A Chaos Disturbed Beetle Antennae Search Algorithm for a Multiobjective Distribution Network Reconfiguration Considering the Variation of Load and DG

As the distributed generation (DG) in a power supply and the user load demand constantly change in an actual distribution network, multiobjective optimal network reconfiguration considering variations in load and DG has become a major concern, which is important and required to make system operations safe and economical. The aim is to minimize the sum of the active power loss, the sum of the load balancing index and the sum of the maximum node voltage deviation index simultaneously during the reconfiguration period. Here, this article proposes a new Chaos Disturbed Beetle Antennae Search (CDBAS) algorithm to reduce the computational time and solve the multiobjective optimal problem of network reconfiguration. To adopt the Chaos Disturbed Beetle Antennae Search algorithm for solving this multiobjective problem, grey target decision-making technology is used to rank the beetles. Additionally, to the enhance the system static voltage stability and voltage quality, a grey target decision-making model is established to achieve a layer relationship between each index and the switching operation index. The plausibility and effectiveness of the presented methodology is verified on the modified IEEE 33, 69 and 118-Bus Test Radial Distribution Network. Finally, compared with other research methods in the literature, the CDBAS algorithm outperforms other algorithms and produces a quality decision solution.


I. INTRODUCTION
The combination of residential, commercial and industrial loads creates differences and time variabilities on the feeders of a distribution network, which make the operational control of the distribution network very complicated and pose challenges for system safety and economy as well as user comfort. Wind power and photovoltaics are increasingly developed sustainable clean energy sources that have attracted increasing interest in the last decade due to increasing power demands and the increasing consumption of fossil fuels. Thus, distributed generation (DG) into distribution networks has become popular considering the The associate editor coordinating the review of this manuscript and approving it for publication was Alexander Micallef .
technical advantages of DG units in reducing network losses, balancing demand overloads, improving the node voltage level and absorbing renewable energy [1]. With the active management of advanced metering facilities and information communication technology in the network topology structure and distributed power supply, a network reconstruction with different factors, which can deal with the time variabilities of load and DG more flexibly, is viable.
Distribution network reconfiguration can also be defined as the status of changing switches to improve system operation. For example, a dispatcher obtains the optimal network topology by opening a closed section switch and then closing an open tie switch under the condition of satisfying constraints such as voltage and current limits, radial structures and power flows. To date, many researchers have investigated network reconfiguration with classic techniques and heuristics for single and multiple objectives with the aim of achieving a safe and reliable operation point of the distribution system [2]- [4].
Many studies have exploited hybrid algorithms based on previously cited researchers who proposed methods for multiobjective distribution system reconfiguration in a structure, as in [5], [6], and other studies have considered the correlation among objectives by the grey correlation method in evolutionary programming and its role in optimizing particles by integrating the objectives. Additionally, some studies have focused on providing the multiobjective problem of network reconfiguration with different techniques, such as fuzzy evaluation theory [8]- [10] and cloud theory [11], and all of these approaches can obtain ideal solutions with better objectives [12], [13]. These methods, which are subject to uncertainty theory, can improve the quality of the solutions, especially for traditional distribution systems, but they cannot achieve the optimal variability of load and DG for a distribution network.
In the literature, studies have examined the concept of optimal multiobjective reconfiguration with the variability of load and DG in the distribution system. The authors of [14] used a multiobjective reconfiguration evolutionary technique of a distribution feeder system with wind turbines and fuel cells to consider the variability of wind power output and load demand for the benefit of minimum losses and better economic costs. Additionally, [15], [16] had the same objective and considered the same distributed generation with different techniques but did not address the calculation time, and other authors have studied multiobjective reconfiguration from the aspect of the convergence rate based on the security index to achieve the best result where the number of iterations is an obstacle [17], [18]. Some of these studies did not consider load time variation [19], [20]. Then, the author of [21] established a bi-objective dynamic reconfiguration model that considers the optimal equilibrium relationship between loss reduction and switching action reduction to obtain the best solution, but the load balance and voltage quality were not considered.
All of the abovementioned papers demonstrate the multiobjective reconfiguration of distribution networks based on specific types of metaheuristic and some new techniques to achieve different objectives, which are converted to a single objective aiming to find a trade-off solution by the most common methods, such as the compromise model [22]- [25], the entropy method [26], and the analytic hierarchy process [21], [27], without considering the mutual constraints among the objectives. However, the simple multiobjective traditional methods are no longer applicable for an automated distribution system with multiple conflicting objectives for scheduling known optimized solutions simultaneously. To solve the trade-off solution problem, we use the grey target decision-making technique to establish a layer relationship between active power loss, the load balancing index and the maximum node voltage deviation index with the number of switching operation indices for optimal multiobjective network reconfiguration. Here, this paper uses a Beetle Antennae Search algorithm based on Chaotic Disturbance (CDBAS) to solve the problem of multiobjective optimal network reconfiguration to reduce the calculation time through the individual advantage of the beetle. The Beetle Antennae Search, first designed and developed by Jiang X Y, is a bioinspired intelligent optimization algorithm, which was inspired by the foraging principle of beetles [28], [29]. The original Beetle Antennae Search (BAS) algorithm often converges slowly, has low accuracy and searches for local optima easily. To avoid the above shortcomings, the original BAS combined with the chaotic disturbance mechanism (CDBAS) approach is proposed. A grey target decision-making technique [30] is adapted to the CDBAS algorithm for determining the order of the beetles with the solution problem of multiobjective optimization network reconstruction. Finally, we simulate the proposed methodology in the modified IEEE 33-bus Distribution Network System and subsequently analyse the optimization results, which verify the feasibility and validity of the method. In addition, compared with other methods in the literature, the proposed methodology can achieve better for solution quality.
The main contributions of this paper are as follows: • We propose a new CDBAS method for a multiobjective optimization network reconstruction model, which enhances a reduction in calculation time.
• We propose a grey target decision-making technique for the order of beetles, which can achieve the actual solution required by personnel.
• To enhance the system voltage quality and stable system operation, we use the grey target decision-making method to achieve the layer relationship between each index and the number of switching operation indexes.
• To ensure safe and economical system operation, we carry out 5 reconfigurations according to the load curve over 24 hours. The logical framework of this work is as follows. Section II provides the load and DG curves. Section III introduces the objective functions and the constraint. Section IV describes the calculation process of the grey target decision-making method. Section V explains the CDBAS algorithm. Section VI demonstrates the simplification and encoding of topology based on a basic loop. Section VII shows the test results of 8 benchmark functions from the CDBAS algorithm, the results from the modified 33, 69 and 118-bus system and a comparison with other methods. Finally, Section VIII presents the paper's conclusions.

