Skip to main content

Real Power Loss Reduction by Extreme Learning Machine Based Leontodon Algorithm

Abstract

In this paper Extreme Learning Machine based Leontodon Algorithm (ELMLA) has been applied for solving the Real Power loss reduction problem. Key objectives of this paper are Real power loss reduction, Voltage stability enhancement and voltage deviation minimization. Leontodon Algorithm (LA) technique is an innovative swarm based optimization algorithm. In evolutionary procedure of LA, the eminence of the seeds engendered by Leontodon is rutted, and the outstanding seeds will be reserved and appraised, whereas the deprived seeds are rejected. In order to define whether a seed is tremendous or not, an enhancement of Leontodon algorithm with extreme learning machine (ELMLA) is projected in this paper. Based on fitness values the Leontodon population is segregated into outstanding and deprived Leontodons. Subsequently outstanding and deprived Leontodons are apportioned corresponding labels as +1 if outstanding or − 1 if deprived), and it has been considered as a training set, which built on extreme learning machine. Lastly, the design is applied to categorize the Leontodon seeds as excellent or deprived. Only outstanding Leontodon seeds are selected to take part in evolution procedure. Legitimacy of the Extreme Learning Machine based Leontodon Algorithm (ELMLA) is substantiated in IEEE 30 bus system (with and devoid of L-index). Actual Real Power loss reduction is reached. Proportion of actual Real Power loss reduction is augmented.

Introduction

At present, the power system network immediately prerequisites to function at complete ability since the demanding venture in power system areas. The optimal reactive power dispatch (ORPD) problem can be deliberated as a vital part of the optimal power flow (OPF) problem. It is a large-scale, nonlinear, discrete, and optimization problem [1] and refers to the reasonable regulation of reactive power through various technologies under the condition of sufficient reactive power, so as to achieve the optimal distribution of reactive power and the reasonable compensation of reactive power for various loads. The control variables of the ORPD problems include the generators, transformers tapings, shunt reactors, and other reactive power sources. In general, it is perplexing work to find an efficient and convenient method to operate a modern power system, because it must be considered the requirement to recompense the structure for repeatedly altering load demand and provide energy of a high superiority [2]. Consequently, proper distribution and efficient management of reactive power are the major issues which need to be solved urgently in our days [3]. ORPD problems have dealt with various decisions by using a large number of classical algorithms like linear programming [4], interior point methods [5], and Lagrange decomposition method [6]. From the survey about the use of classical algorithms, it may be observed that there are obvious drawbacks in these classical algorithms such as insecure convergence, continuity limit, and excessive numerical iterations [7]. Fortunately, as an alternative choice, many researchers have transformed the focus from classical algorithms into computational intelligence-based algorithms which are used to solve the ORPD problems in various power systems [8]. Due to the nonlinear, non-convex, and multimodal nature of the ORPD, most of the methods applied to deal with it are based on computational intelligence-based algorithms [9]. Cuckoo search (CS) algorithm [10], slime mould algorithm (SMA) [11], harmony search algorithm (HSA) [12], particle swarm optimization (PSO) algorithm [13, 14], chaotic krill herd (CKH) algorithm [15], genetic algorithm (GA) [16], immune algorithm (IA) [17], earthworm optimization algorithm (EWA) [18], elephant herding optimization (EHO) [19], moth search (MS) algorithm [20], Harris hawks optimization (HHO) [21], and artificial bee colony (ABC) algorithm [22] have been proposed and used to deal with the ORPD problem. With virtuous tractability, adaptability, and heftiness, they have fascinated countless consideration. Differential evolution (DE) algorithm is one of the most general intelligent optimization algorithms and was first put forward by Storn and Price in 1995 [23]. The foremost feature of DE is its simple structure, the low parameter requirements, and the affluence of use particularly in distributing with cases with nonlinear constraints [24]. Owing to the aforementioned compensations, DE has been extensively used in the power system engineering, such as reactive power dispatching [25,26,27,28,29] and capacitor arrangement [30]. To supplementary progress the convergence speed and optimization consequence of DE in dealing with ORPD, many researchers have made various modifications from different perspectives like regulating scaling factors and crossover factors, mixed with other algorithms and other operations. Awad et al. [31] promulgated an competent DE algorithm for optimal active-reactive power dispatch problems, in which an arithmetic recombination crossover factor and a fresh scaling factor based on Laplace distribution were adapted to augment the performance of the original DE algorithm. Wang et al. [32] proposed a differential evolution algorithm with an adaptive population size adjustment mechanism (SapsDE), which can adaptively determine a more appropriate mutation strategy and its parameter settings according to the previous state at different stages of the evolution process, by this means improving the enactment of DE. Zhang et al. [33] alienated the DE population into manifold groups firstly and presented a self-adaptive strategy for the control parameters for the purpose of achieving better results. Zhang et al. [34] projected a self-adaptive differential evolution algorithm (JADE), which avoids the obligation for prior knowledge of parameter settings, so it can work well deprived of user interaction. Gao et al. [35] projected an original selection mechanism to augment the overall DE algorithm (NSODE), which selects new individuals from N parents and N children as N better solutions to achieve better optimization results. The goalmouth of adapted DE algorithm in [36] was to increase the convergence speed of the original DE by appraising the adaptive scaling factor, which was able to dynamically exchange information for each generation. Wang et al. [37] proposed a hybrid backtracking search optimization algorithm with differential evolution (HBD), which speeds up function convergence. Mohammad et al. [38] projected a new hybrid algorithm based on the cluster center initialization algorithm (CCIA), bee algorithm (BA), and differential evolution (DE) (called CCIA-BADE-K) and evaluated its performance through standard data sets. Surender [39] presented a hybrid DE and harmony search algorithm for the optimal power flow problem. In [40], monarch butterfly optimization (MBO) is used to extricate the problem of optimal power flow (OPF) for standard IEEE 30- and 118-bus test power systems. Pulluri et al. [41] introduced an enhanced self-adaptive DE with mixed crossover algorithm for dealing with the multi-objective optimal power flow (MO-OPF) problems with conflicting objectives. Although some improved DE algorithms for the ORPD problems have been achieved [31,32,33,34,35,36,37,38,39,40,41], further investigation is still indispensable for the complexity of objectives and constraints. Yet many approaches failed to reach the global optimal solution. In this paper Extreme Learning Machine based Leontodon Algorithm (ELMLA) is applied to solve the Real Power loss reduction problem. Leontodon algorithm is a swarm based algorithm, in which Central Leontodon (CL) and Subordinate Leontodons (SL) will form the Leontodon population, and seeding ways of them are dissimilar. For Central Leontodon (CL) possess one seeding way and it will succour to move out of local optimum. Based on a selection strategy appropriate Leontodons are selected to pass in the subsequent iteration. In the course of the evolution of Leontodon Algorithm (LA), outstanding and deprived Leontodon seeds can be engendered while seeding. The outstanding Leontodon seeds would be reserved, and deprived Leontodon seeds are rejected, this process will diminish the depletion of assessments effectually. In the interim, the sum of assessments kept can be used to assess more outstanding Leontodon seeds, which can effectually progress the performance of LA. Consequently, categorizing Leontodon seeds is a virtuous choice for refining performance of LA, and the procedure of differentiating whether the Leontodon seed is outstanding or deprived can be considered as a sorting procedure. Input weight and counterbalances of Extreme learning machine (ELM) are engendered arbitrarily, since it is a neural network learning procedure. In the meantime sum of hidden layer nodes must to be fixed. ELM has quicker learning swiftness and enhanced simplification performance. Owing to its gains, the ELM is applied to LA approach to knob the sorting. Grounded on this, an enhancement of evolution procedure of Leontodon algorithm with extreme learning machine (ELMLA) is projected in this paper. In ELMLA, the Leontodon population will be alienated into outstanding (+1) and deprived (−1) grounded on fitness values and this is considered as training dataset. Sequentially the cataloguing design will be constructed by ELM. Lastly, this sorting design is used to categorize the Leontodon seeds. Then the outstanding Leontodon seeds will be reserved and appraised. Sagacity of Extreme Learning Machine based Leontodon Algorithm (ELMLA) is confirmed by corroborated in IEEE 30 bus system (with and devoid of L-index). Proposed Extreme Learning Machine based Leontodon Algorithm (ELMLA)Appraisal of loss has been done with PSO, modified PSO, improved PSO, comprehensive learning PSO, Adaptive genetic algorithm, Canonical genetic algorithm, enhanced genetic algorithm, Hybrid PSO-Tabu search (PSO-TS), Ant lion (ALO), quasi-oppositional teaching learning based (QOTBO), improved stochastic fractal search optimization algorithm (ISFS), harmony search (HS), improved pseudo-gradient search particle swarm optimization and cuckoo search algorithm. Power loss abridged proficiently and proportion of the power loss lessening has been enriched. Real Power loss reduction is achieved. Proportion of Real power loss reduction is augmented.

