x}</math> states of the system move away from the origin, the energy <math>V_x(\mathbf{x})</math> also grows.
| |
| ** By showing that over time, the energy <math>V_x(\mathbf{x}(t))</math> decays to zero, then the <math>\mathbf{x}</math> states must decay toward <math> \mathbf{x}=\mathbf{0}\,</math>. That is, the origin <math> \mathbf{x}=\mathbf{0}\,</math> will be a '''stable equilibrium''' of the system – the <math>\mathbf{x}</math> states will continuously approach the origin as time increases.
| |
| ** Saying that <math>W(\mathbf{x})</math> is positive definite means that <math>W(\mathbf{x}) > 0</math> everywhere except for <math> \mathbf{x}=\mathbf{0}\,</math>, and <math>W(\mathbf{0})=0</math>.
| |
| ** The statement that <math>\dot{V}_x \leq -W(\mathbf{x})</math> means that <math>\dot{V}_x</math> is bounded away from zero for all points except where <math> \mathbf{x} = \mathbf{0}\,</math>. That is, so long as the system is not at its equilibrium at the origin, its "energy" will be decreasing.
| |
| ** Because the energy is always decaying, then the system must be stable; its trajectories must approach the origin.
| |
| :Our task is to find a control <math>u</math> that makes our cascaded <math>(\mathbf{x},z_1)</math> system also stable. So we must find a ''new'' Lyapunov function '''candidate''' for this new system. That candidate will depend upon the control <math>u</math>, and by choosing the control properly, we can ensure that it is decaying everywhere as well.
| |
| | |
| * Next, by ''adding'' '''and''' ''subtracting'' <math>g_x(\mathbf{x}) u_x(\mathbf{x})</math> (i.e., we don't change the system in any way because we make no ''net'' effect) to the <math>\dot{\mathbf{x}}</math> part of the larger <math>(\mathbf{x},z_1)</math> system, it becomes
| |
| | |
| ::<math>\begin{cases}\dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 + \mathord{\underbrace{\left( g_x(\mathbf{x})u_x(\mathbf{x}) - g_x(\mathbf{x})u_x(\mathbf{x}) \right)}_{0}}\\\dot{z}_1 = u_1\end{cases}</math>
| |
| | |
| :which we can re-group to get
| |
| | |
| ::<math>\begin{cases}\dot{x} = \mathord{\underbrace{\left( f_x(\mathbf{x}) + g_x(\mathbf{x})u_x(\mathbf{x}) \right)}_{F(\mathbf{x})}} + g_x(\mathbf{x}) \underbrace{\left( z_1 - u_x(\mathbf{x}) \right)}_{z_1 \text{ error tracking } u_x}\\\dot{z}_1 = u_1\end{cases}</math>
| |
| | |
| :So our cascaded supersystem encapsulates the known-stable <math>\dot{\mathbf{x}} = F(\mathbf{x})</math> subsystem plus some error perturbation generated by the integrator.
| |
| | |
| * We now can change variables from <math>(\mathbf{x}, z_1)</math> to <math>(\mathbf{x}, e_1)</math> by letting <math>e_1 \triangleq z_1 - u_x(\mathbf{x})</math>. So
| |
| | |
| ::<math>\begin{cases}\dot{\mathbf{x}} = (f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x})) +
| |
| g_x(\mathbf{x}) e_1\\\dot{e}_1 = u_1 - \dot{u}_x\end{cases}</math>
| |
| | |
| : Additionally, we let <math>v_1 \triangleq u_1 - \dot{u}_x</math> so that <math>u_1 = v_1 + \dot{u}_x</math> and
| |
| | |
| ::<math>\begin{cases}\dot{\mathbf{x}} = (f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x}))+g_x(\mathbf{x}) e_1\\\dot{e}_1 = v_1\end{cases}</math>
| |
| | |
| : We seek to stabilize this '''error system''' by feedback through the new control <math>v_1</math>. By stabilizing the system at <math>e_1 = 0</math>, the state <math>z_1</math> will track the desired control <math>u_x</math> which will result in stabilizing the inner <math>\mathbf{x}</math> subsystem.
