Basics WNCEPTS OF SYSTEM THEORY System: the part of the fea) world thet Is of tajerest for @ particular study or application outputs ore system =}-——> & inputs: independert variables * controllable + uncontfollable (disturbances) outputs: dependent variables + controlled + measured Model : mathematical description of the ‘lationship between system inputs and outputs, gor F (uc ») mapping between functions The mapping F can be: © exphest Example: model of 2 resistor fesistamice. a t tyre ve Re - Ohm's Law - mac ‘current (input) voltage (ovtput) TE alti: cast), then w(t)= Reoslt).implicit (eg. 0 differential equation) Example model of o capacitor ae ; dv) -—| be alt)= C = ° dt X Rm fa f = capacitance. IF c(b=coslt), then we have fo solve. dult) 2 A easter) d © Ti begvating both sroes: + t, if dv(t) = J ws(t)de oO fo} a . => v(b)- v0) 22 [ sine) | => vib): v0) + a sin(t) velue of Lhe voltage at time t=0 REMARKS © A model is vor the system Example! incandesen k lanp modelled with Ohm's Jaw 7 Ohm's Law 1 Hot Resistance Area Current (A) Cold Resistance Area Volt)=> Cope with uncertainty in trinsic in a mode | ° paremetric uncertainties e Unmodelled behaviors © The mode) of 2 system is not Unique, W Different models for different: apphcations and levels of acceracy or detail + Using First praciples (eg. laws of physics) + Using measured data => system identification ~ measure wl+) and he carvesponding gO ~ look for a mapping F minimizing the norm of the approximation error I 4c)- Fue) | Given the pairs (44,42), (Ua,42, (49.43), oF (Uy. Yu), find wefficients (m,4) thet minimize Ni 2, ya (He 4-4) \ sum of squared errorsWhat do we do with models? « Simulation - As-Ts arelysis ~- What-Tf analysis « Prediction * Gntvroller design CLASSIFICATION OF SYSTEMS (AND MODELS) e Stanc ve DyYvamicaAL SYSTEMS A system ‘9 static if Lhe output at time t depends only on the input at time + Exampre: the resistor Ohm's Law: ulb= R(t) A system (5 dynamice| iF Ehe ovtput at time t depends onthe history of the input” before time t. Exampre: the capacitor t alt= v)+4 [4@se Co So, | the integral takes (ate accovat al] the valves of the input from time o to time tProsiem: Assume thet the system input uw 15 known From time te to Cime the system input toknown Ts this knowledge sufficiers to determine uniquely the systen output from time ta ? + the enswer is YES for slabic sy stens + the answer 15 NO for dynameal systens Exampre: the capacitor 4 (2 alby= w(0.)+— f4@se oO t, order to debeVvmine vaiguely ult) we need to know (here bozo): fe) (c) fogiocie ae © alo) initial condition at C=O => The voltage 1” 15 8 state veviable of the system (and coincides with an output vaniable isthis case).ExamPLg: 3 car Problem: Determine the space needed bo slop the car if one applies 3 corstant braking force 7 balers _ —— displacenert (input) (otme) == = —_—_— position displacenent ti ; fo AO at time O S(t) attimet, (when one starts applying the bveking force ) Second principle of dynamics: MSQ®)=-F where: + M>0:mass of the car ° foo braking force, constant. + E(t): car acceleration [ZMderivative of stt)]Tategrsting both sides: t t [See ied ey ie > St)=S@) aa C«) 0 M6 tid where: « S(t): Car velocity [4* derivative of s(t)] Totegratiag both sides again: 43 t J s@sc = |[x0-fe]0 8 ° 3lh |e => 5(t)= 50) + $@)t - a cs => | sl)= One (a) x Te determine the displacenest, s(t) uniquely, we need to know the initial velouty of the car => The cor velocity 's a state vaviable of the system (not coinciding with an output variable is this case), Remark: To Solve the taitia) problem, evaluate (a) atts! son , obtained equating («) te 0.DEFINITION -~ The state of a system is Lhe set: of variables whose value at time fo is fequived sa order to determine uniquely the output 4© forall tte, giver the input ult) forall trte. To general, the state of a systen 19 denoted by 2¢, EXAMPLES . Capacitor (Gar c= [v] ne [ 5] y= {vI 4° [s] ¢ GNiTiNUoUs-TiMe vs Discrete-TIME Systems Gonkinvous- time system : bime t belongs toa cantinvous Set eg, the set of nonnegative reals R= jteR: t2o] Discrete -time system: time t belongs toa discrete seb eg, the set of natural numbers IN = Jo4,2,...}@ GNTiNUoUs-stATE vs Discrete- STATE SYSTEMS Gontinuous-state system: the system state takes values in a continuous set Discrete -state system: the system state takes values in a discrete (Countable) set © TiMe- DRIVEN vs Event- Driven SYSTEMS Th event-driven systems, the state changes only upon the (Eypically asynchronous) Occurrence of events In time~ driven sysbens, the state changes synchronously with the system clock. Examece (TIME-DRIVEN System) : Suspension systen => Suspension deflection after @& moving over a dipModel of the suspension: mass- spring - damper system f tx Ui suspension deflection mo+coskuaf Notice that + causes 8 change, ile) = Gey 8% An advancement in time ws. Moreover, notice that, Co compute derivatives, or(t) vnust be a continuous function. Remark: Continvous-time, continuous- state, Lime - driven systens are typically madelled by differential equations. (Quite) geneval form: so(ty= g(x, ult), ) ‘Time iS an independents stole input variable !Exanece (EvENT-DRIVEN S¥sTEM )' queueing system v Uh el queve, Server State 90 = number of customers in the system chenges only when an artival or } departure occurs Atcivals, departures —> everts (occur asynchronously with the clock ) x(t) piecewise- constant (7 sample pathTime instants of interest, are those, where che plot 15 discontinuous =y[ndeed hose Lime instonts where the state ‘sg non-differertiabl, and Lhevefofe cannot be modelled by a differentia] equation H => We will see Ehat for these systens time 5 A dependent variable. (in Lhe meaning of systen theary ) What 1's an Event ? Au event 15 an instantaneaus occurrence. ° specific actions (e.g. , avrival ofa customer, termination of a service, hitting e button on akeylooard,...)° spoataneous occurences (2.g., failure of 3 machine , start of caining , --- ) « fulfilment of lagica| conditions (e.g, overcoming ofa threshold, etc. ) Derinimon: A discrete- event system Is an event-driven, dynamics) system with discrete state