Problem Formulation

Objective function of the problem is mathematically defined in general mode by,

$$ Minimization\ \overset{\sim }{F}\left(\overline{x},\overline{y}\right) $$
(1)

Subject to

$$ E\left(\overline{x},\overline{y}\right)=0 $$
(2)
$$ I\left(\overline{x},\overline{y}\right)=0 $$
(3)

Minimization of the Objective function is the key and it defined by “F”. Both E and I indicate the control and dependent variables. “x” consist of control variables which are reactive power compensators (Qc), dynamic tap setting of transformers –dynamic (T), level of the voltage in the generation units (Vg).

$$ x=\left[{VG}_1,..,{VG}_{Ng};{QC}_1,..,{QC}_{Nc};{T}_1,..,{T}_{N_T}\right] $$
(4)

“y” consist of dependent variables which has slack generator PGslack, level of voltage on transmission lines VL, generation units reactive power QG, apparent power SL .

$$ y=\left[{PG}_{slack};{VL}_1,..,{VL}_{N_{Load}};{QG}_1,..,{QG}_{Ng};{SL}_1,..,{SL}_{N_T}\right] $$
(5)

The fitness function (F1) is defined to reduce the power loss (MW) in the system is written as,

$$ {F}_1={P}_{Min}=\mathit{\operatorname{Min}}\left[{\sum}_m^{NTL}{G}_m\left[{V}_i^2+{V}_j^2-2\ast {V}_i{V}_j\mathit{\cos}{\mathrm{\O}}_{ij}\right]\right] $$
(6)

Number of transmission line indicated by “NTL”, conductance of the transmission line between the ith and jth buses, phase angle between buses i and j is indicated by Øij.