| |
| | |
| * From our existing Lyapunov function <math>V_x</math>, we define the ''augmented'' Lyapunov function ''candidate''
| |
| | |
| ::<math>V_1(\mathbf{x}, e_1) \triangleq V_x(\mathbf{x}) + \frac{1}{2} e_1^2</math>
| |
| | |
| : So
| |
| | |
| ::<math>\dot{V}_1
| |
| = \dot{V}_x(\mathbf{x}) + \frac{1}{2}\left( 2 e_1 \dot{e}_1 \right)
| |
| = \dot{V}_x(\mathbf{x}) + e_1 \dot{e}_1
| |
| = \dot{V}_x(\mathbf{x}) + e_1 \overbrace{v_1}^{\dot{e}_1}
| |
| = \overbrace{\frac{\partial V_x}{\partial \mathbf{x}} \underbrace{\dot{\mathbf{x}}}_{\text{(i.e., }\frac{\operatorname{d}\mathbf{x}}{\operatorname{d}t}\text{)}}}^{\dot{V}_x\text{ (i.e.,} \frac{\operatorname{d}V_x}{\operatorname{d}t}\text{)}} + e_1 v_1
| |
| = \overbrace{\frac{\partial V_x}{\partial \mathbf{x}} \underbrace{\left( (f_x(\mathbf{x}) + g_x(\mathbf{x})u_x(\mathbf{x})) + g_x(\mathbf{x}) e_1 \right)}_{\dot{\mathbf{x}}}}^{\dot{V}_x} + e_1 v_1</math>
| |
| | |
| : By distributing <math>\partial V_x/\partial \mathbf{x}</math>, we see that
| |
| | |
| ::<math>\dot{V}_1 = \overbrace{\frac{\partial V_x}{\partial \mathbf{x}}(f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x(\mathbf{x}))}^{{} \leq -W(\mathbf{x})} + \frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x}) e_1 + e_1 v_1 \leq -W(\mathbf{x})+ \frac{\partial V_x}{\partial \mathbf{x}} g_x(\mathbf{x}) e_1 + e_1 v_1</math>
| |
| | |
| : To ensure that <math>\dot{V}_1 \leq -W(\mathbf{x}) < 0</math> (i.e., to ensure stability of the supersystem), we '''pick''' the control law
| |
| | |
| ::<math>v_1 = -\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})- k_1 e_1</math>
| |
| | |
| : with <math>k_1 > 0</math>, and so
| |
| | |
| ::<math>\dot{V}_1
| |
| = -W(\mathbf{x}) + \frac{\partial V_x}{\partial \mathbf{x}} g_x(\mathbf{x}) e_1 + e_1\overbrace{\left( -\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})-k_1 e_1 \right)}^{v_1}</math>
| |
| | |
| : After distributing the <math>e_1</math> through,
| |
| | |
| ::<math>\dot{V}_1
| |
| =
| |
| -W(\mathbf{x}) + \mathord{\overbrace{\frac{\partial V_x}{\partial \mathbf{x}} g_x(\mathbf{x}) e_1
| |
| - e_1 \frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})}^{0}} - k_1 e_1^2
| |
| = -W(\mathbf{x})-k_1 e_1^2 \leq -W(\mathbf{x})
| |
| < 0</math>
| |
| | |
| : So our ''candidate'' Lyapunov function <math>V_1</math> '''is''' a true [[Lyapunov function]], and our system is '''stable''' under this control law <math>v_1</math> (which corresponds the control law <math>u_1</math> because <math>v_1 \triangleq u_1 - \dot{u}_x</math>). Using the variables from the original coordinate system, the equivalent Lyapunov function
| |
| ::{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>V_1(\mathbf{x}, z_1) \triangleq V_x(\mathbf{x}) + \frac{1}{2} ( z_1 - u_x(\mathbf{x}) )^2</math>
| |
| |align="right"|<math> (2)\,</math>
| |
| |-
| |
| |}
| |
| : As discussed below, this Lyapunov function will be used again when this procedure is applied iteratively to multiple-integrator problem.
| |
| | |
| * Our choice of control <math>v_1</math> ultimately depends on all of our original state variables. In particular, the actual feedback-stabilizing control law
| |
| ::{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>\underbrace{u_1(\mathbf{x},z_1)=v_1+\dot{u}_x}_{\text{By definition of }v_1}=\overbrace{-\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})-k_1(\underbrace{z_1-u_x(\mathbf{x})}_{e_1})}^{v_1} \, + \, \overbrace{\frac{\partial u_x}{\partial \mathbf{x}}(\underbrace{f_x(\mathbf{x})+g_x(\mathbf{x})z_1}_{\dot{\mathbf{x}} \text{ (i.e., } \frac{\operatorname{d}\mathbf{x}}{\operatorname{d}t} \text{)}})}^{\dot{u}_x \text{ (i.e., } \frac{ \operatorname{d}u_x }{\operatorname{d}t} \text{)}}</math>
| |
| |align="right"|<math> (3)\,</math>
| |
| |-
| |
| |}
| |
| : The states <math>\mathbf{x}</math> and <math>z_1</math> and functions <math>f_x</math> and <math>g_x</math> come from the system. The function <math>u_x</math> comes from our known-stable <math>\dot{\mathbf{x}}=F(\mathbf{x})</math> subsystem. The '''gain''' parameter <math>k_1 > 0</math> affects the convergence rate or our system. Under this control law, our system is [[Lyapunov stability|stable]] at the origin <math>(\mathbf{x},z_1)=(\mathbf{0},0)</math>.
| |
| | |
| : Recall that <math>u_1</math> in Equation (3) drives the input of an integrator that is connected to a subsystem that is feedback-stabilized by the control law <math>u_x</math>. Not surprisingly, the control <math>u_1</math> has a <math>\dot{u}_x</math> term that will be integrated to follow the stabilizing control law <math>\dot{u}_x</math> plus some offset. The other terms provide damping to remove that offset and any other perturbation effects that would be magnified by the integrator.