II. LOAD AND DG PATTERN
The variation in the DG power supply and user load demand during a typical day has a significant effect on the safe and economical operation of a power system.

A. LOAD DEMAND PATTERN
Since users have relatively fixed demand proportion curves during a typical day in the same region and season, the loads VOLUME 8, 2020  can be described by demand proportion curves per day [31]. In general, the load obtained by daily loads in summer in a certain region involves three categories: residential, industrial and commercial, as shown in Fig. 1. The load curves of each node are different, and the daily load curve P Li of node 'i' can be described as where S = {s|1,2,3} is the set of load types; s = 1 is residential, s = 2 is industrial, and s = 3 is commercial; P Ni is the rated power of load node 'i'; and θ si represents the ratio of load type 's' in the total load of node 'i'. δ s represents the demand curves of load type 's'.

B. TYPICAL DG OUTPUT
The duration curves of wind power and photovoltaic power generation are shown by curves for a day in summer in a certain region in Fig. 2. The load data of each node and the DG output can be obtained by prediction techniques in the actual operation of the system.

III. PROBLEM FORMULATION A. OBJECTIVE FUNCTION
Considering the optimization objective of network reconfiguration from the aspects of economy, voltage quality and load balance is practical. Here, the sum of the active power loss, the sum of the load balancing index and the sum of maximum node voltage deviation are used to test the performance for the distribution system during hourly or every few hours of reconfiguration.

