Accurate risk identification method and system based on multivariate load situation awareness

Abstract

The invention discloses a risk accurate identification method and system based on multivariate load situation awareness, and the method comprises the steps: collecting data through a WSN node, constructing a load risk association model, and carrying out the quantitative analysis of a load risk in a power system; a novel load identification model is constructed by using an SSA algorithm based on a risk result after quantitative analysis, the load type is identified, and load management in a complex environment is adapted; designing a load management peak shifting plan optimization model by using the identified load type, and minimizing the influence of the peak shifting plan on the user in the load management; and constructing a high-energy-consumption industry load situation sensing model by using a VMD technology to carry out load data noise reduction processing, and proposing an evaluation index of a noise reduction effect. Accurate identification and optimal management of the load risk of the power system are realized, and the accuracy, the adaptability and the real-time performance are remarkably improved. The peak shifting plan optimization model reduces the user influence and improves the operation efficiency. Through application of the VMD technology in load situation awareness, data processing precision and management efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system load management, and specifically to a risk accurate identification method and system based on multi-load situation perception. Background Art

[0002] The development of power system load management and optimization technology can be traced back to the early application stage of power systems. In traditional power systems, load management mainly relies on fixed dispatching strategies and simple optimization algorithms. However, with the expansion of power system scale and the increase of complexity, these traditional methods have gradually exposed their limitations. In order to cope with the increasingly complex power load and the ever-changing power market environment, scientists and engineers continue to explore and develop more advanced load management technologies. For example, optimization methods based on linear programming and dynamic programming were widely used in the mid-20th century. After entering the 21st century, with the rapid development of computer technology and artificial intelligence technology, many load management methods based on intelligent algorithms have sprung up, including genetic algorithms, particle swarm optimization algorithms, and ant colony algorithms. In addition, the introduction of Internet of Things technology and big data analysis technology has made real-time load monitoring and intelligent dispatching possible, providing a new way for efficient operation and optimization management of power systems.

[0003] Deficiencies of Existing Technologies Although the existing load management technologies have made significant progress, there are still many deficiencies that need to be addressed. First, the existing technologies mostly use fixed optimization algorithms, lack adaptive capabilities, and cannot adjust the optimization strategies according to different problem characteristics, resulting in poor optimization effects in complex and dynamic power system environments. Second, due to their reliance on specific data and models, the application scope of existing technologies is relatively narrow, and it is difficult to adapt to diversified and changing load demands. For example, some optimization methods can only work in specific types of power systems, limiting the possibility of their wide application. In addition, existing algorithms usually require a lot of calculation and processing time, which is difficult to meet the requirements of real-time load management, affecting the dispatching efficiency and response speed of the power system. Low intelligence is also a prominent problem. Existing technologies mostly rely on preset rules and models, lack self-learning and optimization capabilities, and cannot achieve true intelligent management. In terms of load risk identification, the existing technologies have limited accurate identification capabilities, making it difficult to accurately identify and handle abnormal load states, and cannot achieve accurate risk identification and management. In addition, the adaptive ability and scalability of traditional load management methods are poor, and they cannot effectively respond to the rapid changes and expansion needs of the power system, such as the widespread access to new energy and the rapid growth of loads. In general, existing load management technologies have significant deficiencies in optimization efficiency, application scope, real-time performance, intelligence, accurate identification, adaptability and scalability, and new technical methods are needed to solve these problems. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: in the power system, load risk identification is the key to ensure the safe and stable operation of the system. The present invention realizes the accurate identification of overload risk, low voltage risk and load loss risk by constructing a load risk association model. Comprehensive and accurate perception of the load of the power system is the basis for realizing intelligent management. The present invention realizes real-time monitoring and management of the load status of the power system through multi-element load situation perception technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a risk accurate identification method based on multi-load situation awareness, comprising:

[0007] Collect data through WSN nodes, build load risk association model, and quantitatively analyze load risks in the power system;

[0008] Based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify load types and adapt to load management in complex environments;

[0009] Using the identified load types, we design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users.

[0010] VMD technology is used to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction, and evaluation indicators for noise reduction effects are proposed.