| |
| | |
| So because this system is feedback stabilized by <math>u_1(\mathbf{x}, z_1)</math> and has Lyapunov function <math>V_1(\mathbf{x},z_1)</math> with <math>\dot{V}_1(\mathbf{x}, z_1) \leq -W(\mathbf{x}) < 0</math>, it can be used as the upper subsystem in another single-integrator cascade system.
| |
| | |
| ===Motivating Example: Two-integrator Backstepping===
| |
| Before discussing the recursive procedure for the general multiple-integrator case, it is instructive to study the recursion present in the two-integrator case. That is, consider the [[dynamical system]]
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = z_2\\
| |
| \dot{z}_2 = u_2
| |
| \end{cases}</math>
| |
| |align="right"|<math> (4)\,</math>
| |
| |-
| |
| |}
| |
| where <math>\mathbf{x} \in \mathbb{R}^n</math> and <math>z_1</math> and <math>z_2</math> are scalars. This system is a cascade connection of the single-integrator system in Equation (1) with another integrator (i.e., the input <math>u_2</math> enters through an integrator, and the output of that integrator enters the system in Equation (1) by its <math>u_1</math> input).
| |
| | |
| By letting
| |
| * <math>\mathbf{y} \triangleq \begin{bmatrix} \mathbf{x} \\ z_1 \end{bmatrix}\,</math>,
| |
| * <math>f_y(\mathbf{y}) \triangleq \begin{bmatrix} f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 \\ 0 \end{bmatrix}\,</math>,
| |
| * <math>g_y(\mathbf{y}) \triangleq \begin{bmatrix} \mathbf{0}\\ 1 \end{bmatrix},\,</math>
| |
| then the two-integrator system in Equation (4) becomes the single-integrator system
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>\begin{cases}
| |
| \dot{\mathbf{y}} = f_y(\mathbf{y}) + g_y(\mathbf{y}) z_2 &\quad \text{( where this } \mathbf{y} \text{ subsystem is stabilized by } z_2 = u_1(\mathbf{x},z_1) \text{ )}\\
| |
| \dot{z}_2 = u_2.
| |
| \end{cases}</math>
| |
| |align="right"|<math> (5)\,</math>
| |
| |-
| |
| |}
| |
| By the single-integrator procedure, the control law <math>u_y(\mathbf{y}) \triangleq u_1(\mathbf{x},z_1)</math> stabilizes the upper <math>z_2</math>-to-<math>\mathbf{y}</math> subsystem using the Lyapunov function <math>V_1(\mathbf{x},z_1)</math>, and so Equation (5) is a new single-integrator system that is structurally equivalent to the single-integrator system in Equation (1). So a stabilizing control <math>u_2</math> can be found using the same single-integrator procedure that was used to find <math>u_1</math>.
| |
| | |
| ===Many-integrator backstepping===
| |
| | |
| In the two-integrator case, the upper single-integrator subsystem was stabilized yielding a new single-integrator system that can be similarly stabilized. This recursive procedure can be extended to handle any finite number of integrators. This claim can be formally proved with [[mathematical induction]]. Here, a stabilized multiple-integrator system is built up from subsystems of already-stabilized multiple-integrator subsystems.
| |
| | |
| * First, consider the [[dynamical system]]
| |
| ::<math>\dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) u_x</math>
| |
| :that has scalar input <math>u_x</math> and output states <math>\mathbf{x} = [x_1, x_2, \ldots, x_n]^{\text{T}} \in \mathbb{R}^n</math>. Assume that
| |
| **<math>f_x(\mathbf{x}) = \mathbf{0}</math> so that the zero-input (i.e., <math>u_x = 0</math>) system is [[stationary point|stationary]] at the origin <math> \mathbf{x} = \mathbf{0}\,</math>. In this case, the origin is called an ''equilibrium'' of the system.
| |
| **The feedback control law <math>u_x(\mathbf{x})</math> stabilizes the system at the equilibrium at the origin.
| |
| **A [[Lyapunov function]] corresponding to this system is described by <math>V_x(\mathbf{x})</math>.
| |
| :That is, if output states <math>\mathbf{x}</math> are fed back to the input <math>u_x</math> by the control law <math>u_x(\mathbf{x})</math>, then the output states (and the Lyapunov function) return to the origin after a single perturbation (e.g., after a nonzero initial condition or a sharp disturbance). This subsystem is '''stabilized''' by feedback control law <math>u_x</math>.