Minimization of Voltage deviation fitness function (F2) is given by,

$$ {F}_2=\mathit{\operatorname{Min}}\left[{\sum}_{i=1}^{N_{LB}}{\left|{V}_{Lk}-{V}_{Lk}^{desired}\right|}^2+{\sum}_{i=1}^{Ng}{\left|{Q}_{GK}-{Q}_{KG}^{Lim}\right|}^2\right] $$
(7)

Load voltage in kth load bus is indicated by VLk, voltage desired at the kth load bus is denoted by \( {V}_{Lk}^{desired} \), reactive power generated at kth load bus generators is symbolized by QGK, then the reactive power limitation is given by \( {Q}_{KG}^{Lim} \), then the number load and generating units are indicated by NLB and Ng.

Then the voltage stability index (L-index) fitness function (OF3)is given by,

$$ {F}_3=\mathit{\operatorname{Min}}\ {L}_{Max} $$
(8)
$$ {L}_{Max}=\mathit{\operatorname{Max}}\left[{L}_j\right];j=1;{N}_{LB} $$
(9)
$$ \left\{\begin{array}{c}{L}_j=1-{\sum}_{i=1}^{NPV}{F}_{ji}\frac{V_i}{V_j}\\ {}{F}_{ji}=-{\left[{Y}_1\right]}^1\left[{Y}_2\right]\end{array}\right. $$
(10)

Such that

$$ {L}_{Max}=\mathit{\operatorname{Max}}\left[1-{\left[{Y}_1\right]}^{-1}\left[{Y}_2\right]\times \frac{V_i}{V_j}\right] $$
(11)

Then the equality constraints are

$$ 0={PG}_i-{PD}_i-{V}_i{\sum}_{j\in {N}_B}{V}_j\left[{G}_{ij}\mathit{\cos}\left[{\mathrm{\O}}_i-{\mathrm{\O}}_j\right]+{B}_{ij}\mathit{\sin}\left[{\mathrm{\O}}_i-{\mathrm{\O}}_j\right]\right] $$
(12)
$$ 0={QG}_i-{QD}_i-{V}_i{\sum}_{j\in {N}_B}{V}_j\left[{G}_{ij}\mathit{\sin}\left[{\mathrm{\O}}_i-{\mathrm{\O}}_j\right]+{B}_{ij}\mathit{\cos}\left[{\mathrm{\O}}_i-{\mathrm{\O}}_j\right]\right] $$
(13)

Inequality constraints

$$ {\mathbf{P}}_{\mathbf{gslack}}^{\mathbf{min}}\le {\mathbf{P}}_{\mathbf{gslack}}\le {\mathbf{P}}_{\mathbf{gslack}}^{\mathbf{max}} $$
(14)
$$ {\mathbf{Q}}_{\mathbf{g}\mathbf{i}}^{\mathbf{min}}\le {\mathbf{Q}}_{\mathbf{g}\mathbf{i}}\le {\mathbf{Q}}_{\mathbf{g}\mathbf{i}}^{\mathbf{max}},\mathbf{i}\in {\mathbf{N}}_{\mathbf{g}} $$
(15)
$$ {\mathbf{VL}}_{\mathbf{i}}^{\mathbf{min}}\le {\mathbf{VL}}_{\mathbf{i}}\le {\mathbf{VL}}_{\mathbf{i}}^{\mathbf{max}},\mathbf{i}\in \mathbf{NL} $$
(16)
$$ {\mathbf{T}}_{\mathbf{i}}^{\mathbf{min}}\le {\mathbf{T}}_{\mathbf{i}}\le {\mathbf{T}}_{\mathbf{i}}^{\mathbf{max}},\mathbf{i}\in {\mathbf{N}}_{\mathbf{T}} $$
(17)
$$ {\mathrm{Q}}_{\mathrm{c}}^{\mathrm{min}}\le {\mathrm{Q}}_{\mathrm{c}}\le {\mathrm{Q}}_{\mathrm{C}}^{\mathrm{max}},\mathrm{i}\in {\mathrm{N}}_{\mathrm{C}} $$
(18)
$$ \left|{SL}_i\right|\le {S}_{L_i}^{max},\mathrm{i}\in {\mathrm{N}}_{\mathrm{TL}} $$
(19)
$$ {\mathrm{VG}}_{\mathrm{i}}^{\mathrm{min}}\le {\mathrm{VG}}_{\mathrm{i}}\le {\mathrm{VG}}_{\mathrm{i}}^{\mathrm{max}},\mathrm{i}\in {\mathrm{N}}_{\mathrm{g}} $$
(20)

Then the multi objective fitness (MOF) function has been defined by,

$$ MOF={F}_1+{x}_i{F}_2+y{F}_3={F}_1+\left[{\sum}_{i=1}^{NL}{x}_v{\left[{VL}_i-{VL}_i^{min}\right]}^2+{\sum}_{i=1}^{NG}{x}_g{\left[{QG}_i-{QG}_i^{min}\right]}^2\right]+{x}_f{F}_3 $$
(21)

Where real power loss reduction fitness function (F1), Minimization of Voltage deviation fitness function (F2) and voltage stability index (L-index) fitness function (F3) are added to construct the multi objective fitness (MOF) function

$$ {VL}_i^{min}=\left\{\begin{array}{c}{VL}_i^{max},{VL}_i>{VL}_i^{max}\\ {}{VL}_i^{min},{VL}_i<{VL}_i^{min}\end{array}\right. $$
(22)
$$ {QG}_i^{min}=\left\{\begin{array}{c}{QG}_i^{max},{QG}_i>{QG}_i^{max}\\ {}{QG}_i^{min},{QG}_i<{QG}_i^{min}\end{array}\right. $$
(23)