[0011] As a preferred solution of the risk accurate identification method based on multi-load situation awareness described in the present invention, wherein: collecting data through WSN nodes includes: the WSN node coverage model describes that a number of WSN nodes are deployed in a two-dimensional plane area S = L × W, there are a number of obstacles in the area, and the entire area to be monitored is discretized into L × W pixels;

[0012] Assume that the WSN node set N = {n1, n2, ..., n k , k represents the number of WSN nodes, and the obstacle set O = {o 1，i o 2，i ..., o j，i}, j represents the number of obstacles, i represents the type of obstacles; let the center of the power system in the two-dimensional plane be the origin of the coordinate system, let (x k ,y k ) is the WSN node n kThe coordinates of the point to be detected are (x g ,y g ), wireless sensor node n k The probability of sensing the g-point is expressed as:

[0013]

[0014] 0 <r e <r

[0015] α=d(n k , g)-(rr e )

[0016] Among them, d(n k , g) represents WSN node n k The Euclidean distance from the target point g, r e It represents the uncertainty measure of the perception ability of WSN nodes in an obstacle environment, and λ and β represent the perception factors;

[0017] Calculate WSN node n k The Euclidean distance d(n k , g) The formula is expressed as:

[0018]

[0019] Among them, (x g ,y g ) represents the coordinates of the target point g;

[0020] The joint perception probability of n WSN nodes to the same target point is obtained by superposition calculation. Node coverage calculation: Calculate the coverage of the entire area;

[0021] Since a target point g can be covered by multiple WSN nodes at the same time, its joint perception probability P(g) is defined as:

[0022]

[0023] Where p(N, g) represents a single WSN node n k The perceived probability of point g;

[0024] The coverage formula of the entire area is expressed as:

[0025]

[0026] Wherein, L represents the length of the area to be monitored, and W represents the width of the area to be monitored;

[0027] In complex environments, there are usually obstacles in the monitored area. If the WSN node deployment location falls into the obstacle area, the node needs to be moved and redeployed. In order to avoid the clustering of WSN node deployment caused by selecting inappropriate constraints, which leads to a decrease in regional coverage, a virtual repulsive field strategy based on artificial potential field is designed to redeploy the WSN node position.

[0028] Establish an obstacle model for WSN node deployment. The actual obstacle is contained in a circle, and the size of the obstacle determines the radius r of the circle. o The size of the circle is also the area where the repulsive field exists. The distance between the WSN node and the obstacle is simplified to the distance from the center of the circle o i If the node falls into the circle, it will be affected by the repulsive force of the obstacle and move along the repulsive direction by a distance corresponding to the repulsive force, so that the WSN node can avoid the obstacle.

[0029] The formula for the repulsive force from obstacles is:

[0030]

[0031] Among them, F no Represents the impact of obstacles on WSN node n i The repulsive force, r o The radius of the circle d(n i , o iceriter ) represents WSN node n i and obstacle center o iceriter The distance between i represents the position vector of the i-th WSN node, o iceriter Represents the center position vector of the obstacle;

[0032] When there is a deployment area boundary around the obstacle, in order to prevent the WSN node from moving beyond the boundary, the node moving toward the boundary will move in the opposite direction of the repulsive force. The magnitude of the repulsive force is defined as:

[0033] F′ no =2r o -F no

[0034] The improved SSA algorithm enables nodes to avoid obstacles and cover more areas, improving the coverage and efficiency of load management.

[0035] As a preferred solution of the risk accurate identification method based on multi-element load situation perception described in the present invention, wherein: constructing a load risk association model, quantitatively analyzing the load risk in the power system, including, since the overload risk, low voltage risk and load loss risk are closely related, performing quantitative correlation analysis, applying the importance ranking and grading of power system branches in the analysis process, and adopting an iterative self-organizing data analysis algorithm;

[0036] Overload risk is the combination of the possibility and severity of the active power of other non-fault branches exceeding their rated values ​​after a power system accident. The formula is:

[0037] Risk(P L |E i )=P(E i )∫f(P L |E i , L)×Sev(P L )dP L

[0038] Among them, P(E i ) represents the probability of branch i being decommissioned, which generally follows the Poisson distribution; P L represents the probability distribution of the system being in a specific operating state; S represents the severity of overload; S ii represents the overload severity of the i-th branch in the j-th state; V represents the low voltage severity, which depends on the voltage of the node after the accident;

[0039] The severity of overload depends on the power flow distribution of all other branches after the accident. The severity function is expressed as:

[0040]

[0041] Among them, Sev is the abbreviation of Severity, which means the severity function. Indicates actual power P and rated power P e The ratio of P represents the actual power, P e Indicates rated power;

[0042] Load loss risk is the combination of the possibility and severity of load loss at a system load node after an accident.

[0043] The formula is:

[0044]

[0045] Where m represents the number of load nodes, P qi Indicates accident E i The load lost by the i-th load node, Risk(P q |Ei ) indicates that in event E i Occurrence load P q The risk of P(E i ) represents event E i The probability of occurrence, Sev(P qi ) represents power P qi The corresponding severity.

