© What is power system security?: Power system security may be looked Probability of the system's operating Point remaining within acceptable ranec (contingencies) and its environment TE Dy Liacco, "The Adaptive Reliability Control System", IEEE Trans. PAS-86, No.5, May 1967, PP. 517-531: Dy Liaceo fi tem may be identified to be abilities of changes in the system st pointed out in 1967 th, operating in a number of statesPreventive State \ —— eee | ican, Emergency State Restorative State Figure I: Thame Scate model ofthe Poner System OperationPreventive or Normal State: © The preventive state is actually the normal state. The term ‘preventive’ was used to the ‘Security’ aspect of the normal operation © Normal operating condition usually means that all the apparatus are running within the prescribed limits, and all the system variables are within acceptable ranges. @ The system should also continue to operate ‘normally’ even in the case of credible continger cies. The operator should ‘foresee’ such contingencies (disturbances) and take preventi control actions (as economically as possible) such that the system integrity and qua’ po’ pply is maintained. Emergency State: @ The powe: em enters an emergency state when some of the components operating limNormal State ~ = Yeqanan (raion) gee 2 Five Seas medal of Pomer Spas OperasionFive-State Model Power system operation can be described differential, and two sets of algebraic. Of the two sets of algebraic equations, one comprises of equality constraints(E), which the balance between generation and load demand. The other set consists of inequality constraints (I) which ensures that the various con ponents in the system and the states (e.g. voltages and currents) remain within safe ¢ acceptable limits. by three sets of generic equations viz. one set ¢ If the generation falls below certain threshold, load increases beyond some limit, or a pote tially dangerous disturbance becomes imminent, the system is said to enter the alert stat} Though the equality (E) and inequality (1) constraints are still maintained, preventive col trols should be brought into action to steer the system out of the alert state.If preventive control fails, or the disturbance is reverse, the system may enter into an eme'| gency state, though the demand is still met by the generation, one or more component ¢ state violate the prescribed operation limits. Emergency control actions should immediatel] be brought into action to bring the system back to the normal state. If the emergency control actions also fail, the system may enter extremis state which characterized by disintegration of the entire system into smaller islands, or a comple! system blackout. It may take anywhere between few seconds to few minutes for a system to enter an extrem. state from a normal state. It may take several hours or even days { The restoration process however is very slow. bring the system back to normal.© The major components of security assessment are: » System monitoring ° Contingency analysis © Preventive and corrective actions © System monitoring: © The prerequisite for security assessment of a power system is the knowledge of the syste states. Monitoring the system is therefore the first step. 2 Measurement devices dispersed throughout the system help in getting a picture of the curre! operating state. The measurements can be in the form of power injections, power flow voltage, current, status of circuit breakers, switches, transformer taps, generator output et which are telemetered to the control centre. Usually a state estimator is used in the control centre to process these telemetered data a compute the best estimates of the system states. Remote control of the circuit breakers, disconnector switches, transformer taps etc. is genera possible. The entire measurement and control system is commonly known as supervise control and data acquisition (SCADA) systeme Contingency analysis: * Once the current operating state is known, the next task is the contingency analysis. Resull of contingency analy: allow the system to be operated defensively. * Major components of contingency analysis are © Contingency definition © Contingency selection ® Contingency evaluation * Contingency definition involves preparing a list of probable contingencies. » Contingency