Extreme Learning Machine Based Leontodon Algorithm

Leontodon algorithm is a swarm based algorithm, in which Central Leontodon (CL) and Subordinate Leontodons (SL) will form the Leontodon population, and seeding methods of them are divergent. The outstanding Leontodon seeds will be reserved, and deprived Leontodon seeds are rejected, this process will reduce the exhaustion of assessments efficaciously. In pro tem, the sum of assessments kept can be used to evaluate more outstanding Leontodon seeds, which can efficiently progress the performance of Leontodon Algorithm (LA). Therefore, categorizing Leontodon seeds is a righteous choice for purifying performance of LA, and the method of distinguishing whether the Leontodon seed is outstanding or deprived can be considered as a categorization process. Input weight and counterbalances of Extreme learning machine (ELM) are created capriciously, since it is a neural network learning procedure. In the meantime sum of hidden layer nodes must to be fixed. ELM has quicker learning swiftness and enhanced simplification performance. Owing to its gains, the ELM is applied to LA approach to knob the sorting. With reference to this an augmented Leontodon algorithm with extreme learning machine (ELMLA) is projected in this paper. In ELMLA, the Leontodon population will be alienated into outstanding (+1) and deprived (−1) grounded on fitness values and this is considered as training dataset. In LA approach around are three types of operators; standard seeding operator, mutation seeding operator and picking plan operator. The sum of Leontodon seeds engendered is determined and for the Leontodon (Yi), the sum can be computed as,

$$ Number\ of\ Leontodon\ Seeds\left({NLS}_i\right)=\left\{\begin{array}{c}{NLS}_{maximum}\times \frac{Fitness_{maximum}- Fitness\left({Y}_i\right)+\varepsilon }{F\mathrm{i}{tness}_{maximum}-{Fitness}_{min imum}+\varepsilon },\kern6.25em {NLS}_i>{NLS}_{min}\\ {}{NLS}_{min imum}\kern23em {NLS}_i\le {NLS}_{min}\end{array}\right. $$
(24)

In LA approach, the Leontodon with least fitness value is called Central Leontodon (CL). Meant for CL, its seeding radius is animatedly computed with Declining factor (D), Growing Tendency (δ) and Growing factor (G) as follows,

$$ Seeding\ Raduis\ \left({SR}_{CL}(t)\right)=\left\{\begin{array}{c} Upper\ Bound- Lower\ Bound\kern0.75em ;t=1\\ {}{SR}_{CL}(t)-1\times (D);\kern2.5em \delta =1,t>1\kern6.25em \\ {}{SR}_{CL}(t)-1\times (G);\delta \ne 1,t>1\ \end{array}\right. $$
(25)
$$ Growing\ Tendency\ \left(\delta \right)=\frac{Fitness_{CL}(t)+\varepsilon }{Fitness_{CL}\left(t-1\right)+} $$
(26)

Seeding radius of the Subordinate Leontodons (SL) is computed by,

$$ Seeding\ Raduis\ \left({SR}_{SL}(t)\right)=\left\{\begin{array}{c} Upper\ Bound- Lower\ Bound,\kern0.75em t=1\\ {} Weight\left(\eta \right)\times {SR}_{SL}\left(t-1\right)+\left({\left\Vert {Y}_{CL}\right\Vert}_{\infty }-{\left\Vert {Y}_i\right\Vert}_{\infty}\right), otherwise\ \end{array}\right. $$
(27)
$$ Weight\ parameter\ \left(\eta \right)=1-\frac{Assessm\mathrm{e} nt\ (A)}{ Assessm ent\ {(A)}_{max}} $$
(28)

The standard Leontodon seeds are engendered as follows,

figurea

The Central Leontodon (CL) has alternative seeding mode, specifically mutation seeding, and engenders mutation Leontodon seeds as shown below,

$$ {LS}_{Mutation}={Y}_{CL}\cdotp \left(1+ Levy. Random\right) $$
(29)

Procedure of engendering mutation Leontodon seeds as follows,

figurec

In Leontodon Algorithm (LA) the Central Leontodon (CL) should be reserved and enduring Leontodons are chosen by the picking plan probability (PPi) as follows,

$$ {PP}_i=\frac{\left| Fitness\left({Y}_i\right)-{Fitness}_{Average}\right|}{\sum_{n=1}^{LSA} Fitness\left({Y}_n\right)} $$
(30)

Initially, through Standard seeding operator and then Mutation seeding operator engenders the standard and mutation Leontodon seeds and then N Leontodons are selected by picking plan operator. Subsequently introduction of standard seeding operator, mutation seeding operator and picking plan operator, there is a common indulgent of complete outline of LA approach as shown below,

  1. a.

    Start

  2. b.

    Input the value of N

  3. c.

    Engender the population of Leontodon

  4. d.

    Compute the fitness value of Leontodon

  5. e.

    Engender the standard Leontodon seeds by applying the standard seeding operator procedure

  6. f.

    Engender the Mutation Leontodon seeds by applying the Mutation seeding operator procedure

  7. g.

    Compute the fitness values of all types of Leontodon seeds

  8. h.

    Choose “N” Leontodons by using picking plan operator

  9. i.

    Continue the process until stop criterion met

  10. j.