| |
| | |
| * Next, connect an [[integrator]] to input <math>u_x</math> so that the augmented system has input <math>u_1</math> (to the integrator) and output states <math>\mathbf{x}</math>. The resulting augmented dynamical system is
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = u_1
| |
| \end{cases}</math>
| |
| :This "cascade" system matches the form in Equation (1), and so the single-integrator backstepping procedure leads to the stabilizing control law in Equation (3). That is, if we feed back states <math>z_1</math> and <math>\mathbf{x}</math> to input <math>u_1</math> according to the control law
| |
| ::<math>u_1(\mathbf{x},z_1)=-\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})-k_1(z_1-u_x(\mathbf{x})) + \frac{\partial u_x}{\partial \mathbf{x}}(f_x(\mathbf{x})+g_x(\mathbf{x})z_1)</math>
| |
| : with gain <math>k_1 > 0</math>, then the states <math>z_1</math> and <math>\mathbf{x}</math> will return to <math>z_1 = 0</math> and <math> \mathbf{x}=\mathbf{0}\,</math> after a single perturbation. This subsystem is '''stabilized''' by feedback control law <math>u_1</math>, and the corresponding Lyapunov function from Equation (2) is
| |
| ::<math>V_1(\mathbf{x},z_1) = V_x(\mathbf{x}) + \frac{1}{2}( z_1 - u_x(\mathbf{x}) )^2</math>
| |
| :That is, under feedback control law <math>u_1</math>, the Lyapunov function <math>V_1</math> decays to zero as the states return to the origin.
| |
| | |
| * Connect a new integrator to input <math>u_1</math> so that the augmented system has input <math>u_2</math> and output states <math>\mathbf{x}</math>. The resulting augmented dynamical system is
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = z_2\\
| |
| \dot{z}_2 = u_2
| |
| \end{cases}</math>
| |
| :which is equivalent to the ''single''-integrator system
| |
| ::<math>\begin{cases}
| |
| \overbrace{ \begin{bmatrix} \dot{\mathbf{x}}\\ \dot{z}_1 \end{bmatrix} }^{\triangleq \, \dot{\mathbf{x}}_1}
| |
| =
| |
| \overbrace{ \begin{bmatrix} f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 \\ 0 \end{bmatrix} }^{\triangleq \, f_1(\mathbf{x}_1)}
| |
| +
| |
| \overbrace{ \begin{bmatrix} \mathbf{0}\\ 1\end{bmatrix} }^{\triangleq \, g_1(\mathbf{x}_1)} z_2 &\qquad \text{ ( by Lyapunov function } V_1, \text{ subsystem stabilized by } u_1(\textbf{x}_1) \text{ )}\\
| |
| \dot{z}_2 = u_2
| |
| \end{cases}</math>
| |
| :Using these definitions of <math>\mathbf{x}_1</math>, <math>f_1</math>, and <math>g_1</math>, this system can also be expressed as
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}}_1 = f_1(\mathbf{x}_1) + g_1(\mathbf{x}_1) z_2 &\qquad \text{ ( by Lyapunov function } V_1, \text{ subsystem stabilized by } u_1(\textbf{x}_1) \text{ )}\\
| |
| \dot{z}_2 = u_2
| |
| \end{cases}</math>
| |
| :This system matches the single-integrator structure of Equation (1), and so the single-integrator backstepping procedure can be applied again. That is, if we feed back states <math>z_1</math>, <math>z_2</math>, and <math>\mathbf{x}</math> to input <math>u_2</math> according to the control law
| |
| ::<math>u_2(\mathbf{x},z_1,z_2)=-\frac{\partial V_1}{\partial \mathbf{x}_1 } g_1(\mathbf{x}_1)-k_2(z_2-u_1(\mathbf{x}_1)) + \frac{\partial u_1}{\partial \mathbf{x}_1}(f_1(\mathbf{x}_1)+g_1(\mathbf{x}_1)z_2)</math>
| |
| :with gain <math>k_2 > 0</math>, then the states <math>z_1</math>, <math>z_2</math>, and <math>\mathbf{x}</math> will return to <math>z_1 = 0</math>, <math>z_2 = 0</math>, and <math> \mathbf{x}=\mathbf{0}\,</math> after a single perturbation. This subsystem is '''stabilized''' by feedback control law <math>u_2</math>, and the corresponding Lyapunov function is
| |
| ::<math>V_2(\mathbf{x},z_1,z_2) = V_1(\mathbf{x}_1) + \frac{1}{2}( z_2 - u_1(\mathbf{x}_1) )^2</math>
| |
| :That is, under feedback control law <math>u_2</math>, the Lyapunov function <math>V_2</math> decays to zero as the states return to the origin.