selection process consists of selecting the set of most probable contingencie ' preferred; they need to be evaluated in terms of potential risk to the system. Usually, fa power flow solution techniques such as DC power flow are used to quickly evaluate the ris associated with each contingency. ° Finally, the selected contingencies are ranked in order of their security, till no violation operating limits is observed.Security Assessment » Preventive and corrective actions: © Preventive and corrective actions are needed bring it to 2 secure operating state. * Corrective actions such as switching of VAR compensating devices, changing transformer tay and phase shifters etc. are mainly automatic in nature, and involve short duration. = Preventive actions such as generation rescheduling involve longer time scales. Securit constrained optimal power flow is an example of rescheduling the generations in the syste! in order to ensure a secure operation. Online Security Assessment: ® In earlier days, security assessment in a power system was mainly offline in nature. Predefine set of rules or nomographs were used to assist the operators in the decision-making proce to maintain a secure operation of a system or t ® How , due to the highly interconnected nature of modern power systems, and deregula’ y market scenarios, operating conditions and even the topology of a power syste ently. Off-line techniques for security assessment are therefore no- long’ eliable in modern power systems. @ On-line security assessment techniques use near-real-time measurements from different cations in a power system, and continuously update the security assessment of the systSecurity Assessment. fie Power system security analysis,’ N. Balu et.al, "Ont "Proc. of the IEEE, Vol 80, No. 2, Fet| 1992, pp. 262-280N. Balu et.al., "On-line power system security analysis,"Proc. of the IEEE, Vol 80, No. 2, Feb 1992, pp. 262-280. Power System Security Based on Parametric] @ Line security: This is also known as rea| power security. This ensures the real po} flow limit within the maximum value @ Voltage security: This is also known a reactive power security. This ensures th magnitude of voltage with the limite.Flow.or Reali Rower: Flaw! Based Ranking Example - 1 Determine the ranking of con For «°" contingency: gencies using line flow or real power flow ba i pe approach that utilizes DCLF for the sysij pp —>y > 2 Tae shown in Fig. 4 (PD = Soon | Bese @ 1 0.02 + 70.04 where, N; = Total no. of lines, Pi” = Power Gi flow in line-| following a contingency - i, Pn Max. limit on real power flowin line - I, n = in- |_oo+s003 Stack Bus 3 ger value, 2n = order of performance: index, a kbele .) — relative weighing factor of line - 1, which is decided by the operator by examining the oper- sting conditions. In most of the literature, the value of n is taken as 1, and the order becomes— — i ncasiiiniitimeele rim 2 © S Gl ge Security Based Ranking: z AV 1) => ne SR (SE) ow Where, w, — relative weight of bus - 7, N = Total number of buses in the system avy= = AV = yi _ yt iim m yma. fv > vim pai Vere V If it is not violating the limit, then the limit closer he FV CV” FO = fO +a4-AP, A” = ay -APi + Yay +2 (4) (2) Following the outage of it* generator, ont deficit in generation is met by remaining generators (in the ratio of their ratings) Substituting the above expression of AP; in aDC Sensitivity Factors , an | For j'* generator j + i, a distribution factor can 2 Generator Shift Factor: ee eancite Assumption: (1) Following the outage of it” generator, slack bus generator picks up the load Afi BP, (3) a where, AP, = —P{2). From equation (2): Change inline power flow can be represented as No +a; na AR” = ay AP, + Do ay -AP; (5) 4) jal (2) Following the outage of i** generator, a#t deficit in generation is met by remaining generators (in the ratio of their ratings) As) = au-AP, > A = 5 Substituting the above expression of AP, in (5)Ng AR = au- AP: —Yoaz-7%,-AP: (6) =i ou No FO =F +a AP Yo ai; jal at © Line Outage Distribution Factors: dix = ‘a @) FO =f 4 ana fOtdn- AO (9) Calculation of DC Sen: Generation Outage Distribution Factors: Equation (3) can be represented as: Aft, Oh AP; ~ OP; Let us assume that the line / is connected be- tween buses n and m, then We can write fr as: ay © fi=+(6n-5n) Eo 