    Return the Central Leontodon (CL)

Extreme learning machine (ELM) is applied and learning speed of feed-forward neural networks is poised of input, hidden and output layer.

For N capricious examples (yi, Ti); yi = [yi1, yi2, .., yidn]T ∈ SRdn, Ti = [Ti1, Ti2, .., Tidn]T ∈ SRdn, (31)

$$ {\sum}_{i=1}^N{\beta}_i\cdotp kn\left({\omega}_i{y}_j+{a}_i\right)={T}_j,j=1,2,3,..,N $$
(32)
$$ Output\ matrix\ (Q)\cdotp Output\ weight\left(\beta \right)=T $$
(33)
$$ Q\left({y}_1,..{y}_L;{\omega}_1,..,{\omega}_L;{a}_1,..,{a}_l\right)=\left[\begin{array}{ccc} kn\left({\omega}_1{y}_1+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_1+{a}_L\right)\\ {}\vdots & \ddots & \vdots \\ {} kn\left({\omega}_1{y}_N+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_N+{a}_L\right)\end{array}\right] $$
(34)
$$ \beta ={Q}^{-1}\cdotp T $$
(35)

The procedure of the Extreme Learning Machine (ELM) is given as below

  1. a.

    Begin

  2. b.

    Input the complete data

  3. c.

    Test and Training set are created from the general data

  4. d.

    With reference to the “training set” of data - compute the value of Output matrix (Q)

    $$ Q\left({y}_1,..{y}_L;{\omega}_1,..,{\omega}_L;{a}_1,..,{a}_l\right)=\left[\begin{array}{ccc} kn\left({\omega}_1{y}_1+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_1+{a}_L\right)\\ {}\vdots & \ddots & \vdots \\ {} kn\left({\omega}_1{y}_N+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_N+{a}_L\right)\end{array}\right] $$
  5. e.

    Compute the value of weight (output)

    $$ \beta ={Q}^{-1}\cdotp T $$
  6. f.

    With reference to the “test set” of data - compute the value of Output matrix (Q)

    $$ Q\left({y}_1,..{y}_L;{\omega}_1,..,{\omega}_L;{a}_1,..,{a}_l\right)=\left[\begin{array}{ccc} kn\left({\omega}_1{y}_1+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_1+{a}_L\right)\\ {}\vdots & \ddots & \vdots \\ {} kn\left({\omega}_1{y}_N+{a}_1\right)& \cdots & kn\left({\omega}_L{y}_N+{a}_L\right)\end{array}\right] $$
  7. g.

    Compute the real value by β and Q

  8. h.

    Calculation of error value

  9. i.

    Comparison of real value with expected value

  10. j.

    Return the error value

  11. k.

    End

To speed up the performance of the Leontodon Algorithm (LA) in this paper Extreme learning machine (ELM) has been integrated. In the basic procedure of LA approach in each generation "M " Leontodon seeds are engendered, if kn(kn ≤ M)Leontodon seeds outstanding (+1) means then M − kn Leontodon seeds is in deprived stage (−1). By utilizing ELM approach into the procedure, if Y = y1, y2, . . , yM is the set of Leontodon seeds then first level of kn Leontodon seeds are in outstanding (+1) and the other Leontodon seeds (yk + 1, yk + 2, .., yM) are in deprived stage (−1). This mode partition will diminish the assessment time, since the deprived (−1) Leontodon seeds will be rejected and fitness value will be calculated only for outstanding (+1) Leontodon seeds. Mainly additional assessment will be utilized for more appraisal of outstanding (+1) Leontodon seeds.

Extreme Learning Machine based Leontodon Algorithm (ELMLA) and fig. 1 shows the flowchart of the algorithm

figureb
Fig. 1
figure1

Flow chart of Extreme Learning Machine based Leontodon Algorithm

Simulation Results

With considering L- index (voltage stability), Extreme Learning Machine based Leontodon Algorithm (ELMLA) is substantiated in IEEE 30 bus system [49, 50]. Assessment of real power loss has been done with PSO, amended PSO, enhanced PSO, widespread learning PSO, Adaptive genetic algorithm, Canonical genetic algorithm, enriched genetic algorithm, Hybrid PSO-Tabu search (PSO-TS), Ant lion (ALO), quasi-oppositional teaching learning based (QOTBO), improved stochastic fractal search optimization algorithm (ISFS), harmony search (HS), improved pseudo-gradient search particle swarm optimization and cuckoo search algorithm. Power loss abridged competently and proportion of the Real Power loss reduction has been enriched. Predominantly voltage constancy enrichment achieved with minimized voltage deviation. In Table 1 shows the loss appraisal, Table 2 shows the voltage deviation evaluation and Table 3 gives the L-index assessment. Figures 2, 3, 4 gives graphical appraisal.

Table 1 Assessment of Real Power loss reduction
Table 2 Evaluation of voltage deviation
Table 3 Assessment of voltage constancy
Fig. 2
figure2

Appraisal of actual power loss

Fig. 3
figure3

Appraisal of Voltage deviation

Fig. 4
figure4

Appraisal of voltage constancy

Then Projected Extreme Learning Machine based Leontodon Algorithm (ELMLA) is corroborated in IEEE 30 bus test system deprived of L- index. Loss appraisal is shown in Tables 4. Figure 5 gives graphical appraisal between the approaches with orientation to Real power loss.

