Method for Identifying Electricity Theft Users in Low-Voltage Distribution Areas Based on Improved Whale Optimization Algorithm

Abstract

The present invention belongs to the field of power distribution technology and relates to a method for identifying electricity theft users in low-voltage power distribution areas based on an improved whale optimization algorithm, which includes the following steps: calculating the calculated value of the total power consumption of the power distribution area within a time period; establishing an objective function by using the calculated value of the total power consumption of the power distribution area and the measured value of the total power consumption of the power distribution area; solving the objective function through the improved whale optimization algorithm; and analyzing the power loss rates of each user obtained by the solution by using a distance-based outlier detection algorithm, and outputting the users suspected of electricity theft. The present invention improves the traditional whale algorithm by introducing a dynamically changing convergence factor, a random position update coefficient and a directional search behavior, then uses the improved whale optimization algorithm to estimate the power loss rates of each user in the power distribution area, and finally identifies the users with abnormal power loss rates through a distance-based outlier detection algorithm.

Description

Technical Field

[0001] The present invention belongs to the technical field of power distribution, and particularly relates to a method for identifying electricity theft users in low-voltage power distribution areas based on an improved whale optimization algorithm. Background Art

[0002] With the rapid development of social economy, the social demand for electric energy has increased rapidly. Electric energy is not only an important support for the development of the national economy, but also an important guarantee for power companies to maintain their own development. The abnormal electricity consumption of users in the power distribution area not only damages the interests of power companies and seriously affects their healthy development, but also brings great hidden dangers to the safety and electricity consumption of the power grid.

[0003] The traditional manual inspection method has low efficiency, high difficulty, and consumes a large amount of manpower and material resources. Most of the existing electricity theft detection models rely on large datasets, but there are few low-voltage electricity theft work orders in the power grid marketing system, which is difficult to meet the needs of model training.

[0004] CN110824270A discloses a method for identifying electricity theft users by combining substation line loss and abnormal events, including: obtaining substation and user data of at least one substation to be inspected, where the substation and user data includes substation line loss data and basic data of each electricity user in the substation; applying a substation line loss abnormal detection method and the substation line loss data of the at least one substation to be inspected to determine an abnormal line loss substation with electricity theft suspect users during a specified power consumption period; for any abnormal line loss substation with electricity theft suspect users, determining a set of electricity theft suspect users by K-means clustering, a set of electricity theft suspect users by support vector machine, and a set of electricity theft suspect users by Bayesian algorithm; and comprehensively evaluating to determine a list of electricity theft suspect users in the abnormal line loss substation.

[0005] CN109947815A discloses an electricity theft identification method based on an outlier algorithm, including the steps of: obtaining daily electricity consumption data of users, preprocessing the data, calculating the sample volatility CV, determining the centroid, as well as parameters p and D, performing electricity theft discrimination by the outlier algorithm, determining electricity theft sample points, and setting an electricity theft alarm. This patent combines the electricity consumption volatility and an improved distance-based outlier mining algorithm to complete the identification of user electricity theft.

[0006] The Whale Optimization Algorithm is an algorithm proposed based on the behavior of whales hunting prey. Whales are social mammals that cooperate with each other to drive and surround prey during hunting. The Whale Optimization Algorithm is a newly emerging optimization algorithm with few research and application cases. In the Whale Optimization Algorithm, the position of each whale represents a feasible solution. During the hunting process of the whale group, each whale has two behaviors. One is to surround the prey, and all whales move towards other whales. The other is to blow bubbles in a net-like pattern, and the whales swim in a circle and blow bubbles to drive the prey. During each generation of swimming, the whales will randomly choose these two behaviors to hunt. During the behavior of whales surrounding the prey, the whales will randomly choose whether to swim towards the whale at the optimal position or randomly select a whale as their target and approach it.

[0007] There is little research on the application of the Whale Optimization Algorithm in the power field. CN113281620A discloses a fault section location method, system and medium based on an adaptive Whale Optimization Algorithm for fault section location. CN110110930A discloses a short-term power load forecasting method for a recurrent neural network based on an improved Whale Optimization Algorithm for power load forecasting. There is no application report on the Whale Optimization Algorithm in anti-stealing electricity in the prior art. Summary of the Invention

[0008] The purpose of the present invention is to solve the problem of identifying electricity-stealing users in low-voltage power distribution areas, and to provide a method for identifying electricity-stealing users in low-voltage power distribution areas based on an improved Whale Optimization Algorithm. The core idea is to use the improved Whale Optimization Algorithm to estimate the power loss rate of each user, and then use the distance-based outlier detection algorithm to screen users with abnormal power loss rates; by analyzing the power loss rate of each user, the efficient identification of electricity-stealing users in low-voltage power distribution areas is realized.