[0046] As a preferred solution of the risk accurate identification method based on multi-element load situation awareness described in the present invention, wherein: based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify the load type and adapt to load management in complex environments, including dividing the population into leaders and followers, the role of the leader is to lead the followers to find food, let the food be G, and the position matrix of the salp group in the d-dimensional space is set to Sn×d, where the element si,j represents the position of the i-th salp in the j-dimensional search space, and n represents the salp population capacity;

[0047] The leader in the SSA algorithm updates its position based on the food position. The leader's position update formula is expressed as:

[0048]

[0049] Among them, u j represents the upper limit of the jth dimension, G j Indicates the position of the leader, s i，j represents the current position of the jth individual, r1 represents the upper limit of the jth dimension, l j Indicates the lower limit value of the j-th dimension;

[0050] r1, r2, and r3 all represent adjustment coefficients. The value thresholds of r1 and r2 are both between 0 and 1. k represents the number of iterations. max Indicates the maximum number of iterations;

[0051] The follower moves along with the leader, and the follower's position update formula is expressed as:

[0052]

[0053] Among them, s i，j represents the position of the jth individual at the i-th iteration;

[0054] The MSISSA algorithm dynamically and adaptively adjusts the size of the leader population. It introduces a nonlinear update mechanism for the proportion of leader individuals based on a fuzzy inference system. The two input variables of the fuzzy inference system that defines the leader proportion factor are population diversity. and algorithm iteration stage k, where population diversity It represents the Euclidean distance between each salp individual and the current optimal solution individual, and the formula is expressed as:

[0055]

[0056] in, represents the average Euclidean distance metric, N represents the population size, D represents the dimension of the problem, and x i，j (t) represents the value of the jth dimension of the ith individual at time t, x be s t，j (t) represents the value of the best individual in the jth dimension at time t;

[0057] Iteration stage k represents the ratio of the current number of iterations of the algorithm to the maximum number of iterations, which is defined as:

[0058] k=t / T max

[0059] Among them, T max Indicates the maximum number of iterations of the algorithm.

[0060] As a preferred solution of the risk accurate identification method based on multi-element load situation awareness described in the present invention, wherein: using the identified load type, designing a load management peak shifting plan optimization model, minimizing the impact of the peak shifting plan in load management on users includes, taking the minimum index of users affected by the peak shifting plan in load management as the objective function, constructing a load management peak shifting plan optimization model;

[0061] The formula is:

[0062]

[0063] Among them, f represents the impact index of users in the peak shifting plan in load management, N represents the total number of users, T i represents the change time of the electricity consumption behavior of off-peak user i, W i represents the daily electricity consumption of off-peak user i, g(T i ) represents the weighted coefficient of the time of change of the user's electricity consumption behavior. The longer the change time, the greater the weighted coefficient. Considering that the impact of electricity consumption behavior in advance on users is greater than that in postponement, the following weighted method is adopted, and the formula is expressed as:

[0064]

[0065] Branch flow constraint. In the power system, due to the constraints of wire diameter and material, the power transmitted by each branch is capped. Branch flow constraint means that the branch transmission power shall not exceed its transmission limit power. The formula of branch flow constraint is expressed as:

[0066] S j ≤Sj，max

[0067] Among them, S j represents the load of the jth branch, S j，max represents the maximum load or power of the jth branch;

[0068] Node voltage constraint, the node voltage is an important indicator to characterize the stability and power quality of the power system. In order to ensure that the system node fluctuation value is within a certain range during load management, the upper and lower limits of the node voltage must be constrained. The formula for the node voltage constraint is expressed as:

[0069] U min ≤U i ≤U max

[0070] Among them, U i represents the voltage amplitude of node i, U min Indicates the minimum value of the voltage amplitude at node i, U max represents the maximum value of the voltage amplitude at node i;

[0071] The power network equation with node voltage and branch current as variables, in the rectangular coordinate system, when the branch conductivity to the ground is ignored, the power network can be described as a mixed form of branch current and node voltage equations, and the formula is expressed as:

[0072]

[0073] For node i, the formula is:

[0074]

[0075] Wherein, L represents the number of branches l=1, 2, ..., L; N represents the number of nodes, i, j=1, 2, ..., N; lei represents that branch ι is associated with node i; represents the real part of the current in branch l, Represents the imaginary part of the current in branch l; U i represents the voltage amplitude at node i, θ i represents the phase angle of the voltage at node i; R ij represents the resistance of branch 1, X ij Represents the reactance of branch l; B l represents the 1 / 2 susceptance of branch l to ground; p i represents the active power injected by the node, q i Represents the reactive power injected by the node;