6, = Xp Pi ++ Sy = Xr Pp too +XniPi t+ + XmiPi +Substituting 0, and 6,, in above equation of a;: Of _ 1 (= = oe) — (Xni = Xmi) “OP, ~ i \OP, OP; a (10) Line Outage Distribution Factors: & (Xin — Xjn — Xim + Xjm] t —[Xnn + Xmm — 2Xnm] (11) dijnm = dik = 0 i ieand dy, in above equation of aj; 5m _ (Xni = Xmi) ” OP; ) Z (10) led by adding system, one at end of the line to be dropped. The eft in the system and the effects are njections. he system shown in Fig. 5 e hn and m ysis Using Sensitivi aus Nes To REMAINDER OF NETWORKAnalysis Using Sensitivity Factors fe If APa = Pam, where Prm is equal t power flowing over the line, and Al —Pam, we will still have no current ng through the circuit t they are closec | closed , butce) \ Faby)Ain =XandPa+XamAPm (43) siberes pet A= XERAPiE Xen SPin nm =, an Om) define Then, Sn:dm:Pnm — to exist before the outage, where Pam is the flow on line k from bus n to bus m Abn = (Xnn — Xnm) SPn Gn, Adin, APnm to be the incremental changes Aém = (Xinm — Xmn) AP resulting from the outage 5nedm+Pam — to exist after the outage and bn =n + Abn Pam = &Pn = —APrm (12) i254 a5 Giving, alos P. UE 1 1 ym = — (bn — bm) = — (bn — Sn) + — (45, — Adm) oh Tk Tk : 1 => Pam = Pam + — (Xan + Ximm — 2Xnm) AP, = ji2 Using the above relationships, the sensitivit _|p tor Si.nm can be represented as nm) | Al | ae (16) = (Xin = Xim fine a sensitivity factor s as the ratio of the tk — (Xnn + Xmm — 2X ange in phase angle 5, anywhere in the system the onginal power Pnm flowing over a line before it was dropped. That is, Then the expression dy, can be rey ; Ah = (Aéd; — Ad As, aa a = (17) Pram 1 (M5 AS Pe aml ase angle Ad; can be repre. a NY Pram a AI XimAP, (18) i‘neither / nor j is a reference bu: Ac, _ | [mek One He mL te — (Xan + Xmm ~ 2Xnm) o The fact that the « and d factors are linear models of the power system allows us to use to extend them. ul extension is to use the a to model the power system in uta that is, to generate 5 that model the system's sensitivity ° nsitivity enerator bus s, then dij.nm = dy. can be given as: jm) ke z (X c B @ This is calculated by first assuming the change in generation on bus «42, | direct effect on line / and an in t through its influence on the power f on line k, which, in turn, influences I when line k is out. Then Af = iAP; + du MS However, we know that= aj AP; + da: AP; = (ar; + deans) AP; ( n refer to aj; + diay: as the "compensated generation shift sensitivity" OePerformance Index for Contingency Ranking Line Flow or Real Power Fi = low Based Ranking: For i contingency: a oe eno =e (z ) a) on where, Ny = Total no. of lines, P\ —~ Power flow in line-t following a contingency - i, Py"*= = Max. limit on real power flowin line - /, n — in- teger value, 2n = order of performance index, 1 = relative weighing factor of line - !, which is decided by the operator by examining the oper- ating conditions. In most of the literature, the value of nm is taken as 1, and the order becomes Example - 1 Determine the ranking of cor Bencies using line flow or real power flow b approach that utilizes DCLF for the sy: shown in Fig. 4 sxe. Mag 20) [Ms [= 108L [ (Xin = Xim) te — (Xin = Xjm) =e] _ 2 (Xin = Xin —Xim + Xjm atl te— (Xan + Xmm — 2Xnm) 24 =) F that the a and d factors are linear ls of the power system allows us to use Superposition to extend them. Dne very useful extension is to use the a vd d factors to model the power system in post-outage state; that is, to generate \ors that model the system's sensitivity a branch has been lost se one desired to have the sensitivity between line | and generator bus / ine i was opened @ This is calculated by first assuming that change in generation on bus i. AP, hy direct effect on line | and an indirect ef through its influence on the power flov on line k, which, in turn, influences 11 when line & is out. Then Afi = ai AP, + de ASe However, we know that Af = ans AP:her / nor j is a reference bus, then dij.nm = diz can be given as: —Xjm)e] — (Xin — Xjn—Xim + Xjm) 1 [ (Xin = Nim) te = (Xin tk — (Xan + Xmm — 2Xam) tk — (Xan + Ximm — e The fact that the a and d factors are linear models of the power system allows us to use superposition to extend them © This is calculated by first assuming that change in generation on bus i. AP. h direct effect on line / and an indirect © One very useful extension is to use the a through its influence on the power flo and d factors to model the power system in on line k, which, in turn, influences |i its post-outage state; that is, to generate when line k is out. Then factors that model the system's sensitivity Afi = ay AP, + de dS after a branch has been lost ; a Suppose one desired to have the sensitivity However, we know that ctor between line / and generator bus 7 hen line k was opened Afi = am AP,Afi = AP, + deter MP, = (ans + detie) AP, Iiyays a8 the “compensated generation shift sensitivity vity Factors: ator Outage Voltage Distribution (0) Factors: \ J) P;, + jQ. AP, \© Generator Outage Line (MW) Distribu- tion Factors: ) cline — <— Eee Puy agar 0 Pome Bow repeat af. AF! fe A Gt aE ENO) pitrew) — pl 4. kA, + a, SQ Where, p Pie = Pej Pr 5 Qu -Q Coa and PT and Qf are average rea power flow through line /Power Sytem Seay & Contng Outage Analysis Using Sensitivity Factors © Line Outage Voltage Distribution Fac- tors: ' @ = ty = BO Pp (27) ofp a ome yee = V9 42 PT O42 QM (28) e Line Outage Line Flow Distribution Fac i pee “ee ~ pre > pliner) _ pTlold) 4 7 Qro 3” + L OQ, ail AN outagean hia =Problems with PI Based Contingency Ranking: e@ Masking Effect: With the use of lower Misranking Effect: This happens order performance index (PI), contingency to the use of approximate methods cases which gives power flow just below the by using approximate methods some t line rating be found to be greater than con- contingencies are ranked wrongly tingency cases which gives power in one line this can be overcome by minim: just above the line rating and which results [pry — (Pry |. this hapt into wrong contingency selection and called masking effect. This can be overcome by sing higher order Pl by optimally readjusting the weighing ‘tors. Assignment approachvation © The main goal of the system operator is to maintain the system in the normal ——\ as the operating conditions vary during the daily operation secure state | » Accomplishing this goal requires continuous monitoring of the system conditions, ation of the operating state and determination of the necessary preventive actior m state is found to be insecure quence of actions is referred to as the security analysis of the system The first step of security analysis is to monitor the current state of the system and th involves acquisition of measurements from all parts of the system and then processing t to determine the system states. measurements may be both of analog and digital (on/off status of devices)type tations are equipped with devices called remote terminal units (RTU) which colle es of measurements from the field and are responsible for transmitting them iE called intelligent electronic devices (IED) RTUs.main goal of the system operator is to maintain the system in the normal secure stat ss the operating conditions vary during the daily operation * Accomplishing this goal requires continuous monitoring of the system conditions, identifi ation of the operating state and determination of the necessary preventive actions in cas the system state is found to be insecure. @ This sequence of actions is referred to as the security analysis of the system @ The first step of security analysis is to monitor the current state of the system and th involves acquisition of measurements from all parts of the system and then proce n order to determine the system states. @ The measurements may be both of analog and digital (on/off status of device ° ions are equipped with devices called remote terminal units (RTU) which coll arious types of measurements from the field and are responsible for transmitting them t ntrol center ° Mo ntly, the so-called intelligent electronic devices (IED) are replacing or comp isting RTUs@ It is possible to have a mixture of these devices: connected to a local area network LAN) alang with a SCADA front end com~ puter, which supports the communication of the callected measurements to the host computer at the control center. @ The SCADA host computer at the control nter receives measurements from all the monitored substations’ SCADA systems via sne of many possible types of communica- cs received at the control cen- ine power flaws, bus voltage nitudes, generator out- r and switch status information, transformer tap positi¢ and switchable capacitor bank values.