[0009] The technical solution adopted by the present invention is: a method for identifying electricity-stealing users in low-voltage power distribution areas based on an improved Whale Optimization Algorithm, including the following steps:

[0010] Step 1, calculate the calculated value of the total power consumption of the power distribution area within a time period;

[0011] Step 2, establish an objective function using the calculated value of the total power consumption of the power distribution area and the measured value of the total power consumption of the power distribution area;

[0012] Step 3, solve the objective function through the improved Whale Optimization Algorithm;

[0013] Step 4, analyze the power loss rate of each user obtained in Step 3 using the distance-based outlier detection algorithm, and output users suspected of stealing electricity.

[0014] Further preferably, in the step 1, the user power consumption is corrected by using the user power loss rate, and the corrected power consumptions of all users in the substation area are added to obtain the calculated value of the total power consumption in the substation area:

[0015] M0 = M1θ1 + M2θ2 + … + M n θ n (1)

[0016] In the formula, θ1, θ2, …, θ n are the power loss rates of the 1st, 2nd, …, nth users respectively, and M1, M2, …, M n are the measured values of the electricity meters of the 1st, 2nd, …, nth users respectively;

[0017] Substituting the measured values of the electricity meters of each user in 95 time periods of a day into formula (1) respectively, the calculated values of the total power consumption in the substation area in 95 time periods can be obtained

[0018]

[0019] In the formula, respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users in the 1st time period; respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users in the 2nd time period; respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users in the 95th time period.

[0020] Further preferably, in the step 2, by using the calculated values of the total power consumption in the substation area in 95 time periods and the measured values of the total power consumption in the substation area in 95 time periods a target function is established:

[0021]

[0022] In the formula: F is the error between the sequence of the calculated values of the total power in the substation area in 95 time periods after correcting the user power consumption and the sequence of the measured values of the total power in the substation area in 95 time periods; is the sequence of the calculated values of the total power consumption in the substation area in 95 time periods after correcting the user power consumption, is the sequence of the measured values of the total power consumption in the substation area in 95 time periods.

[0023] Further preferably, in the step 3, the process of solving the target function by improving the whale optimization algorithm is as follows:

[0024] Parameter A of the basic whale optimization algorithm:

[0025] A = 2a * r1 - a (4)

[0026] r1 is a random number in [0, 1], and a is a dynamically changing convergence factor;

[0027]

[0028] t is the current iteration number, and t max is the maximum iteration number;

[0029] To make full use of the position of the prey, that is, the optimal solution θ * (t), a random position update coefficient m is proposed as follows:

[0030] θ(t + 1) = m·θ * (t) - A·K |A| < 1, p < 0.5 (7)

[0031] θ(t + 1) = m·θ rand - A·K |A| ≥ 1, p < 0.5 (8)

[0032] θ(t + 1) = K’·e bl ·cos(2πl) + (1 - m)·θ * (t) p ≥ 0.5 (9)

[0033] K = |C·θ rand - θ(t)| (10)

[0034] K’ = |θ rand - θ(t)| (11)

[0035] C = 2·r2 (12)

[0036] In the formula, θ(t) is the position of the whale individual in the current iteration, θ(t + 1) is the new position of the whale individual in the next iteration, θ rand is the position vector representing the randomly selected whale, m is a random number between, p and r2 are both random numbers in [0, 1], b is a constant used to define the shape of the logarithmic spiral, here it takes 1, l is a random number in [0, 1], both b and l jointly control the spiral position update method of the whale individual, K is the moving step size, and C is a random number in [0, 2];

[0037] In addition, to expand the search space of the algorithm and further improve the optimization ability of the algorithm, the present invention also proposes a directional search behavior as follows:

[0038]

[0039]

[0040] It is the position after the movement behavior of the three worst whale individuals in the current iteration towards the optimal individual, θ 1 , θ 2 , θ 3 are the positions of the three worst whale individuals in the current iteration, θ * is the optimal whale individual in the current iteration, and step is the movement step size. After each iteration, the three worst whale individuals in the current iteration move one step towards the optimal individual, thus expanding the search range, improving the global search ability of the algorithm in the early stage of the algorithm, improving the optimization accuracy of the algorithm in the later stage, and reducing the possibility of the algorithm falling into local optimum.