Table 4 Assessment of true power loss
Fig. 5
figure5

Appraisal of Real Power Loss

Table 5 shows the convergence characteristics of Extreme Learning Machine based Leontodon Algorithm (ELMLA). Figure 6 shows the graphical representation of the characteristics.

Table 5 Convergence characteristics
Fig. 6
figure6

Convergence characteristics of ELMLA

Conclusion

Extreme Learning Machine based Leontodon Algorithm (ELMLA) abridged the Real power loss dexterously. ELMLA substantiated in IEEE 30- bus test system with L- index and devoid of L-index. For Central Leontodon (CL) possess one seeding way and it will succour to move out of local optimum. Based on a selection strategy appropriate Leontodons are selected to pass in the subsequent iteration. To speed up the performance of the Leontodon Algorithm (LA) in this paper Extreme learning machine (ELM) has been integrated. In the basic procedure of LA approach in each generation "M " Leontodon seeds are engendered, if kn(kn ≤ M)Leontodon seeds outstanding (+1) means then M − kn Leontodon seeds is in deprived stage (−1). By utilizing ELM approach into the procedure, if Y = y1, y2, . . , yM is the set of Leontodon seeds then first level of kn Leontodon seeds are in outstanding (+1) and the other Leontodon seeds (yk + 1, yk + 2, .., yM) are in deprived stage (−1). This mode partition will diminish the assessment time, since the deprived (−1) Leontodon seeds will be rejected and fitness value will be calculated only for outstanding (+1) Leontodon seeds. Mainly additional assessment will be utilized for more appraisal of outstanding (+1) Leontodon seeds. Therefor categorizing Leontodon seeds is a righteous choice for purifying performance of LA, and the method of distinguishing whether the Leontodon seed is outstanding or deprived can be considered as a categorization process. Input weight and counterbalances of Extreme learning machine (ELM) are created capriciously, since it is a neural network learning procedure. Extreme Learning Machine based Leontodon Algorithm (ELMLA) commendably reduced the power loss and proportion of Real Power loss reduction has been upgraded. Convergence characteristics show the better performance of the proposed ELMLA algorithm. Assessment of power loss has been done with other customary reported algorithms. Real Power loss reduction achieved by ELMLA is 4.5005 (MW) with considering the Voltage stability index. And without considering Voltage stability index Real Power loss reduction achieved by ELMLA is13.90 (MW). Percentage of power loss reduction attained is 20.79. In future proposed Extreme Learning Machine based Leontodon Algorithm (ELMLA) can be applied in other areas of power system optimization problem like unit commitment, Economic dispatch problem. Mainly the proposed algorithm can be applied in practical test systems. And in other areas of engineering like image processing, medical field also proposed algorithm can be applied.

Data availability

No data available.

References

  1. 1.

    Sakr WS, El-Sehiemy RA, Azmy AM (2017) Adaptive differential evolution algorithm for efficient reactive power management. Appl Soft Comput 53:336–351

    Article  Google Scholar 

  2. 2.

    Abaci K, Yamacli V (2016) Differential search algorithm for solving multi-objective optimal power flow problem. Int J Electr Power Energy Syst 79:1–10

    Article  Google Scholar 

  3. 3.

    Medani K, Sayah S, Bekrar A (2018) Whale optimization algorithm based optimal reactive power dispatch: a case study of the Algerian power system. Electr Power Syst Res 163:696–705

    Article  Google Scholar 

  4. 4.

    Deeb NI, Shahidehpour SM (1988) An efficient technique for reactive power dispatch using a revised linear programming approach. Electr Power Syst Res 15(2):121–134

    Article  Google Scholar 

  5. 5.

    Zhu JZ, Xiong XF (2003) Optimal reactive power control using modified interior point method. Electr Power Syst Res 66(2):187–192

    Article  Google Scholar 

  6. 6.

    Sousa VA, Baptista EC, Costa GRM (2012) Optimal reactive power flow via the modified barrier Lagrangian function approach. Electr Power Syst Res 84(1):159–164

    Article  Google Scholar 

  7. 7.

    Ghasemi M, Taghizadeh M, Ghavidel S, Aghaei J, Abbasian A (2015) Solving optimal reactive power dispatch problem using a novel teaching-learning-based optimization algorithm. Eng Appl Artif Intell 39:100–108

    Article  Google Scholar 

  8. 8.

    Sayah S (2018) Modified differential evolution approach for practical optimal reactive power dispatch of hybrid AC-DC power systems. Appl Soft Comput 73:591–606

    Article  Google Scholar 

  9. 9.

    Rojas DG, Lezama JL, Villa W (2016) Metaheuristic techniques applied to the optimal reactive power dispatch: a review. IEEE Lat Am Trans 14(5):2253–2263

    Article  Google Scholar 

  10. 10.

    Raha SB, Mandal KK, Chakraborty N (2018) Hybrid SMES based reactive power dispatch by cuckoo search algorithm. IEEE Trans Ind Appl 55(1):907–917

    Article  Google Scholar 

  11. 11.

    Manar M, Hegazy R, Mokhtar A, Ahmed EM (2020) A new strategy based on slime mould algorithm to extract the optimal model parameters of solar PV panel, Sustainable Energy Technologies and Assessments, vol. 42, Article ID 100849

  12. 12.

    Karthikaikannan D, Ravi G (2018) Optimal reactive power dispatch considering multi-type FACTS devices using harmony search algorithms. Automatika 59(3–4):311–322

    Article  Google Scholar 

  13. 13.