| |
| | |
| * Connect an integrator to input <math>u_2</math> so that the augmented system has input <math>u_3</math> and output states <math>\mathbf{x}</math>. The resulting augmented dynamical system is
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = z_2\\
| |
| \dot{z}_2 = z_3\\
| |
| \dot{z}_3 = u_3
| |
| \end{cases}</math>
| |
| :which can be re-grouped as the ''single''-integrator system
| |
| ::<math>\begin{cases}
| |
| \overbrace{ \begin{bmatrix} \dot{\mathbf{x}}\\ \dot{z}_1\\ \dot{z}_2 \end{bmatrix} }^{\triangleq \, \dot{\mathbf{x}}_2}
| |
| =
| |
| \overbrace{ \begin{bmatrix} f_x(\mathbf{x}) + g_x(\mathbf{x}) z_2 \\ z_2 \\ 0\end{bmatrix} }^{\triangleq \, f_2(\mathbf{x}_2)}
| |
| +
| |
| \overbrace{ \begin{bmatrix} \mathbf{0}\\ 0\\ 1\end{bmatrix} }^{\triangleq \, g_2(\mathbf{x}_2)} z_3 &\qquad \text{ ( by Lyapunov function } V_2, \text{ subsystem stabilized by } u_2(\textbf{x}_2) \text{ )}\\
| |
| \dot{z}_3 = u_3
| |
| \end{cases}</math>
| |
| :By the definitions of <math>\mathbf{x}_1</math>, <math>f_1</math>, and <math>g_1</math> from the previous step, this system is also represented by
| |
| ::<math>\begin{cases}
| |
| \overbrace{ \begin{bmatrix} \dot{\mathbf{x}}_1\\ \dot{z}_2 \end{bmatrix} }^{\dot{\mathbf{x}}_2}
| |
| =
| |
| \overbrace{ \begin{bmatrix} f_1(\mathbf{x}_1) + g_1(\mathbf{x}_1) z_2 \\ 0\end{bmatrix} }^{f_2(\mathbf{x}_2)}
| |
| +
| |
| \overbrace{ \begin{bmatrix} \mathbf{0}\\ 1\end{bmatrix} }^{g_2(\mathbf{x}_2)} z_3 &\qquad \text{ ( by Lyapunov function } V_2, \text{ subsystem stabilized by } u_2(\textbf{x}_2) \text{ )}\\
| |
| \dot{z}_3 = u_3
| |
| \end{cases}</math>
| |
| :Further, using these definitions of <math>\mathbf{x}_2</math>, <math>f_2</math>, and <math>g_2</math>, this system can also be expressed as
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}}_2 = f_2(\mathbf{x}_2) + g_2(\mathbf{x}_2) z_3 &\qquad \text{ ( by Lyapunov function } V_2, \text{ subsystem stabilized by } u_2(\textbf{x}_2) \text{ )}\\
| |
| \dot{z}_3 = u_3
| |
| \end{cases}</math>
| |
| :So the re-grouped system has the single-integrator structure of Equation (1), and so the single-integrator backstepping procedure can be applied again. That is, if we feed back states <math>z_1</math>, <math>z_2</math>, <math>z_3</math>, and <math>\mathbf{x}</math> to input <math>u_3</math> according to the control law
| |
| ::<math>u_3(\mathbf{x},z_1,z_2,z_3)=-\frac{\partial V_2}{\partial \mathbf{x}_2 } g_2(\mathbf{x}_2)-k_3(z_3-u_2(\mathbf{x}_2)) + \frac{\partial u_2}{\partial \mathbf{x}_2}(f_2(\mathbf{x}_2)+g_2(\mathbf{x}_2)z_3)</math>
| |
| :with gain <math>k_3 > 0</math>, then the states <math>z_1</math>, <math>z_2</math>, <math>z_3</math>, and <math>\mathbf{x}</math> will return to <math>z_1 = 0</math>, <math>z_2 = 0</math>, <math>z_3 = 0</math>, and <math> \mathbf{x}=\mathbf{0}\,</math> after a single perturbation. This subsystem is '''stabilized''' by feedback control law <math>u_3</math>, and the corresponding Lyapunov function is
| |
| ::<math>V_3(\mathbf{x},z_1,z_2,z_3) = V_2(\mathbf{x}_2) + \frac{1}{2}( z_3 - u_2(\mathbf{x}_2) )^2</math>
| |
| :That is, under feedback control law <math>u_3</math>, the Lyapunov function <math>V_3</math> decays to zero as the states return to the origin.
| |
| | |
| * This process can continue for each integrator added to the system, and hence any system of the form
| |
| ::<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 &\qquad \text{ ( by Lyapunov function } V_x, \text{ subsystem stabilized by } u_x(\textbf{x}) \text{ )}\\
| |
| \dot{z}_1 = z_2\\
| |
| \dot{z}_2 = z_3\\
| |
| \vdots\\
| |
| \dot{z}_i = z_{i+1}\\
| |
| \vdots\\
| |
| \dot{z}_{k-2} = z_{k-1}\\
| |
| \dot{z}_{k-1} = z_k\\
| |
| \dot{z}_k = u
| |
| \end{cases}</math>
| |
| :has the recursive structure
| |