[0041] First, initialize the positions of the whale individuals in the whale optimization algorithm; initialize the position state of the whale individuals as the vector θ = (θ1, θ2,..., θ n ), θ1, θ2,... θ n are respectively a set of possible user power loss rates for the corresponding users. The current fitness of the whale individual is f = f(θ), corresponding to the error in the objective function. Set the initial parameters of the whale optimization algorithm, the population size N and the maximum number of iterations t max ; randomly generate each whale individual, and each whale represents a set of possible user power loss rates;

[0042] Calculate the optimal individual and record it on the bulletin board; select the power data acquisition sequences of the main meter and n power users in a substation area for 95 time periods in a day, calculate the fitness according to the objective function, compare the fitness corresponding to each individual, and select the optimal individual. The optimal individual has the smallest error value, and record its current position and fitness on the bulletin board;

[0043] After each individual respectively performs random simulation to surround the prey, bubble net attack, search and prey, and directional search, check the fitness of the current position of the individual and compare it with the value recorded on the bulletin board. If it is better than the value recorded on the bulletin board, update the bulletin board. Then judge whether the maximum number of iterations is reached. If it is reached, output the result and obtain a set of user power loss rates θ1, θ2,... θ n .

[0044] For further optimization, in step 4, assume that the number of neighbors of a sample point is greater than the set threshold, then the sample point is a normal point. If the number of neighbors is less than the set threshold, then the sample point is an abnormal point; the method to find the number of neighbors of any sample point is to calculate its Euclidean distance D from all other points. Compare the Euclidean distance D with the distance threshold r. If D < r, it is a neighbor of the sample. If D > r, it is not a neighbor of the sample; determine whether the analyzed sample point is an abnormal point.

[0045] Advantages of the present invention: The power loss rate of electricity-stealing users must be abnormal. This method first improves the traditional whale algorithm by introducing a dynamically changing convergence factor and a random position update coefficient, then uses the improved whale optimization algorithm to estimate the power loss rate of each user in the low-voltage power distribution area, and finally identifies the users with abnormal power loss rate through a distance-based outlier detection algorithm. Brief Description of the Drawings

[0046] Figure 1 It is a flowchart of the present invention.

[0047] Figure 2 Are the power loss rates of each user.

[0048] Figure 3 Are the identification results of electricity-stealing users. Detailed Embodiment

[0049] The present invention will be further elaborated in detail below with reference to the drawings.

[0050] Refer to Figure 1 , a method for identifying electricity-stealing users in a low-voltage power distribution area based on an improved whale optimization algorithm, includes the following steps:

[0051] Step 1, calculate the calculated value of the total power consumption of the low-voltage power distribution area within a time period.

[0052] Use the power loss rate of the user to correct the power consumption of the user, and add up the corrected power consumption of all users in the low-voltage power distribution area to obtain the calculated value of the total power consumption of the low-voltage power distribution area:

[0053] M0 = M1θ1 + M2θ2 + … + M n θ n (1)

[0054] In the formula, θ1, θ2, …, θ n Are the power loss rates of the 1st, 2nd, …, nth users respectively, and M1, M2, …, M n Are the measured values of the electricity meters of the 1st, 2nd, …, nth users respectively;

[0055] Substitute the measured values of the electricity meters of each user within 95 time periods of a day into formula (1) respectively, and the calculated value of the total power consumption of the low-voltage power distribution area within 95 time periods can be obtained

[0056]

[0057] In the formula, Respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 1st time period; Respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 2nd time period; respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 95th period.

[0058] Step 2: Establish an objective function by using the calculated value of the total electricity consumption of the transformer substation area and the measured value of the total electricity consumption of the transformer substation area.

[0059] Using the calculated values of the total electricity consumption of the transformer substation area within 95 periods and the measured values of the total electricity consumption of the transformer substation area within 95 periods Establish an objective function, considering both the overall error and the error deviation in each dimension.

[0060]

[0061] In the formula: F is the error between the sequence of calculated values of the total electricity energy of the transformer substation area in 95 periods of a day after correcting the electricity consumption of users and the sequence of measured values of the total electricity energy of the transformer substation area in 95 periods of a day. is the sequence of calculated values of the total electricity consumption of the transformer substation area in 95 periods of a day after correcting the electricity consumption of users, is the sequence of measured values of the total electricity consumption of the transformer substation area in 95 periods of a day.

[0062] Step 3: Solve the objective function by improving the whale optimization algorithm.