[0076] For node j, there is an equation of the same form as the above two equations;

[0077] set up represents the real part of the node injected current, let represents the imaginary part of the current injected into the WSN node;

[0078] To get B i0 =∑ 1∈i The real and imaginary parts of the WSN node voltage of B1 are expressed as:

[0079]

[0080] Among them, x i represents the reactance associated with node i, y i represents the conductance associated with node i; s i，i represents the position of the jth individual at the i-th iteration; s i-1，j represents the position of the jth individual at the i-1th iteration;

[0081] The node voltage expression is expressed by the branch current and the node injection power. The generator node is usually defined as a PV node;

[0082] The formula for the boundary condition of the power flow without solution is expressed as:

[0083]

[0084] U min ≤U i ≤U max

[0085] When the amplitude of the node injection current is in a circle with O as the center, When it is outside a circle with a radius of , the equation has high and low pressure solutions, and the system is solvable by flow.

[0086] Only when the equality sign holds, the equation has a unique solution, which is on the circle. When the condition is met, the equation has no real solution, the nodes are located inside the circle, and the system is unsolvable by flow.

[0087] As a preferred solution of the risk accurate identification method based on multi-element load situation awareness described in the present invention, wherein: based on the risk results after quantitative analysis, the SSA algorithm is used to construct a new load identification model to identify the load type, and to adapt to the load management in a complex environment. It also includes that for the optimization adjustment strategy with no solution for the power flow, the minimum node load adjustment amount is used as the objective function, and the formula is expressed as follows:

[0088]

[0089]

[0090] Among them, p α represents the active power boundary value of a node, q αrepresents the reactive power boundary value of a node, pb represents the initial load value given by the node, q b represents the initial value of the node load, x i represents the reactance of node i, y i represents the conductance B of node i i0 represents the reference reactive power of node i, q i represents the reactive power of node i, p i represents the active power of node i;

[0091] The minimum Euclidean distance is used as the optimization target. The load adjustment direction with the minimum adjustment amount is obtained. The inequality constraint condition is:

[0092]

[0093] The results are expressed as:

[0094]

[0095] The optimization model with the minimum node load adjustment as the objective function and constraint condition forms the Lagrangian equation L(s, t).

[0096] The Kuhn-Tucker conditional formula of L(s, t) is expressed as:

[0097]

[0098] Among them, α and β represent Lagrangian factors;

[0099] The formula for the change direction of the minimum load margin of node b is expressed as:

[0100]

[0101] The minimum load margin is expressed by the Euclidean distance formula:

[0102]

[0103] Where s represents the active and reactive power of node a, t represents the real and imaginary parts of the current associated with node a, and η b represents the minimum load margin of node b, p a represents the active power of node a, p b represents the active power of node b, q a represents the reactive power of node a, q b Represents the reactive power of node b.

[0104] As a preferred solution of the risk accurate identification method based on multi-element load situation awareness described in the present invention, wherein: VMD technology is used to construct a load situation awareness model for high-energy-consuming industries to perform load data denoising, and the evaluation index of the denoising effect is proposed, including using variational mode decomposition to realize the denoising process of the bus load, which consists of two parts: construction and decomposition. For the original signal X(t), the intrinsic mode function uk(t) with the smallest bandwidth is sought, and the sum of the intrinsic mode functions can be equal to the original function. Let δ(t) represent the Dirichlet function, * is the convolution symbol, and the established constraint conditions are:

[0105]

[0106] Among them, u k (t) represents the kth signal or control variable, represents an oscillating signal;

[0107] After assuming the constraints of the intrinsic mode function, the Lagrange multiplier method is used to unconstrain the constrained problem and the following function expression is constructed:

[0108]

[0109] Initialize the object to be solved iteratively On this basis, using iteration

[0110] The function is iteratively updated under ω ≥ 0 ω k and

[0111]

[0112]

[0113] The VMD noise reduction evaluation index is the reconstructed signal formula after the noisy signal X(t) is decomposed by variational mode:

[0114]

[0115] Among them, M(t) represents the remainder, and n represents the IMF quantity;

[0116] Assume that the noisy signal is expressed as:

[0117] x=(x1,x2,…,x m )

[0118] The mean square error of the noise reduction deviation is defined as:

[0119] Define the correlation as:

[0120]

[0121] Defining Standard Deviation for

[0122]

[0123] MSE is a relative error index, and the correlation coefficient p is used to characterize the accuracy of the reconstructed curve of the intrinsic mode function after denoising and the bus load curve before denoising. The closer the correlation coefficient is to 1, the higher the overlap with the bus load curve before denoising.