@ It is possible to have a mixture of these devices connected to a local area network (LAN) along with a SCADA front end com- puter, which supports the communication of the collected measurements to the host computer at the control center. @ The SCADA host computer at the control center receives measurements from all the monitored substations’ SCADA systems via one of many possible types of communica- tion links such as fiber optics, satellite, mi- crowave, etc. @ Measurements received at the control cen- ter will include line power flows, bus voltage and line current magnitudes, generator out- puts, loads, circuit breaker and switch status information, transformer tap positio| and switchable capacitor bank values igre 1 EMS cr SCADA Sytem Conbgurton© These raw data and measurements are pro- essed by the state estimator in order to fil- the measurement noise and detect gross errors State estimator solution will provide an op- timal estimate of the system state based on the available measurements and on the as- sumed system model, This will then be passed on to all the management system (EMS) ‘appli- in. functi n such as the contingency automatic generation control, load g and optimal pow flow, etc. @ The same information will also be ava via LAN connection to the corporate o where other planning and analysis func can be executed off-line Initially, power systems were mon by supervisory control systems control systems which and control the status of circuit the substations. @ Generator outputs and the quency v Automatic g Economic Diste Estimation Introduction Weighted Least Squares State Estimation4 © The foregoing concerns were first recog- nized and subsequently addressed by Fred Schweppe, who proposed the idea of state estimation in power systems. © Schweppe F.C. and Wildes J., "Power Sys- tem Static-State Estimation, Part |:Exact Model", IEEE Transactions on Power Ap- some and Systems, Vol.PAS-89, January 1970, pp. 120-125. © Schweppe F.C. and Rom D.B.,"Power Sys- tem Static-State Estimation, Part Il: Ap- proximate Model", IEEE Transactions on Power Apparatus and Systems, Vol.PAS-89, January 1970, pp-125-130. © Schweppe F.C., "Power System Stat State Estimation, Part III: Implementati , IEEE Transactions on Power Appar and Systems, Vol.PAS-89, January 19 pp. 130-13 @ Introduction of the state estimation tion broadened the capabilities of SCADA system computers, leading t establishment of the Energy Manage Systems (EMS), which would no equipped with, among other applica functions, an on-line State Estimator (Introduction to State Estimation @ In order to identify the current operating state of the system, state estimators facili- ate accurate and efficient monitoring of op- erational constraints on quantities such as the transmission line loadings or bus volt- age magnitudes. © They provide a reliable real-time data base of the system, including the existing state based on which, security assessment func- tions can be reliably deployed in order to an- contingencies, and to d corrective tions. @ The state estimators typically include the functions: jowin Topology processor: Gathers status about the circuit breakers and switches, a configures the one-line diagram of the tem @ Observability analysis: Determines if state estimation solution for the entire 5] tem can be obtained using the availab! of measurements. Identifies the unobs¢ able branches, and the observable island the system if any exist. State estimation solution: Det the optimal estimate for the system which is composed of complex bus vol in the entire power m, network model and the gathe ments from the system mi ed meastAlso provides the best estimates for all the line flows, loads, transformer taps, and generator outputs. e Bad data processing: Detects the exis- tence of gross errors in the measurement set. Identifies and eliminates bad measure- ments provided that there is enough redun- dancy in the measurement configuration. e Parameter and structural error process- ing: Estimates various network parameters, such as transmission line model parame- s, tap changing transformer parameters, shunt Capacitor or reactor parameters. De- ‘ts structural errors in the network config: n and identifies the erroneous breaker provided