[0063] Parameter A of the basic whale optimization algorithm:

[0064] A = 2a * r1 - a (4)

[0065] a = 2 - 2t / t max (5)

[0066] r1 is a random number in [0, 1], a is the convergence factor, and the change of the convergence factor a determines the value of parameter A. When the convergence factor is larger, it is beneficial to the global search of the basic whale optimization algorithm and reduces the possibility of falling into the local optimum. When the convergence factor is smaller, it is beneficial to the local search of the basic whale optimization algorithm and speeds up the convergence rate of the basic whale optimization algorithm. In the basic whale optimization algorithm, the convergence factor decreases linearly with the increase of the iteration times, which makes the convergence rate of the basic whale optimization algorithm slow. Therefore, the present invention proposes a dynamically changing convergence factor, which improves the convergence rate of the whale optimization algorithm and does not affect the global search and local search performance of the algorithm. The specific formula is as follows:

[0067]

[0068] t is the current iteration number, t max is the maximum iteration number. In the early stage of iteration, a larger parameter A is generated by the convergence factor, which is beneficial to the global search of the whale optimization algorithm and improves the convergence rate of the algorithm; in the later stage of iteration, a smaller parameter A is generated by the convergence factor, which is beneficial to the local search of the whale optimization algorithm.

[0069] In addition, in order to make full use of the position of the prey, that is, the optimal solution θ * (t), to improve the optimization accuracy of the whale optimization algorithm, the present invention proposes a random position update coefficient m, as shown in the following formula:

[0070] θ(t + 1) = m·θ * (t) - A·K |A| < 1, p < 0.5 (7)

[0071] θ(t + 1) = m·θ rand - A·K |A| ≥ 1, p < 0.5 (8)

[0072] θ(t + 1) = K’·e bl ·cos(2πl) + (1 - m)·θ * (t) p ≥ 0.5 (9)

[0073] K = |C·θ rand - θ(t)| (10)

[0074] K’ = |θ rand - θ(t)| (11)

[0075] C = 2·r2 (12)

[0076] In the formula, θ(t) is the position of the whale individual in the current iteration, θ(t + 1) is the new position of the whale individual in the next iteration, θ rand is the position vector representing the randomly selected whale, m is a random number between, p and r2 are both random numbers between [0, 1], b is a constant used to define the shape of the logarithmic spiral, here it is taken as 1, l is a random number between [0, 1], both b and l jointly control the spiral position update method of the whale individual, K is the moving step size, and C is a random number between [0, 2].

[0077] m is the position update coefficient of the optimal solution. When p < 0.5, the whale optimization algorithm is in the stage of surrounding the prey and random search. At this time, the position update coefficient increases with the increase of the number of iterations, so that the optimal solution can fully play its role; when p ≥ 0.5, the whale optimization algorithm is in the stage of spiral position update. As the whale optimization algorithm iterates continuously, the whale gets closer and closer to the prey. A smaller update coefficient 1 - m is used to change the position of the prey, effectively improving the local search ability of the whale optimization algorithm, and thus improving the optimization accuracy.

[0078] In addition, in order to expand the search space of the algorithm and further improve the optimization ability of the algorithm, the present invention also proposes a directional search behavior, as shown in the following formula:

[0079] θ inext = θ i + step i = 1, 2, 3 (13)

[0080]

[0081] is the position after the movement behavior of the three worst whale individuals towards the optimal individual in the current iteration, θ 1 , θ 2 , θ 3 are the positions of the three worst whale individuals in the current iteration, θ* is the optimal whale individual in the current iteration, and step is the movement step size. After each iteration, the three worst whale individuals move one step towards the optimal individual, thereby expanding the search range, improving the global search ability of the algorithm in the early stage of the algorithm, improving the optimization accuracy of the algorithm in the later stage, and reducing the possibility of the algorithm falling into local optimum.

[0082] First, initialize the positions of the whale individuals in the whale optimization algorithm. Initialize the position state of the whale individuals as the vector θ = (θ1, θ2, …, θ n ), θ1, θ2, … θ n are respectively a set of possible user power loss rates for the corresponding users. The current fitness of the whale individual is f = f(θ), corresponding to the error in the objective function. Set the initial parameters of the whale optimization algorithm: the population size N and the maximum number of iterations t max ; randomly generate each whale individual, and each whale represents a set of possible user power loss rates.

[0083] Calculate the optimal individual and record it on the bulletin board; select the power data acquisition sequences of the master meter and n power users in a substation area for 95 time periods in a day, calculate the fitness according to the objective function, compare the fitness corresponding to each individual, and select the optimal individual. The optimal individual has the smallest error value, and record its current position and fitness on the bulletin board.