[0124] Combining the noise reduction bias mean square error with the correlation coefficient, the judgment criterion is:

[0125]

[0126] Among them, α represents the influencing factor of the algorithm approximation, and its value range is between 0 and 1;

[0127] If the impact factor is increased, that is, the proportion of the mean square error of the deviation in the judgment criterion is adjusted, the judgment criterion will pay more attention to the impact of the mean square error on the degree of noise reduction.

[0128] The risk accurate identification system based on multi-element load situation awareness includes: data acquisition module, which collects data through WSN nodes, builds a load risk association model, and quantitatively analyzes the load risk in the power system;

[0129] Type identification module, based on the risk results after quantitative analysis, uses the SSA algorithm to build a new load identification model to identify load types and adapt to load management in complex environments;

[0130] The load management module uses the identified load types to design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users;

[0131] The noise reduction processing module uses VMD technology to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction processing, and proposes evaluation indicators for noise reduction effects.

[0132] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0133] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any one of the methods of the present invention.

[0134] Beneficial effects of the present invention: The risk accurate identification method based on multi-element load situation awareness provided by the present invention realizes accurate identification and optimized management of load risks in the power system by constructing a load risk association model and adopting a salp group optimization algorithm, which significantly improves accuracy, adaptability and real-time performance. The peak-shifting plan optimization model reduces user impact and improves operating efficiency. The application of VMD technology in load situation awareness improves data processing accuracy and management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0135] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0136] Figure 1 An overall flow chart of a method for accurate risk identification based on multi-load situation awareness provided by the first embodiment of the present invention;

[0137] Figure 2 Three load risk correlation diagrams in the risk accurate identification method based on multi-load situation awareness provided by the first embodiment of the present invention;

[0138] Figure 3 A flow solution boundary condition diagram for the risk accurate identification method based on multi-load situation awareness provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0139] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0140] Example 1

[0141] Reference Figure 1 , is an embodiment of the present invention, and provides a risk accurate identification method based on multi-load situation awareness, including:

[0142] S1: Collect data through WSN nodes, build a load risk association model, and quantitatively analyze the load risk in the power system.

[0143] The data collected through WSN nodes includes: the WSN node coverage model describes that a number of WSN nodes are deployed in a two-dimensional plane area S = L × W, there are a number of obstacles in the area, and the entire area to be monitored is discretized into L × W pixels;

[0144] Assume that the WSN node set N = {n1, n2, ..., n k , k represents the number of WSN nodes, and the obstacle set O = {o 1，i o 2，i ..., o j，i}, j represents the number of obstacles, i represents the type of obstacles; let the center of the power system in the two-dimensional plane be the origin of the coordinate system, let (x k ,y k ) is the WSN node n k The coordinates of the point to be detected are (x g ,y g ), wireless sensor node n k The probability of sensing the g-point is expressed as:

[0145]

[0146] 0 <r e <r

[0147] α=d(n k , g)-(rr e )

[0148] Among them, d(n k , g) represents WSN node n k The Euclidean distance from the target point g, r e It represents the uncertainty measure of the perception ability of WSN nodes in an obstacle environment, and λ and β represent the perception factors;

[0149] Calculate WSN node n k The Euclidean distance d(n k , g) The formula is expressed as:

[0150]

[0151] Among them, (x g ,y g ) represents the coordinates of the target point g;

[0152] The joint perception probability of n WSN nodes to the same target point is obtained by superposition calculation. Node coverage calculation: Calculate the coverage of the entire area;

[0153] Since a target point g can be covered by multiple WSN nodes at the same time, its joint perception probability P(g) is defined as:

[0154]

[0155] Where p(N, g) represents a single WSN node n k The perceived probability of point g;

[0156] The coverage formula of the entire area is expressed as:

[0157]

[0158] Wherein, L represents the length of the area to be monitored, and W represents the width of the area to be monitored;

[0159] In complex environments, there are usually obstacles in the monitored area. If the WSN node deployment location falls into the obstacle area, the node needs to be moved and redeployed. In order to avoid the clustering of WSN node deployment caused by selecting inappropriate constraints, which leads to a decrease in regional coverage, a virtual repulsive field strategy based on artificial potential field is designed to redeploy the WSN node position.

[0160] Establish an obstacle model for WSN node deployment. The actual obstacle is contained in a circle, and the size of the obstacle determines the radius r of the circle. o The size of the circle is also the area where the repulsive field exists. The distance between the WSN node and the obstacle is simplified to the distance from the center of the circle o i If the node falls into the circle, it will be affected by the repulsive force of the obstacle and move along the repulsive direction by a distance corresponding to the repulsive force, so that the WSN node can avoid the obstacle.