1) POWER LOSS INDEX
The calculation formula of active power loss for a distribution system is described as where P m represents the active power of the mth branch, Q m represents the reactive power of the mth branch, R m represents the resistance of the mth branch, V m represents the initial voltage of the mth branch, D m is the binary value, D m represents the closed switch status in the mth branch when D m = 1, D m is the opening switch status for the ith branch when D m = 0, M represents the number of branches, and TL represents the time length.

2) LOAD BALANCING INDEX
Load balancing requires the transfer of the negative load on the heavy line to the light line for network reconstruction, the load-balancing index is calculated as where S max m represents the maximum complex power of branch 'm'.

3) VOLTAGE DEVIATION INDEX
where V i represents the virtual voltage of the ith node, V ir represents the rated voltage of the ith node, and N represents the total number of nodes.
where P DGi represents the active power of the DG connected to the ith node.

2) NETWORK NODES VOLTAGE LIMITS
where V min i and V max i are the allowable minimum and maximum values of the voltage at node i, respectively.

3) NETWORK LINE CAPACITY LIMITS
where S max i represents the allowable maximum capacity of branch 'k'.

4) RADIAL NETWORK CONSTRAINT
The network structure should meet radial topological constraints.

IV. GREY TARGET DECISION-MAKING MODEL
Because of the conflicting indexes, in this study, we use a grey target decision-making technique to solve the multiobjective problem to find the best reconfiguration solution. The optimization solution decision model is established by the grey target decision-making technique [30] considering the relationship between each index and the number of switching operation indexes.
Here, we define decision scheme set S = {S 1 , S 2 , · · · , S i , · · · , S n }, which is obtained by the CDBAS method, so switching operation index f D is modeled as follows: where D ji is the status of switch 'j' of reconfiguration scheme S i , D j,i+1 and D j,i−1 are the statuses of switch 'j' for the other two reconfiguration schemes S i , and f D is the switch operation index of reconfiguration scheme 'i'. Additionally, the grey target decision-making technique can be defined as finding the shortest distance between the solution of the candidate scheme set and the ideal solution, which is calculated as the decision value determined by the objective function. First, the effect sample matrix T = [t ij ] n×m is formed, and the defined factor Z j is calculated by the effect value determined for each index.
where t ij (i = 1, 2, · · · , n; j = 1, 2, · · · , m) is the effect value of decision scheme S i with respect to objective j. f ij is the evaluation index of scheme S i with respect to objective 'j'. ω i (where i = 1, 2) is calculated by the entropy method [26].
Then, the decision matrix R = [r ij ] n×m is formed according to decision values (11), (12), and (13) with the benefit objective function, cost objective function, and interval objective function.
where r ij represents the decision value when the objective function is defined as the benefit type index. (12) where r ij represents the decision value when the objective function is defined as the cost type index.
where r ij ∈ [−1, 1] represents the decision value when the objective function is defined as the interval [a,b] index. However, we define the decision vector to be r 0 = r 0 1 , r 0 2 , · · · , r 0 m , that is, the centre of the grey target, corresponding to the ideal solution S 0 , where r 0 According to the grey target decision-making definition, the closer the target centre moment of solution S i is to ideal solution S 0 , the better the solution S i is, and the distance between solution S i and ideal solution S 0 is expressed as where γ j , (j = 1, 2, · · · , m) is the weight of each decision-making objective obtained by combining the entropy method with the analytic hierarchy process [32].

V. THE BASIC CONCEPT OF BEETLE ANTENNAE SEARCH (BAS) A. BAS
The Beetle Antennae Search is a bioinspired intelligent optimization algorithm inspired by the foraging principle of beetles, which is based on the smell of food using two antennae. If the left antennae receives a stronger odour than the right antenna, then the next step is to fly to the left and not the right. The beetle can determine the position of the food effectively according to this principle. The smell is regarded as a function, and the optical value of the function can be found by collecting the odour value of two points near the beetle.
The beetle produces a random vector method to determine the direction of food, and the random direction is generated as follows: where rand(n,1) represents a random function and n represents the dimension of space. Then, according to the direction VOLUME 8, 2020 vector, the left and right antennae of the beetle are defined as where x t l and x t r respectively represent the location of the search area on the left and right of the beetle at time t, x t represents the position of the beetle at time t, and d 0 is the length between the left and right antennae at time t. The odour intensity of the left and right antennae can be defined as f (x t l ) and f (x t r ) function, and then we can calculate the position of the beetle at the next moment as follows: where f (·) is the fitness function, δ t represents the step size at time t, and sign(·) is a sign function. To search the optimal solution, the step size decrease with the number of iterations increases as follows: where λ is the attenuation coefficient of the step length δ. The length d between the right and left antennae, for practical calculation, is variable with step size, and this can be generated by the following formula where c is a constant, generally from 2 to 10.