that there is enough mea- edundancy @ Thus, power system state estimator con: tutes the core of the on-line security ana sis function. It acts like a filter between Taw measurements received from the tem and all the application functions t} require the most reliable data b. current state of the system. for Figure 2 describes the data and funct interfaces between the various applica functions involved in the on-line st curity assessment proceduree@ Power system is assumed to operate in the steady state under balanced conditions. This implies that all bus loads and branch power flows will be three phase and bal- anced, all transmission lines are fully trans- posed, and all other series or shunt devices are symmetrical in the three phases. © These assumptions allow the use of single phase positive sequence equivalent circuit for modeling the entire power system he solution that will be obtained by us- ing such a network model, will also be the positive sequence component of the system state during balanced steady state opera- tion @ Asin the case of the power flow, all net. data as well as the network variables expressed in the per unit system Example - 1 1 2. acl@ Form the bus admittance matrix, Y for ider the 4-bus power system whose one-line the entire system diagram is given in Figure 3. Network data and ee ea iirsca te nee iniges power inject th dy state bus voltages are listed below. ee susceptance of the shunt capacitor at bus 3 (Bue) Voltage Mag epeeeeraie is given as 0.5 per unit. os = e Write the nodal equations for the 2-port T0000 | 0.¢ z-model of the transformer connected 0.9620 «| een bus-2 and -4. 0.9597 0.9742 To Ree (eae Total Line | Tap Bus pu | pu Charging Susceptance ae 2 Daze] Ole eee OTR 3 0.02 0.06 0.25 - 0.00GW. Stagg, A. H. Ei-Abiad, Computer Methods in Power System Analysis, MEDTECH 2019. G.LKusic, Computer Aided Power Systems Analysis, Prentice Hall, 1986. A.J.Wood and B.F.Wollenberg, Power Generation, Operation and Control, John Wiley, 1984. O.1.Elgerd, Electric Energy Systems Theory, McGraw Hill, 1971. A. Abur and A. G. Exposito, Power System State Estimation: Theory and Implementatior Marcel Dekker, 2004.Expanding the non-linear function g(.r) into its Taylor series around the state vector .r* yields: o(«) =9 (2*) +E (x*) (e—2*) +---=0 Neglecting the higher order terms leads to an iterative solution scheme known as the Gauss- Newton method as shown below: where, k is the iteration index, x* is the solution rat iteration k, and© The objective of state estimation is to de termine the most likely state of the system based on the quantities that are measured. © One way to accomplish this is by maximum likelinood estimation (MLE), a method widely used in statistics. @ The measurement errors are assumed to a known probability distribution with unknown parameters. The joint probability density function for all the measurements can then be written in terms of these un- known parameters. This function is referred to as the likelihood function and will attain its peak value when the unknown parame- ters are chosen to be closest to their actual hay alues © Hence, an optimization problem can be up in order to maximize the likelihood fu tion as a function of these unknown par eters. @ The solution will give the maximum lil hood estimates for the parameters of in est. @ The measurement errors are common sumed to have a Gaussian (Normal) di bution and the parameters for such a tribution are its mean (j:) and its varia (a7). The problem of maximum likelih estimation is then solved for these wo rameters- on the parameters j: and we probability density function ng shape can be standardi: responding probability ing change of variables n (d.f.) will be discuss priefly before describing the maximum like- hood estimation method = Gaussian (Normal) probability density function: h Jd The Normal probability density function for a Which yields andom variable = is defined as 1 E(u) E(=)-p 0 {mame Gaussian p.d-f. The join’ P dfcan be expressed as the product of individ) \S W each measurement is assumed to DS \ dent of the rest , whereEssentially it is a measure of the probability of observing the particular set of measurements in the vector =. @ The objective of maximum likelihood esti- mation is to maximize this likelihood func- tion by