[0084] When the whale individuals respectively perform random simulation of surrounding prey, bubble net attack, search and predation, and directional search behaviors, check the fitness of the current position of the individual and compare it with the value recorded on the bulletin board. If it is better than the value recorded on the bulletin board, update the bulletin board. Then judge whether the maximum number of iterations is reached. If so, output the result and obtain a set of user power loss rates θ1, θ2, …, θ n .

[0085] Step 4, use the distance-based outlier detection algorithm to analyze the user power loss rates obtained in Step 3, and output the users suspected of electricity theft.

[0086] Assume that the number of neighbors of a sample point is greater than the set threshold, then the sample point is a normal point; if the number of neighbors is less than the set threshold, then the sample point is an abnormal point. The method for finding the number of neighbors of any sample point is to calculate its Euclidean distance D from all other points. The Euclidean distance D is compared with the distance threshold r. If D < r, it is a neighbor of the sample; if D > r, it is not a neighbor of the sample. Finally, determine whether the analyzed sample point is an abnormal point according to the above assumption. Since the object of research and analysis is the user's power loss rate, the one-dimensional Euclidean distance is used here. For two one-dimensional sample data θ1 and θ2, the Euclidean distance formula is defined as:

[0087] D(θ1, θ2) = |θ1 - θ2| (15)

[0088] Using one-dimensional samples greatly reduces the complexity of data processing and improves the efficiency of algorithm execution. Therefore, compared with other electricity theft identification algorithms, this method has certain advantages.

[0089] There are 20 households in XX Community of a certain province. Obtain the electricity data of 96 points of the total meter and each user's sub-meter in the community on December 20, 2020, with the unit of kilowatt-hour. After preprocessing, the electricity consumption in 95 time periods is obtained. Part of it is shown in Table 1.

[0090] Table 1 Partial electricity data of the community

[0091]

[0092] Substitute the electricity data of 95 time periods into Equation (2) to calculate the calculated value of the total electricity of the transformer substation Then use the calculated value of the total electricity of the low-voltage transformer substation and the actual value to establish the objective function

[0093]

[0094] Then use the whale optimization algorithm to solve for the power loss rate θ of each user. The solution results are as Figure 2 shown. It can be seen from Figure 2 that the power loss rates of some users are relatively large. Then use the distance-based outlier detection algorithm to identify the outliers. The distance threshold r = 0.2 and the score threshold is 0.3. The final output results are as Figure 3 shown. Mark the outliers as 1 and the normal users as 0. It can be seen from the algorithm output results that the user numbered 16 is suspected of being an electricity thief.

[0095] The above description only represents the preferred embodiments of the present invention and does not limit the present invention in other forms. Any person skilled in the relevant art may make changes or modifications to equivalent embodiments with equivalent changes by using the content disclosed above. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for identifying electricity theft users in low-voltage power distribution areas based on an improved whale optimization algorithm, characterized in that, It includes the following steps: Step 1, calculate the calculated value of the total power consumption of the transformer area within a time period; Step 2, establish an objective function by using the calculated value of the total power consumption of the transformer area and the measured value of the total power consumption of the transformer area; Step 3, solve the objective function by improving the whale optimization algorithm; Parameter A of the basic whale optimization algorithm: A = 2a * r1 - a (4) r1 is a random number in [0, 1], and a is a dynamically changing convergence factor; t is the current iteration number, t max is the maximum number of iterations; In order to make full use of the position of the prey, i.e., the optimal solution θ * (t), a random position update coefficient m is proposed as follows: θ(t + 1) = m·θ * (t) - A·K |A| < 1, p < 0.5 (7) θ(t + 1) = m·θ rand -A·K |A|≥1, p<0.5 (8) θ(t + 1) = K′·e bl ·cos(2πl) + (1 - m)·θ * (t) p ≥ 0.5 (9) K = |C·θ rand - θ(t)| (10) K′ = |θ rand - θ(t)| (11) C＝2·r2 (12) where, θ(t) is the position of the whale individual in the current iteration, θ(t + 1) is the new position of the whale individual in the next iteration, θ rand is the position vector randomly selected to represent the whale, m is a random number between, p and r2 are both random numbers between [0, 1], b is a constant, l is a random number in [0, 1], both b and l jointly control the spiral position update method of the whale individual, K is the moving step size, C is a random number between [0, 2]; A directional search behavior is proposed, as shown in the following formula: θ i next = θ i + step i = 1, 2, 3 (13) is the position after the movement behavior of the three worst whale individuals in the current iteration towards the optimal individual, θ 1 , θ 2 , θ 3 are the positions of the three worst whale individuals in the current iteration, θ * is the optimal whale individual in the current iteration, and step is the movement step length; after each iteration, the three worst whale individuals in the current iteration move one step towards the optimal individual, thus expanding the search range, improving the global search ability of the algorithm in the early stage of the algorithm, improving the optimization accuracy of the algorithm in the later stage, and reducing the possibility of the algorithm falling into local optimum; each whale represents a group of possible user power loss rates, and the finally output is a group of user power loss rates θ1, θ2, …, θ n ; Step 4, use the distance-based outlier detection algorithm to analyze the power loss rate of each user obtained in Step 3, and output the users suspected of stealing electricity.