B. CHAOTIC DISTURBANCE BAS
In the BAS algorithm, because a beetle population is difficult to maintain in the optimization process, the antennae must be an individual search, and it often converges to local optima. To address this shortcoming, a BAS with chaotic disturbance (CDBAS) that combines BAS with a chaotic local search is proposed in this study. First, a beetle that needs chaotic disturbance is found according to the similarity, which is calculated by the similar function (20). Chaotic sequences are generated by the logistic map in this paper (22), which is described as follows: where d(i,j) is the Euclidean distance from individual i to j, i is the beetle with the best fitness, d min is the Euclidean distance of the nearest individual to i, and d max is the Euclidean distance of the beetle farthest from i. The similarity is described as follows: Logistic map is described as follows: where X k is the k-dimensional variable of chaotic sequence X , that is, X k ∈ [0, 1], µ ∈ (3.569, 4), and N is the number of populations. The chaos initial population is scaled into [0,1], and the chaotic sequence G = {X 1 , . . . . . . , X S } can be obtained in accordance with iteration (23), where b max and b min are the upper and lower boundary values of k-dimensional variable newX k , respectively.

C. MULTIOBJECTIVE CDBAS BASED ON GREY TARGET DECISION-MAKING METHOD
With the aim of adopting the CDBAS algorithm for solving multiobjective optimization problems, a grey target decision-making theory is applied to achieve the power of the beetles, that is for sorting the beetles.
Here, the proposed method randomly generates the initial population of 'pbest' beetles in the specified search range. The corresponding indexes of these initial beetles are calculated and then sorted by using the grey target decision-making technique. Furthermore, each beetle generates the left and right antennae spatial position coordinates depends upon its decision value. The chaotic disturbance mechanism is carried out for better beetles depending on the best beetle and each beetle generates a new position according to the left and right antennae spatial position coordinates. The population is again ranked, and this computation continues until the specified target condition is reached. Moreover, Table 1 and Fig. 3 show the procedure process and the flowchart of the CDBAS method.

VI. SIMPLIFICATION AND ENCODING OF TOPOLOGY BASED ON BASIC LOOP A. OPTIMAL NETWORK RECONFIGURATION
In general, the optimal network reconfiguration is achieved by changing the status of the switches including normally closed switches and normally open switches, while satisfying the constraint in any distributed system. Only tie-line switches are allowed to be open, and the other switches are closed to generate a solution vector to reduce the computational speed of the system. With the aim of generating fewer infeasible solutions, all switches are initially closed, and then only one switch is selected to be open for each loop under the condition of satisfying the constraints, which are the radiation and connectivity of the network structure. In addition, the number of open switches (i.e., tie switches) is defined as the size of the problem decision variable. For example, the modified IEEE 33-bus radial network which is represented in Fig. 4 below is considered to evaluate the performance of the proposed methodology. In this network, the number of normally open switches is 5, and the CDBAS algorithm generates a 5-dimensional vector as the size of the problem decision variable based on the condition of satisfying the constraints. However, the indexes are calculated according to the generated feasible solution which is to maintain the radial and connected network structure, and the generated unfeasible solution can be replaced by generating a new solution immediately. To detect the radial network structure, an incidence matrix is generated according to the distribution network topology in Fig. 4.
We adopt the encoding method of particles based on the loop topology by which only one switch can be selected in each loop. Table 2 shows loop vectors formed according to the distribution network topology in Fig. 4. The binary numbers 0 and 1 are used to encode the open and closed states of the contact switch. In (24), (1 or 0) means take '1' or '0', and D ij represents the opening and closing state of the j section switch in loop 'i'. When (1or0) is taken to 1, the contact switch of loop 'i' is closed, and D ij indicates that the segment switch 'j' of loop '1' is disconnected. When (1 or 0) is taken to 0, the segment switch 'j' in D ij takes a random number, in other words, the contact switch in loop 'i' is disconnected, and the segment switch is closed.