| ::<math>\begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 &\qquad \text{ ( by Lyapunov function } V_x, \text{ subsystem stabilized by } u_x(\textbf{x}) \text{ )}\\
| |
| \dot{z}_1 = z_2
| |
| \end{cases}\\
| |
| \dot{z}_2 = z_3
| |
| \end{cases}\\
| |
| \vdots
| |
| \end{cases}\\
| |
| \dot{z}_i = z_{i+1}
| |
| \end{cases}\\
| |
| \vdots
| |
| \end{cases}\\
| |
| \dot{z}_{k-2} = z_{k-1}
| |
| \end{cases}\\
| |
| \dot{z}_{k-1} = z_k
| |
| \end{cases}\\
| |
| \dot{z}_k = u
| |
| \end{cases}</math>
| |
| :and can be feedback stabilized by finding the feedback-stabilizing control and Lyapunov function for the single-integrator <math>(\mathbf{x},z_1)</math> subsystem (i.e., with input <math>z_2</math> and output <math>\mathbf{x}</math>) and iterating out from that inner subsystem until the ultimate feedback-stabilizing control <math>u</math> is known. At iteration <math>i</math>, the equivalent system is
| |
| ::<math>\begin{cases}
| |
| \overbrace{ \begin{bmatrix} \dot{\mathbf{x}}\\ \dot{z}_1\\ \dot{z}_2 \\ \vdots \\ \dot{z}_{i-2} \\ \dot{z}_{i-1} \end{bmatrix} }^{\triangleq \, \dot{\mathbf{x}}_{i-1}}
| |
| =
| |
| \overbrace{ \begin{bmatrix} f_{i-2}(\mathbf{x}_{i-2}) + g_{i-2}(\mathbf{x}_{i-1}) z_{i-2} \\ 0 \end{bmatrix} }^{\triangleq \, f_{i-1}(\mathbf{x}_{i-1})}
| |
| +
| |
| \overbrace{ \begin{bmatrix} \mathbf{0}\\ 1\end{bmatrix} }^{\triangleq \, g_{i-1}(\mathbf{x}_{i-1})} z_i &\quad \text{ ( by Lyap. func. } V_{i-1}, \text{ subsystem stabilized by } u_{i-1}(\textbf{x}_{i-1}) \text{ )}\\
| |
| \dot{z}_i = u_i
| |
| \end{cases}</math>
| |
| :The corresponding feedback-stabilizing control law is
| |
| ::<math>u_i(\overbrace{\mathbf{x},z_1,z_2,\dots,z_i}^{\triangleq \, \mathbf{x}_i})=-\frac{\partial V_{i-1}}{\partial \mathbf{x}_{i-1} } g_{i-1}(\mathbf{x}_{i-1}) \, - \, k_i(z_i \, - \, u_{i-1}(\mathbf{x}_{i-1})) \, + \, \frac{\partial u_{i-1}}{\partial \mathbf{x}_{i-1}}(f_{i-1}(\mathbf{x}_{i-1}) \, + \, g_{i-1}(\mathbf{x}_{i-1})z_i)</math>
| |
| :with gain <math>k_i > 0</math>. The corresponding Lyapunov function is
| |
| ::<math>V_i(\mathbf{x}_i) = V_{i-1}(\mathbf{x}_{i-1}) + \frac{1}{2}( z_i - u_{i-1}(\mathbf{x}_{i-1}) )^2</math>
| |
| :By this construction, the ultimate control <math>u(\mathbf{x},z_1,z_2,\ldots,z_k) = u_k(\mathbf{x}_k)</math> (i.e., ultimate control is found at final iteration <math>i=k</math>).
| |
| Hence, any system in this special many-integrator strict-feedback form can be feedback stabilized using a straightforward procedure that can even be automated (e.g., as part of an [[adaptive control]] algorithm).
| |
| | |
| ==Generic Backstepping==
| |
| | |
| Systems in the special [[strict-feedback form]] have a recursive structure similar to the many-integrator system structure. Likewise, they are stabilized by stabilizing the smallest cascaded system and then ''backstepping'' to the next cascaded system and repeating the procedure. So it is critical to develop a single-step procedure; that procedure can be recursively applied to cover the many-step case. Fortunately, due to the requirements on the functions in the strict-feedback form, each single-step system can be rendered by feedback to a single-integrator system, and that single-integrator system can be stabilized using methods discussed above.
| |
| | |
| ===Single-step Procedure===
| |
| | |
| Consider the simple [[strict-feedback form|strict-feedback]] [[dynamical system|system]]
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = f_1(\mathbf{x}, z_1) + g_1(\mathbf{x}, z_1) u_1
| |
| \end{cases}</math>
| |
| |align="right"|<math> (6)\,</math>
| |
| |-
| |
| |}
| |
| where
| |
| * <math>\mathbf{x} = [x_1, x_2, \ldots, x_n]^{\text{T}} \in \mathbb{R}^n</math>,
| |
| * <math>z_1</math> and <math>u_1</math> are [[scalar (mathematics)|scalar]]s,
| |
| * For all <math>\mathbf{x}</math> and <math>z_1</math>, <math>g_1(\mathbf{x},z_1) \neq 0</math>.