[0161] The formula for the repulsive force from obstacles is:

[0162]

[0163] Among them, F no Represents the impact of obstacles on WSN node n i The repulsive force, r o The radius of the circle d(n i , o iceriter ) represents WSN node n i and obstacle center o iceriter The distance between i represents the position vector of the i-th WSN node, o iceriter Represents the center position vector of the obstacle;

[0164] When there is a deployment area boundary around the obstacle, in order to prevent the WSN node from moving beyond the boundary, the node moving toward the boundary will move in the opposite direction of the repulsive force. The magnitude of the repulsive force is defined as:

[0165] F′ no =2r o -F no

[0166] The improved SSA algorithm enables nodes to avoid obstacles and cover more areas, improving the coverage and efficiency of load management.

[0167] The improved Salp Alveolus algorithm enables nodes to avoid obstacles and cover more areas, improving the coverage and efficiency of load management. At the same time, the algorithm has the ability to dynamically adapt to environmental changes, and can adjust node positions and paths in real time to adapt to the emergence of new obstacles, ensuring that nodes cover key areas and enhancing the reliability and flexibility of load management. WSN node coverage optimization based on obstacle environments can improve data collection efficiency, adapt to complex environments, dynamically adjust node positions, and provide early warning of potential risks in power grid replication management, which helps to enhance the power grid security management and risk control capabilities.

[0168] S2: Based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify load types and adapt to load management in complex environments.

[0169] Since overload risk, low voltage risk and load loss risk are closely related, a quantitative correlation analysis is performed, and the importance ranking and classification of power system branches are applied in the analysis process, as well as an iterative self-organizing data analysis algorithm;

[0170] Overload risk is the combination of the possibility and severity of the active power of other non-fault branches exceeding their rated values ​​after a power system accident. The formula is:

[0171] Risk(P L |E i )=P(E i )∫f(P L |E i , L)×Sev(P L )dP L

[0172] Among them, P(E i ) represents the probability of branch i being decommissioned, which generally follows the Poisson distribution; P L represents the probability distribution of the system being in a specific operating state; S represents the severity of overload; S ij represents the overload severity of the i-th branch in the j-th state; V represents the low voltage severity, which depends on the voltage of the node after the accident;

[0173] The severity of overload depends on the power flow distribution of all other branches after the accident. The severity function is expressed as:

[0174]

[0175] Among them, Sev is the abbreviation of Severity, which means the severity function. Indicates actual power P and rated power P e The ratio of P represents the actual power, P e Indicates rated power;

[0176] Load loss risk is the combination of the possibility and severity of load loss at a system load node after an accident.

[0177] The formula is:

[0178]

[0179] Where m represents the number of load nodes, P qi Indicates accident E i The load lost by the i-th load node, Risk(P q |E i ) indicates that in event E i Occurrence load P q The risk of P(E i ) represents event E i The probability of occurrence, Sev(P qi ) represents power P qi The corresponding severity.

[0180] The relationship between overload risk, low voltage risk and load loss risk is as follows: Figure 2 shown.

[0181] The Salp Swarm Algorithm (SSA) is a new type of heuristic optimization algorithm proposed by Mirjalili et al. in 2017. It is based on the foraging behavior of salps. Salps are marine animals that are very similar to jellyfish. Their movement methods and body tissues are almost the same. During the foraging process, salps are connected head to tail in a chain to move, and individuals cooperate with each other to eventually find food. The population is divided into leaders and followers. The role of the leader is to lead the followers to find food. Let the food be G, and the position matrix of the salp group in the d-dimensional space is Sn×d, where the element si,j represents the position of the i-th salp in the j-dimensional search space, and n represents the salp population capacity;

[0182] The leader in the SSA algorithm updates its position based on the food position. The leader's position update formula is expressed as:

[0183]

[0184] Among them, u j represents the upper limit of the jth dimension, G j Indicates the position of the leader, s i，j represents the current position of the jth individual, r1 represents the upper limit of the jth dimension, l j Indicates the lower limit value of the j-th dimension;

[0185] r1, r2, and r3 all represent adjustment coefficients. The value thresholds of r1 and r2 are both between 0 and 1. k represents the number of iterations. max Indicates the maximum number of iterations;

[0186] The follower moves along with the leader, and the follower's position update formula is expressed as:

[0187]

[0188] Among them, s i，j represents the position of the jth individual at the i-th iteration;