B. CDBAS METHODOLOGY FOR OPTIMAL MULTIOBJECTIVE RECONFIGURATION
With the CDBAS methodology used for solving the multiobjective network reconfiguration problem for the distribution system, each beetle is defined as a solution vector consisting of open switches. The randomly generated dimension of the initial population is the number of switches to be opened in the system, and it moves to the optimal reconfiguration solution step by step through the iteration update of the orientation position. The position of the beetle is used to simulate the switching state of each one-dimensional switch. The fitness value in the CDBAS approach for optimal network reconfiguration is the distance d i of the grey target decision-making design for the three indexes in the system including active power loss, load balancing and maximum node voltage deviation, as described in Section III. The solution corresponding to the minimum value of distance d i is obtained in accordance with the ranking of the distance d i . Next, the population is VOLUME 8, 2020 ranked by using grey target decision-making theory, and the population is updated by using the CDBAS method. Furthermore, the generated feasible solution obtained for optimal network reconfiguration by the proposed method must satisfy the radial constraint and be feasible to provide services for all the loads with the level values of voltage and current. If the generated solution violates one of the constraints set in this study, the solution is discarded.

VII. EXPERIMENTAL DESIGN AND RESULTS ANALYSIS A. EXPERIMENTAL DESIGN 1) BENCHMARK FUNCTIONS
To verify the performance of the proposed CDBAS algorithm, 8 benchmark functions are tested in the simulation. Table 3 shows the expressions, dimensions, search ranges and theoretical optimal values of 8 benchmark functions. The functions g1-g3 are typical unimodal functions, which are mainly used to test the accuracy, convergence rate and global search ability of the algorithm. The g4-g8 functions are continuous multimodal functions and have an infinite number of local minimum points centered on the global minimum and the radial ring region. Therefore, these functions are often used to test the global optimization ability, the convergence rate and the local optimal avoidance ability of the algorithm.
Here, to set the parameters of the CDBAS and BAS algorithms, the initial population size of the beetles is 100, the maximum number of iterations is 200, the descending coefficient of longicorn step size is λ = 0.95, the degradation factor of step size to beetle antennae is c = 5, and the initial maximum step size is δ = 1.
The two algorithms run independently 30 times, and the search dimension of each function is 5, 10 and 30. Then, the performance of the algorithm is compared and analysed by comparing the accuracy of the results and the evolution curve.

2) ALGORITHM ACCURACY COMPARISON
The precision and robustness of the algorithm are evaluated by the optimal value, the worst value, the average value and square difference tested benchmark functions with 5, 10 and 30 dimensions, which are shown in Table 4, Table 5 and Table 6, respectively.
Taking the functions g1-g3 in Table 4, Table 5 and Table 6 as an example, the precision of the CDBAS algorithm is higher than that of the BAS algorithm and the robustness is better. From the comparison results for the functions g4-g8, the CDBAS also enhanced the global search ability; it was very easy to avoid the local optimal pole since the temporal perturbation decreased continuously with the increase of the supernumerary number.

VIII. TEST AND ANALYSIS OF CDBAS ALGORITHM FOR DISTRIBUTION NETWORK RECONSTRUCTION
To validate the presented method and compare it with other methods, the optimal network reconfiguration is solved for the following cases: Case-4. Simultaneous minimization of active power loss, load balancing index and maximum node voltage deviation with reconfiguration one time over 24 hours Case-5. Simultaneous minimization of active power loss, load balancing index and maximum node voltage deviation with reconfiguration five times over 24 hours Case-6. Simultaneous minimization of active power loss, load balancing index and maximum node voltage deviation with reconfiguration twenty-four times over 24 hours

A. 33-BUS TEST SYSTEM
The performance and effectiveness of the presented methodology is verified on the modified IEEE 33-Bus system and compared with other methods. Fig. 4 shows single diagram of this network with 4 DG networks. The specific parameters and the locations of DG access are shown in Table 7. Table 8 and Table 9 show the proportion load and distribution in hours for three types of load in each bus and for that considered in the study case on the modified IEEE 33 bus network [33]. Fig. 5 shows the output distribution of DGs in 24 hours.