    Devi S, Geethanjali M (2014) Optimal location and sizing determination of distributed generation and DSTATCOM using particle swarm optimization algorithm. Int J Electr Power Energy Syst 62:562–570

    Article  Google Scholar 

  14. 14.

    Aljohani TM, Ebrahim AF, Mohammed O (2019) Single and multiobjective optimal reactive power dispatch based on hybrid artificial physics-particle swarm optimization. Energies 12(12):2333

    Article  Google Scholar 

  15. 15.

    Wang G-G, Guo L, Gandomi AH, Hao G-S, Wang H (2014) Chaotic krill herd algorithm. Inf Sci 274:17–34

    MathSciNet  Article  Google Scholar 

  16. 16.

    Subbaraj P, Rajnarayanan PN (2009) Optimal reactive power dispatch using self-adaptive real coded genetic algorithm. Electr Power Syst Res 79(2):374–381

    Article  Google Scholar 

  17. 17.

    Taher SA, Afsari SA (2014) Optimal location and sizing of DSTATCOM in distribution systems by immune algorithm. Int J Electr Power Energy Syst 60:34–44

    Article  Google Scholar 

  18. 18.

    Ghosh I, Roy PK (2019) Application of earthworm optimization algorithm for solution of optimal power flow, 2019 international conference on opto-electronics and applied optics, Optronix, Article ID 8862335

  19. 19.

    Dhillon SS, Agarwal S, Wang G-G, Lather JS (2020) Automatic generation control of interconnected power systems using elephant herding optimization. Intell Comput Tech Smart Energy Syst 607:9–18

    Article  Google Scholar 

  20. 20.

    Singh P, Bishnoi SK, Meena NK (2018) Moth search optimization for optimal integration of DERs for annual energy loss minimization in distribution systems, in Proceedings of the India International Conference on Power Electronics, Article ID 8709410

  21. 21.

    Qu C, He W, Peng X, Peng X (2020) Harris hawks optimization with information exchange. Appl Math Model 84:52–75

    MathSciNet  Article  Google Scholar 

  22. 22.

    He X, Wang W, Jiang J, Xu L (2015) An improved artificial bee colony algorithm and its application to multi-objective optimal power flow. Energies 8(4):2412–2437

    Article  Google Scholar 

  23. 23.

    Storn R, Price K (1997) Differential evolution: a simple and efficient heuristic for global optimization over continuous space. J Glob Optim 11(4):341–359

    MathSciNet  Article  Google Scholar 

  24. 24.

    Ela AAAE, Abido MA, Spea SR (2011) Differential evolution algorithm for optimal reactive power dispatch. Electr Power Syst Res 81(2):458–464

    Article  Google Scholar 

  25. 25.

    Ramirez JM, Gonzalez JM, Ruben TO (2011) An investigation about the impact of the optimal reactive power dispatch solved by DE. Int J Electr Power Energy Syst 33(2):236–244

    Article  Google Scholar 

  26. 26.

    Li Y, Wang Y, Li B (2013) A hybrid artificial bee colony assisted differential evolution algorithm for optimal reactive power flow. Int J Electr Power Energy Syst 52(1):25–33

    Article  Google Scholar 

  27. 27.

    Varadarajan M, Swarup KS (2008) Differential evolutionary algorithm for optimal reactive power dispatch. Int J Electr Power Energy Syst 30(8):435–441

    Article  Google Scholar 

  28. 28.

    Ghasemi M, Ghanbarian MM, Ghavidel S, Rahmani S, Mahboubi Moghaddam E (2014) Modified teaching learning algorithm and double differential evolution algorithm for optimal reactive power dispatch problem: a comparative study, Inf Sci, vol. 278, pp. 231–249, 18 Mathematical Problems in Engineering

  29. 29.

    Sanam J, Ganguly S, Panda AK, Hemanth C (2017) Optimization of energy loss cost of distribution networks with the optimal placement and sizing of DSTATCOM using differential evolution algorithm. Arab J Sci Eng 42(7):2851–2865

    Article  Google Scholar 

  30. 30.

    Shaheen AM, El-Sehiemy RA, Farrag SM, Farrag SM (2018) Adequate planning of shunt power capacitors involving transformer capacity release benefit. IEEE Syst J 12(1):373–382

    Article  Google Scholar 

  31. 31.

    Awad NH, Ali MZ, Mallipeddi R, Suganthan PN (2019) An efficient differential evolution algorithm for stochastic OPF based active-reactive power dispatch problem considering renewable generators. Appl Soft Comput 76:445–458

    Article  Google Scholar 

  32. 32.

    Wang X, Zhao SG (2013) Differential evolution algorithm with self-adaptive population resizing mechanism, Mathematical Problems in Engineering, vol. 2013, Article ID 419372, 14 pages

  33. 33.

    Zhang X, Chen W, Dai C, Cai W (2010) Dynamic multigroup self-adaptive differential evolution algorithm for reactive power optimization. Int J Electr Power Energy Syst 32(5):351–357

    Article  Google Scholar 

  34. 34.

    Zhang JQ, Sanderson AC (2007) Self-adaptive Differential Evolution with Fast and Reliable Convergence Performance, 2007, pp. 2251–2258, IEEE Congress on Evolutionary Computation, Singapore

  35. 35.

    Gao D, Wang GG, Pedrycz W (2020) Solving fuzzy job-shop scheduling problem using DE algorithm improved by a selection mechanism. IEEE Trans Fuzzy Syst 28(12):3265–3275

    Article  Google Scholar 

  36. 36.