[0189] Considering that the traditional salp algorithm has some shortcomings in power grid load management, such as the inability to consider the restrictions of obstacles on node layout and the lack of adaptability to complex environments, this scheme proposes an improved salp swarm algorithm based on WSN node coverage optimization and multi-strategy fusion in obstacle environment:

[0190] In the SSA algorithm, the leader represents the dominant group of the population, but during the algorithm iteration search process, the size of the leader's salp population remains unchanged. However, in the early stages of the algorithm iteration, a larger leader is conducive to a more comprehensive global search of the SSA algorithm and accelerates the convergence of the algorithm; in the later stages of the algorithm iteration, more emphasis should be placed on local development, and the size of the leader should be reduced to facilitate the SSA algorithm to carry out fine development and improve the convergence accuracy of the algorithm.

[0191] The MSISSA algorithm dynamically and adaptively adjusts the size of the leader population. It introduces a nonlinear update mechanism for the proportion of leader individuals based on a fuzzy inference system. The two input variables of the fuzzy inference system that defines the leader proportion factor are population diversity. and algorithm iteration stage k, where population diversity It represents the Euclidean distance between each salp individual and the current optimal solution individual, and the formula is expressed as:

[0192]

[0193] in, represents the average Euclidean distance metric, N represents the population size, D represents the dimension of the problem, and x i，j (t) represents the value of the jth dimension of the ith individual at time t, x best，j (t) represents the value of the best individual in the jth dimension at time t;

[0194] Iteration stage k represents the ratio of the current number of iterations of the algorithm to the maximum number of iterations, which is defined as:

[0195] k=t / T max

[0196] Among them, T max Indicates the maximum number of iterations of the algorithm.

[0197] S3: Using the identified load types, design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users.

[0198] Taking the minimum impact index of users in the peak-shifting plan in load management as the objective function, an optimization model for the peak-shifting plan of load management is constructed;

[0199] The formula is:

[0200]

[0201] Among them, f represents the impact index of users in the peak shifting plan in load management, N represents the total number of users, T i represents the change time of the electricity consumption behavior of off-peak user i, W i represents the daily electricity consumption of off-peak user i, g(T i ) represents the weighted coefficient of the time of change of the user's electricity consumption behavior. The longer the change time, the greater the weighted coefficient. Considering that the impact of electricity consumption behavior in advance on users is greater than that in postponement, the following weighted method is adopted, and the formula is expressed as:

[0202]

[0203] Branch flow constraint. In the power system, due to the constraints of wire diameter and material, the power transmitted by each branch is capped. Branch flow constraint means that the branch transmission power shall not exceed its transmission limit power. The formula of branch flow constraint is expressed as:

[0204] S j ≤S j，max

[0205] Among them, S j represents the load of the jth branch, S j，max represents the maximum load or power of the jth branch;

[0206] Node voltage constraint, the node voltage is an important indicator to characterize the stability and power quality of the power system. In order to ensure that the system node fluctuation value is within a certain range during load management, the upper and lower limits of the node voltage must be constrained. The formula for the node voltage constraint is expressed as:

[0207] U min ≤U i ≤U max

[0208] Among them, U i represents the voltage amplitude of node i, U min Indicates the minimum value of the voltage amplitude at node i, U max represents the maximum value of the voltage amplitude at node i;

[0209] The power network equation with node voltage and branch current as variables, in the rectangular coordinate system, when the branch conductivity to the ground is ignored, the power network can be described as a mixed form of branch current and node voltage equations, and the formula is expressed as:

[0210]

[0211] For node i, the formula is:

[0212]

[0213] Wherein, L represents the number of branches l=1, 2, ..., L; N represents the number of nodes, i, j=1, 2, ..., N; lei represents that branch ι is associated with node i; represents the real part of the current in branch l, Represents the imaginary part of the current in branch l; U i represents the voltage amplitude at node i, θ i represents the phase angle of the voltage at node i; R ij represents the resistance of branch 1, X ij Represents the reactance of branch l; B l represents the 1 / 2 susceptance of branch l to ground; p i represents the active power injected by the node, q i Represents the reactive power injected by the node;

[0214] For node j, there is an equation of the same form as the above two equations;

[0215] set up represents the real part of the node injected current, let represents the imaginary part of the current injected into the WSN node;

[0216] To get B i0 =∑ 1∈iThe real and imaginary parts of the WSN node voltage of B1 are expressed as:

[0217]

[0218] Among them, x i represents the reactance associated with node i, y i represents the conductance associated with node i; s i，j represents the position of the jth individual at the i-th iteration; s i-1，j represents the position of the jth individual at the i-1th iteration;

[0219] The node voltage expression is expressed by the branch current and the node injection power. The generator node is usually defined as a PV node;

[0220] The formula for the boundary condition of the power flow without solution is expressed as:

[0221]

[0222] When the amplitude of the node injection current is in a circle with O as the center, When it is outside a circle with a radius of , the equation has high and low pressure solutions, and the system is solvable by flow.