1) VALIDATION OF CDBAS ALGORITHM
The simulation results of IEEE 33-node systems with DG for Cases 1, 2, 3 and 4 are given in Table 10. From the results shown for Cases 1, 2 and 3 in Table 10, the optimal solution obtained by the proposed algorithm is better than the results obtained in reference [16] and [34]. We can see that the indexes conflict with each other to minimize each index simultaneously to achieve the optimal solution. A comprehensive solution with three objectives considering the numbers of the switching operations index for the optimal operation of the distribution system is studied and obtained in Case 4. In addition, the total running time for optimal reconfiguration proposed algorithm in Case 4 is 90.99 s which has been reduced 132.73 s and 95.23 s in comparison with algorithm used in [16] and [34], respectively. Fig. 6, Fig. 7, and Fig. 8 show the node voltage distribution (standard unitary value), power loss and load balance of the network before and after reconfiguration with DG based on VOLUME 8, 2020   Case 4 using CDBAS, Particle Swarm Optimization (PSO) in [16], Genetic Algorithm (GA) in [34] and BAS for the distribution network. The results show that the three indexes of the proposed methodology are better than those of the    other three methods for distribution network reconfiguration considering variations in load and DG over 24 hours.  . Global convergence curves of the proposed algorithm and algorithms in reference [16] and [34].

2) VERIFICATION OF ALGORITHM SUPERIORITY
The voltage profiles of the 33 bus networks before and after reconfiguration during for 24 hours are illustrated in Fig. 6. It is clear that the minimum voltage amplitude of each algorithm is 0.9450 p.u. at 11 a.m, 0.9650 p.u. at 9 p.m, 0.9635p.u. at 9 p.m, 0.9620 p.u. at 9 a.m, and 0.9631 p.u. at 9 a.m. Fig. 7 shows the power loss profile of the 33 bus network before and after reconfiguration with DG for 24 hours. It is clear that the maximum total power loss of each algorithm for the distribution network over 24 hours is 857.54 kWh, 574.77 kWh, 574.77 kWh, 580.86 kWh, and 581.08 at 12 a.m.
From the results of Fig. 6 voltage and Fig. 7 active power loss, we can see that the voltage and active power loss results of CDBAS algorithm are slightly better to compare the PSO algorithm [16], the GA algorithm [34] and the BAS algorithm. In other words, the CDBAS algorithm is better in global optimization ability. Fig. 8 shows the load balance before and after reconfiguration for 24 hours. The results show that the total load balance value for each algorithm over 24 hours is 2.7733, 2.2551, 2.2572, 2.2600 and 2.2437

3) EFFICIENCY VERIFICATION OF ALGORITHMS
From the results shown for the population evolution of the four algorithms in Fig. 9, the number of iterations and global optimization capability of the CDBAS algorithm are better than those of the other three algorithms. The minimum number of iterations of the CDBAS algorithm is  19 times and 17 times for the IEEE 33-bus distribution system without DG in Fig. 9 (a) and with DG in Fig. 9 (b), respectively.

4) OPTIMAL RECONFIGURATION IN TIME SEGMENTS
The number of network reconfigurations is set to five for one day over 24 hours considering the limit of switching action based on the load demand curve. In Fig. 10 shows the total demand profile over 24 hours. The five network reconfiguration periods in a day are 1-4, 5-12, 13-15, 16-20, and 21-24. The reconfiguration results of Cases 5 and 6 are given in Table 11 to compare the two cases for different switching times during 24 hours. It is clear that the total power loss of 557.59 kWh, the total load balancing index of 2.2520 and the total maximum node voltage deviation of 0.4981 in Case 5 are both better than total power loss of 564.87 kWh, the total load balancing index of 2.2587 and the total maximum node voltage deviation of 0.5207 in Case 6 over 24 hours.