    Sakr WS, EL-Sehiemy RA, Azmy AM (2017) Adaptive differential evolution algorithm for efficient reactive power management. Appl Soft Comput 53:336–351

    Article  Google Scholar 

  37. 37.

    L. J. Wang, Y. W. Zhong, Y. L. Yin, W. T. Zhao, B. O. Wang, and Y. L. Xu (2015) A hybrid backtracking search optimization algorithm with differential evolution, Mathematical Problems in Engineering, vol. 2015, Article ID 769245, 16 pages

  38. 38.

    Mohammad BB, Hashim SZM, Nor ENB, Ahmed KZA (2015) An effective hybrid of bees algorithm and differential evolution algorithm in data clustering, Mathematical Problems in Engineering, vol. 2015, Article ID 240419, 17 pages

  39. 39.

    Surender SR (2019) Optimal power flow using hybrid differential evolution and harmony search algorithm. Int J Mach Learn Cybern 10:1077–1091

    Article  Google Scholar 

  40. 40.

    Yadav V, Ghoshal SP (2018) Optimal power flow for IEEE 30 and 118-bus systems using monarch butterfly optimization, in Proceedings of the 2018 Technologies for Smart-City Energy Security and Power, pp. 1–6, ICSESP), Bhubaneswar, India

  41. 41.

    Pulluri H, Naresh R, Sharma V (2017) An enhanced selfadaptive differential evolution based solution methodology for multiobjective optimal power flow. Appl Soft Comput 54:229–245

    Article  Google Scholar 

  42. 42.

    Sahli Z, Hamouda A, Bekrar A, Trentesaux D (2014) Hybrid PSO-tabu search for the optimal reactive power dispatch problem,” in Proceedings of the IECON 2014-40th Annual Conference of the IEEE Industrial Electronics Society, Dallas, TX, USA

  43. 43.

    Mouassa S, Bouktir T, Salhi A (2017) Ant lion optimizer for solving optimal reactive power dispatch problem in power systems. Eng Sci Technol Intl J 20(3):885–895

    Google Scholar 

  44. 44.

    Mandal B, Roy PK (2013) Optimal reactive power dispatch using quasi-oppositional teaching learning based optimization. Int J Electr Power Energy Syst 53:123–134

    Article  Google Scholar 

  45. 45.

    Khazali H, Kalantar M (2011) Optimal reactive power dispatch based on harmony search algorithm. Int J Electr Power Energy Syst 33(3):684–692

    Article  Google Scholar 

  46. 46.

    Tran HV, Pham TV, Pham LH, Le NT, Nguyen TT (2019) Finding optimal reactive power dispatch solutions by using a novel improved stochastic fractal search optimization algorithm. Telecommun Comput Electron Control 17(5):2517–2526

    Google Scholar 

  47. 47.

    Polprasert J, Ongsakul W, Dieu VN (2016) Optimal reactive power dispatch using improved pseudo-gradient search particle swarm optimization. Electric Power Components Syst 44(5):518–532

    Article  Google Scholar 

  48. 48.

    Duong TL , Duong MQ, Phan V-D, Nguyen TT (2020) Optimal Reactive Power Flow for Large-Scale Power Systems Using an Effective Metaheuristic Algorithm Hindawi Journal of Electrical and Computer Engineering Volume, Article ID 6382507, 11 pages https://doi.org/10.1155/2020/6382507

  49. 49.

    Muhammad Y, Khan R, Raja MAZ, Ullah F, Chaudhary NI, He Y (2020) Solution of optimal reactive power dispatch with FACTS devices: a survey. Energy

  50. 50.

    Illinois Center for a Smarter Electric Grid (ICSEG). Available online: https://icseg.iti.illinois.edu/ieee-30-bussystem/ (accessed on 25 February 2019)

  51. 51.

    Dai C, Chen W, Zhu Y, Zhang X (2009) Seeker optimization algorithm for optimal reactive power dispatch. IEEE T Power Syst 24(3):1218–1231

    Google Scholar 

  52. 52.

    Pandya S, Roy R (2015) Particle swarm optimization based optimal reactive power dispatch. Proceeding of the IEEE international conference on electrical, computer and communication technologies (ICECCT), pp: 1–5

  53. 53.

    Ali Nasser Hussain, Ali Abdulabbas Abdullah and Omar Muhammed Neda, (2018), Modified Particle Swarm Optimization for Solution of Reactive Power Dispatch, Res J Appl Sci Eng Technol 15(8): 316–327, 2018 DOI:https://doi.org/10.19026/rjaset.15.5917

  54. 54.

    Vishnu, Mini, Sunil. “An Improved Solution for Reactive Power Dispatch Problem Using Diversity-Enhanced Particle Swarm Optimization”. Energies. 13. 2862. Pp 2–21, 2020 https://doi.org/10.3390/en13112862

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Lenin Kanagasabai.

Ethics declarations

All ethics are followed

Conflicts of interest

No conflict of interest.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kanagasabai, L. Real Power Loss Reduction by Extreme Learning Machine Based Leontodon Algorithm. Technol Econ Smart Grids Sustain Energy 6, 16 (2021). https://doi.org/10.1007/s40866-021-00110-1

Download citation

Keywords

  • Optimal reactive power
  • Transmission loss
  • Extreme learning machine
  • Leontodon