varying the assumed parameters of the density function, namely its mean jr and indard deviation o. mining the optimum parameter val- Jes, the function is commonly replaced by its logarithm, in order to simplify the opti- mization procedure. The modified function Is called the Log-Likelihood Function £ and is given by: £ = log fm(=) = >» log f (=i) i= MLE will maximize the likelihood (or | likelihood)function for a given set of obse} tions =). ++-+>2m- Hence, it can be obtai by solving the following problem maximize log fm (=) OR minimize > ( aMaximum Likelihood Estimation This: minimization problem can be re-written in sms of the residual r; of measurement i, which s defined as — ne mean ju;, or the expected value E(z:) measurement = can be expressed as a nonlinear function relating the system zctor x to the it measurement. Square Bch residual r2 is weighted by Wit = 7; > Which is inversely related to the assumed error ariance for that measurement Hence, the minimization problem of Equat| (4) will be equivalent to minimizing the weigh sum of squares of the residuals or solving the lowing optimization problem for the state ve minimize )_ Wis subject to =; =hi(x) +7 The solution of the above optimization prob is called the weighted least squares (WLS) ¢ mator for xon problem can be re-written in residual r, of measurement i, which — E(z) xpected value E(>; an be expressed as ting the system urement. Square s weighted by W 5 assumed error Hence, the minimization problem of Equat (4) will be equivalent to minimizing the weigh! sum of squares of the residuals or solving the lowing optimization problem for the state v minimize S> Wis subject to =, =hy(.r) The solution of the al is called the weight mator for fe optimization pro st squares (WLSThis minimization problem can be re-written in terms of the residual r; of measurement i, which is defined as: Wi =a — E(x) where the mean j1;, or the expected value E(=;) of the measurement =; can be expressed as h,(), a nonlinear function relating the system state vector x to the it* measurement. Square of each residual r? is weighted by Wis = 7; >, which is inversely related to the assumed error for that measurement. Maximum Likelihood Estimation Hence, the minimization problem of Equa: (4) will be equivalent to minimizing the wei sum of squares of the residuals or solving the lowing optimization problem for the state ved zoe SWar? ~ subject to =; = hj(x) + ri, The solution of the above optimization pro is called the weighted least squares (WLS mator for .r.Consider the set of measurements given by the vector =: =m Iam (214302)---» 2m) em \ AE = [hry (a) hale) -.- Men (©)] hh,(-r) is the nonlinear function relating measurement i to the state vector :r is the system state vector is the vector of measurement errors.The following assumptions are commonly made,regarding the statistical properties of the measurement errors: E (e;) =0, i=1,...,m leasurement errors are independent, i.e. The standard deviation o; of each measurement is calculated to reflect the expected accuracy of the corresponding meter used. The WLS estimator will minimize the following objective function: Measurement Model and Assumptions 4 —hi(a))” Ris 1 =[e=h(@)P ROE — A(z) At the minimum, the first-order optimality c| ditions will have to be satisfied. These can expressed in compact form as follows: aS) _ te ‘z)R[z — h(x) (2) = where: h(x) ] | H(x) =he following assumptions are commonly made,regarding the statistical properties of the measurement errors: ° E(e,) =0, i=1,.....m @ Measurement errors are independent, i.e. Eifesesi— 0} Hence, Cov(e) [e- o2} mi i standard deviation 7; of each measurement alculated to reflect the expected accuracy onding meter used rr LS estimator will minimize the following ctive function Ta) =o (zi = hile)? / Ris 1 =[e—h(x)|7 R= — (a At the minimum, the first-order optimality ¢ ditions will have to be satisfied. These can expressed in compact form as follows A(xExpanding the non-linear function g(.r) into its Taylor series around the state vector x* yields: g(x) = 9 (x*) G (2*) (c—2*) +-- lecting the higher order terms leads to an terative solution scheme known as the Gauss- ton method as shown below: fee)! -9@) ©) here, k: is the iteration index, x* is the solution tor at iteration k, and = HT (r*)- Ro - H (2*)