B. 69-BUS TEST SYSTEM
In this section, multi-objective reconfiguration of the distribution network is done, with the objectives of active power loss, load balancing index and Maximum node voltage deviation. The CDBAS algorithm is used for optimization and reconfiguration are implemented on the modified IEEE 69-bus system. In 69 bus standard system there are five loops so it should have five tie switches to remain network radial in [35]. The specific parameters and the locations of DG access are shown in Table 12.
The proportion load and distribution in hours for three types of load in each bus are presented in [35] and for that considered in the study case on the modified IEEE 69 bus network. In Fig. 10 shows the total demand profile over 24 hours. The optimization results of the improved algorithm for Case 4 are given and compared with results of [16] in Table 13.
From Table 13 for Case 4, the best solution given by proposed method is better than the results obtained in reference [16]. The total active power loss is reduced by 39.62%, minimum node voltage is improved from 0.9278 to 0.9536 in Fig. 11 and the number of switching operations is 14.
The reconfiguration results of Cases 5 and 6 are given in Table 14 to compare the two cases for different switching times during 24 hours. It is clear that the total power loss of 781.77 kWh, load balancing of 2.6710 and the total maximum node voltage deviation of 0.7580 in Case 5 are both slightly   better than total power loss of 783.16 kWh load balancing of 2.6784 and the total maximum node voltage deviation of 0.7591 in Case 6 over 24 hours.

C. 118-BUS TEST SYSTEM
The IEEE 118-bus system consists of 118 buses, 11 kV, and radial distribution system [36]. In 118 bus standard system there are fifteen loops so it should have 117 sectional switches and 15 tie switches to remain network radial. The CDBAS algorithm is used for optimization and reconfiguration are implemented on the modified IEEE 118-bus system. In this section, the reconfiguration results of Cases 4, 5 and 6 by the CDBAS algorithm are presented and compared with the results of [16]. The specific parameters and the locations of DG access are shown in Table 15 [37]. In Fig. 12 shows the total demand profile over 24 hours.
From Table 16 for Case 4, the best solution given by proposed method is better than the results obtained in reference [16]. The total active power loss is reduced by 10.99%, minimum node voltage is improved from 0.8847 to 0.9335 in Fig. 13 and the number of switching operations is 104.
The reconfiguration results of Cases 5 and 6 are given in Table 17 and 18 to compare the two cases for different switching times during 24 hours. It is clear that the total power loss of 6454.10 kWh, the load balancing of 1.8900 and the      the voltage at a certain time meets the constraint condition. However, the reconfiguration results of Case 5 can satisfy the voltage constraint in the whole reconstruction period, so that the total active power loss increase slightly, the load balancing decreases slightly, and the voltage deviation increases slightly with the variation of load and DG.

IX. CONCLUSION
Considering multiobjective network reconfiguration based on variations in load and distributed generation (DG) for distribution systems has a strong effect on ensuring the safe and economical operation of a power system. In this study, minimizing the sum of active power loss, the load balancing index and the maximum node voltage deviation are considered as the objectives, and then the optimization problem of the layering relationship between each index and the number of switching operations is solved using grey target decisionmaking technology. Beetle Antennae Search (BAS) has the advantages of efficient optimization and small computation; however, it often converges to local optima. The improved Chaotic Disturbed Beetle Antennae Search (CDBAS) algorithm, which guarantees efficient calculation, is used to optimize multiobjective network reconfiguration in distribution systems.
Here, the proposed method can achieve the optimization of each objective in other words, it is valid and feasible through simulation in the modified IEEE 33, 69 and 118-bus system with variation of load and DG over 24 hours. Finally, compared with other methods in terms of speed and accuracy, the CDBAS method is accurate and converges very rapidly, so it can be used to solve optimization problems in multiple fields, such as the distribution network-planning schemes considering load change, multi-source system schedules, and optimal storage allocations. Additionally, the proposed methodology can be applied to problems in which the objective functions are differentiable, non-differentiable, convex and non-convex with continuous and discrete variables.