2. The method for identifying electricity theft users in a low-voltage power distribution area based on an improved whale optimization algorithm according to claim 1, wherein, In the said Step 1, the electricity consumption of users is corrected by using the power loss rate of users, and the corrected electricity consumption of all users in the transformer area is added up to obtain the calculated value of the total power consumption of the transformer area: M0 = M1θ1 + M2θ2 + … + M n θ n (1) where θ1, θ2, …, θ n are the power loss rates of the 1st, 2nd, …, nth users respectively, and M1, M2, …, M n are the measured values of the electricity meters of the 1st, 2nd, …, nth users respectively; Substitute the measured values of each user's electricity meter in 95 time periods of a day into Equation (1) respectively, and the calculated values of the total electricity consumption in the region for 95 time periods can be obtained. In the formula, respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 1st period of time; respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 2nd period of time; respectively represent the measured values of the electricity meters of the 1st, 2nd, …, nth users within the 95th period of time.

3. The method for identifying electricity theft users in low-voltage power distribution areas based on an improved whale optimization algorithm according to claim 2, wherein In the said step 2, the calculated value of the total power consumption of the power distribution area within 95 time periods and the measured value of the total power consumption of the power distribution area within 95 time periods are used to establish an objective function: Where: F is the error between the sequence of calculated values of the total power energy of the transformer substation area for 95 time periods in a day after correcting the user's power consumption and the sequence of measured values of the total power energy of the transformer substation area for 95 time periods in a day; is the sequence of calculated values of the total power consumption of the transformer substation area for 95 time periods in a day after correcting the user's power consumption, is the sequence of measured values of the total power consumption of the transformer substation area for 95 time periods in a day.

4. A method for identifying electricity theft users in low-voltage power distribution areas based on an improved whale optimization algorithm according to claim 3, characterized in that In step 3, first, initialize the positions of the whale individuals in the whale optimization algorithm; initialize the position state of the whale individuals as the vector θ = (θ1, θ2, …, θ n ), θ1, θ2, … θ n are respectively a set of possible user power loss rates for the corresponding users. The current fitness of the whale individual is f = f(θ), corresponding to the error in the objective function. Set the initial parameters of the whale optimization algorithm: the population size N and the maximum number of iterations t max ; randomly generate each whale individual; Calculate the optimal individual and record it on the bulletin board; select the power data acquisition sequence of the main meter and n household power users in a transformer area at 95 time periods in a day, calculate the fitness according to the objective function, compare the fitness corresponding to each individual, and select the optimal individual. The optimal individual has the smallest error value, and record its current position and fitness on the bulletin board; After each individual respectively performs random simulation to surround the prey, foam net attack, search for prey, and directional search, the fitness of the individual's current position is examined and compared with the value recorded on the bulletin board. If it is better than the value recorded on the bulletin board, the bulletin board is updated; then it is judged whether the maximum number of iterations is reached. If so, the result is output to obtain a set of user power loss rates θ1, θ2, …, θ that meet the requirements. n .

5. The method for identifying electricity theft users in a low-voltage power distribution area based on an improved whale optimization algorithm according to claim 4, characterized in that, In the said Step 4, assume that the number of neighbors of a sample point is greater than the set threshold, then this sample point is a normal point. If the number of neighbors is less than the set threshold, then this sample point is an outlier; the method to find the number of neighbors of any sample point is to calculate its Euclidean distance D from all other points. Compare the Euclidean distance D with the distance threshold r. If D < r, it is a neighbor of this sample. If D > r, it is not a neighbor of this sample; determine whether the analyzed sample point is an outlier.

Images