| |
| Rather than designing feedback-stabilizing control <math>u_1</math> directly, introduce a new control <math>u_{a1}</math> (to be designed ''later'') and use control law
| |
| :<math>u_1( \mathbf{x}, z_1 )
| |
| =
| |
| \frac{ 1 }{ g_1( \mathbf{x}, z_1 ) }
| |
| \left( u_{a1} - f_1(\mathbf{x},z_1) \right)</math>
| |
| which is possible because <math>g_1 \neq 0</math>. So the system in Equation (6) is
| |
| :<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = f_1(\mathbf{x}, z_1) + g_1(\mathbf{x}, z_1) \overbrace{\frac{ 1 }{ g_1( \mathbf{x}, z_1 ) }
| |
| \left( u_{a1} - f_1(\mathbf{x},z_1) \right)}^{u_1(\mathbf{x}, z_1)}
| |
| \end{cases}</math>
| |
| which simplifies to
| |
| :<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1\\
| |
| \dot{z}_1 = u_{a1}
| |
| \end{cases}</math>
| |
| This new <math>u_{a1}</math>-to-<math>\mathbf{x}</math> system matches the ''single-integrator cascade system'' in Equation (1). Assuming that a feedback-stabilizing control law <math>u_x(\mathbf{x})</math> and [[Lyapunov function]] <math>V_x(\mathbf{x})</math> for the upper subsystem is known, the feedback-stabilizing control law from Equation (3) is
| |
| :<math>u_{a1}(\mathbf{x},z_1)=-\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})-k_1(z_1-u_x(\mathbf{x})) + \frac{\partial u_x}{\partial \mathbf{x}}(f_x(\mathbf{x})+g_x(\mathbf{x})z_1)</math>
| |
| with gain <math>k_1 > 0</math>. So the final feedback-stabilizing control law is
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>u_1(\mathbf{x},z_1) = \frac{1}{ g_1(\mathbf{x},z_1) } \left( \overbrace{-\frac{\partial V_x}{\partial \mathbf{x}}g_x(\mathbf{x})-k_1(z_1-u_x(\mathbf{x})) + \frac{\partial u_x}{\partial \mathbf{x}}(f_x(\mathbf{x})+g_x(\mathbf{x})z_1)}^{u_{a1}(\mathbf{x},z_1)} \, - \, f_1(\mathbf{x}, z_1) \right)</math>
| |
| |<math> (7)\,</math>
| |
| |-
| |
| |}
| |
| with gain <math>k_1 > 0</math>. The corresponding Lyapunov function from Equation (2) is
| |
| :{| border="0", width="75%"
| |
| |-
| |
| |align="left"|<math>V_1(\mathbf{x},z_1) = V_x(\mathbf{x}) + \frac{1}{2} ( z_1 - u_x(\mathbf{x}) )^2</math>
| |
| |<math> (8)\,</math>
| |
| |-
| |
| |}
| |
| Because this ''strict-feedback system'' has a feedback-stabilizing control and a corresponding Lyapunov function, it can be cascaded as part of a larger strict-feedback system, and this procedure can be repeated to find the surrounding feedback-stabilizing control.
| |
| | |
| ===Many-step Procedure===
| |
| | |
| As in many-integrator backstepping, the single-step procedure can be completed iteratively to stabilize an entire strict-feedback system. In each step,
| |
| # The smallest "unstabilized" single-step strict-feedback system is isolated.
| |
| # Feedback is used to convert the system into a single-integrator system.
| |
| # The resulting single-integrator system is stabilized.
| |
| # The stabilized system is used as the upper system in the next step.
| |
| That is, any ''strict-feedback system''
| |
| :<math>\begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 &\qquad \text{ ( by Lyapunov function } V_x, \text{ subsystem stabilized by } u_x(\textbf{x}) \text{ )}\\
| |
| \dot{z}_1 = f_1( \mathbf{x}, z_1 ) + g_1( \mathbf{x}, z_1 ) z_2\\
| |
| \dot{z}_2 = f_2( \mathbf{x}, z_1, z_2 ) + g_2( \mathbf{x}, z_1, z_2 ) z_3\\
| |
| \vdots\\
| |
| \dot{z}_i = f_i( \mathbf{x}, z_1, z_2, \ldots, z_i ) + g_i( \mathbf{x}, z_1, z_2, \ldots, z_i ) z_{i+1}\\
| |
| \vdots\\
| |
| \dot{z}_{k-2} = f_{k-2}( \mathbf{x}, z_1, z_2, \ldots z_{k-2} ) + g_{k-2}( \mathbf{x}, z_1, z_2, \ldots, z_{k-2} ) z_{k-1}\\
| |
| \dot{z}_{k-1} = f_{k-1}( \mathbf{x}, z_1, z_2, \ldots z_{k-2}, z_{k-1} ) + g_{k-1}( \mathbf{x}, z_1, z_2, \ldots, z_{k-2}, z_{k-1} ) z_k\\
| |
| \dot{z}_k = f_k( \mathbf{x}, z_1, z_2, \ldots z_{k-1}, z_k ) + g_k( \mathbf{x}, z_1, z_2, \ldots, z_{k-1}, z_k ) u
| |
| \end{cases}</math>
| |
| has the recursive structure
| |
| :<math>\begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \begin{cases}
| |
| \dot{\mathbf{x}} = f_x(\mathbf{x}) + g_x(\mathbf{x}) z_1 &\qquad \text{ ( by Lyapunov function } V_x, \text{ subsystem stabilized by } u_x(\textbf{x}) \text{ )}\\
| |
| \dot{z}_1 = f_1( \mathbf{x}, z_1 ) + g_1( \mathbf{x}, z_1 ) z_2
| |
| \end{cases}\\
| |
| \dot{z}_2 = f_2( \mathbf{x}, z_1, z_2 ) + g_2( \mathbf{x}, z_1, z_2 ) z_3
| |
| \end{cases}\\
| |
| \vdots\\
| |
| \end{cases}\\
| |
| \dot{z}_i = f_i( \mathbf{x}, z_1, z_2, \ldots, z_i ) + g_i( \mathbf{x}, z_1, z_2, \ldots, z_i ) z_{i+1}
| |
| \end{cases}\\
| |
| \vdots
| |
| \end{cases}\\
| |
| \dot{z}_{k-2} = f_{k-2}( \mathbf{x}, z_1, z_2, \ldots z_{k-2} ) + g_{k-2}( \mathbf{x}, z_1, z_2, \ldots, z_{k-2} ) z_{k-1}
| |
| \end{cases}\\
| |
| \dot{z}_{k-1} = f_{k-1}( \mathbf{x}, z_1, z_2, \ldots z_{k-2}, z_{k-1} ) + g_{k-1}( \mathbf{x}, z_1, z_2, \ldots, z_{k-2}, z_{k-1} ) z_k
| |
| \end{cases}\\
| |
| \dot{z}_k = f_k( \mathbf{x}, z_1, z_2, \ldots z_{k-1}, z_k ) + g_k( \mathbf{x}, z_1, z_2, \ldots, z_{k-1}, z_k ) u
| |
| \end{cases}</math>
| |
| and can be feedback stabilized by finding the feedback-stabilizing control and Lyapunov function for the single-integrator <math>(\mathbf{x},z_1)</math> subsystem (i.e., with input <math>z_2</math> and output <math>\mathbf{x}</math>) and iterating out from that inner subsystem until the ultimate feedback-stabilizing control <math>u</math> is known. At iteration <math>i</math>, the equivalent system is
| |
| :<math>\begin{cases}
| |
| \overbrace{ \begin{bmatrix} \dot{\mathbf{x}}\\ \dot{z}_1\\ \dot{z}_2 \\ \vdots \\ \dot{z}_{i-2} \\ \dot{z}_{i-1} \end{bmatrix} }^{\triangleq \, \dot{\mathbf{x}}_{i-1}}
| |
| =
| |
| \overbrace{ \begin{bmatrix} f_{i-2}(\mathbf{x}_{i-2}) + g_{i-2}(\mathbf{x}_{i-2}) z_{i-2} \\ f_{i-1}(\mathbf{x}_i) \end{bmatrix} }^{\triangleq \, f_{i-1}(\mathbf{x}_{i-1})}
| |
| +
| |
| \overbrace{ \begin{bmatrix} \mathbf{0}\\ g_{i-1}(\mathbf{x}_i)\end{bmatrix} }^{\triangleq \, g_{i-1}(\mathbf{x}_{i-1})} z_i &\quad \text{ ( by Lyap. func. } V_{i-1}, \text{ subsystem stabilized by } u_{i-1}(\textbf{x}_{i-1}) \text{ )}\\
| |
| \dot{z}_i = f_i(\mathbf{x}_i) + g_i(\mathbf{x}_i) u_i
| |
| \end{cases}</math>
| |
| By Equation (7), the corresponding feedback-stabilizing control law is
| |
| :<math>u_i(\overbrace{\mathbf{x},z_1,z_2,\dots,z_i}^{\triangleq \, \mathbf{x}_i})
| |
| =
| |
| \frac{1}{g_i(\mathbf{x}_i)}
| |
| \left( \overbrace{-\frac{\partial V_{i-1}}{\partial \mathbf{x}_{i-1} }
| |
| g_{i-1}(\mathbf{x}_{i-1})
| |
| \, - \,
| |
| k_i\left( z_i \, - \, u_{i-1}(\mathbf{x}_{i-1}) \right)
| |
| \, + \,
| |
| \frac{\partial u_{i-1}}{\partial \mathbf{x}_{i-1}}(f_{i-1}(\mathbf{x}_{i-1})
| |
| \, + \,
| |
| g_{i-1}(\mathbf{x}_{i-1})z_i) }^{\text{Single-integrator stabilizing control } u_{a\;\!i}(\mathbf{x}_i)}
| |
| \, - \,
| |
| f_i( \mathbf{x}_{i-1} )
| |
| \right)</math>
| |
| with gain <math>k_i > 0</math>. By Equation (8), the corresponding Lyapunov function is
| |
| :<math>V_i(\mathbf{x}_i) = V_{i-1}(\mathbf{x}_{i-1}) + \frac{1}{2} ( z_i - u_{i-1}(\mathbf{x}_{i-1}) )^2</math>
| |
| By this construction, the ultimate control <math>u(\mathbf{x},z_1,z_2,\ldots,z_k) = u_k(\mathbf{x}_k)</math> (i.e., ultimate control is found at final iteration <math>i=k</math>).
| |
| Hence, any strict-feedback system can be feedback stabilized using a straightforward procedure that can even be automated (e.g., as part of an [[adaptive control]] algorithm).
| |
| | |
| ==See also==
| |
| * [[Nonlinear control]]
| |
| * [[Strict-feedback form]]
| |
| * [[Robust control]]
| |
| * [[Adaptive control]]
| |
| | |
| ==References==
| |
| {{Reflist}}
| |
| | |
| [[Category:Control theory]]
| |
| [[Category:Nonlinear control]]
| |