[0223] Similar to PQ nodes, PV nodes also have the problem of dual solutions, so other nodes in the system except the balance nodes can use the same method to study the power flow no-solution adjustment strategy.

[0224] Boundary conditions for unsolvable flow:

[0225]

[0226] When the amplitude of the node injection current is outside the circle with O as the center and radius as Figure 3 The equation has high and low pressure solutions, and the system is flow solvable.

[0227] Only when the equality sign holds, the equation has a unique solution, which is on the circle, such as Figure 3 When the condition is met, the equation has no real solution and the node is located inside the circle, such as Figure 3 At point b, the system is trend-insoluble.

[0228] In this case, the system can be adjusted to the solution region by adjusting the generator output, investing in reactive power compensation devices, changing the load level, etc. The most basic requirement is to adjust to point a on the boundary of the solution. Of course, there are multiple such points a in different adjustment directions (such as Figure 3 Finding the nearest point represents the minimum adjustment amount. It is the optimization adjustment target when the equation has no solution.

[0229] Research on power system load management through the Salp Algorithm can give full play to the advantages of the algorithm in solving complex optimization problems, improve the efficiency and stability of power system load management, and provide support and guarantee for the sustainable development of the power system.

[0230] Compared with other similar models, the power flow optimization model has the advantages of higher robustness and adaptability in solving problems such as overload and line congestion in power system load, and can achieve optimal load configuration. Therefore, research on the optimization of internal functions of the load management system is carried out based on the power flow optimization model.

[0231] For the optimization adjustment strategy with no solution for power flow, the minimum node load adjustment is used as the objective function, and the formula is expressed as:

[0232]

[0233] Among them, p α represents the active power boundary value of a node, q α represents the reactive power boundary value of a node, pb represents the initial load value given by the node, q b represents the initial value of the node load, x i represents the reactance of node i, y i represents the conductance B of node i i0 represents the reference reactive power of node i, q i represents the reactive power of node i, p i represents the active power of node i;

[0234] The minimum Euclidean distance is used as the optimization target. The load adjustment direction with the minimum adjustment amount is obtained. The inequality constraint condition is:

[0235]

[0236] The results are expressed as:

[0237]

[0238] The optimization model with the minimum node load adjustment as the objective function and constraint condition forms the Lagrangian equation L(s, t).

[0239] The Kuhn-Tucker conditional formula of L(s, t) is expressed as:

[0240]

[0241] Among them, α and β represent Lagrangian factors;

[0242] The formula for the change direction of the minimum load margin of node b is expressed as:

[0243]

[0244] The minimum load margin is expressed by the Euclidean distance formula:

[0245]

[0246] Where s represents the active and reactive power of node a, t represents the real and imaginary parts of the current associated with node a, and η b represents the minimum load margin of node b, p a represents the active power of node a, p b represents the active power of node b, q a represents the reactive power of node a, q b Represents the reactive power of node b.

[0247] S4: Use VMD technology to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction, and propose evaluation indicators for noise reduction effects.

[0248] The process of bus load noise reduction is realized by using variational mode decomposition, which consists of two parts: construction and decomposition. In order to find the intrinsic mode function uk(t) with the smallest bandwidth for the original signal X(t), and to make the sum of the intrinsic mode functions equal to the original function, let δ(t) represent the Dirichlet function, * is the convolution symbol, and the established constraints are:

[0249]

[0250] Among them, u k (t) represents the kth signal or control variable, represents an oscillating signal;

[0251] After assuming the constraints of the intrinsic mode function, the Lagrange multiplier method is used to unconstrain the constrained problem and the following function expression is constructed:

[0252]

[0253] Initialize the object to be solved iteratively On this basis, the iterative function is used to iteratively update ω k and

[0254]

[0255] The VMD noise reduction evaluation index is the reconstructed signal formula after the noisy signal X(t) is decomposed by variational mode:

[0256]

[0257] Among them, M(t) represents the remainder, and n represents the IMF quantity;

[0258] Assume that the noisy signal is expressed as:

[0259] x=(x1,x2,…,x m )

[0260] The mean square error of the noise reduction deviation is defined as:

[0261] Define the correlation as:

[0262]

[0263] Define standard deviation δ A , δ B for

[0264]

[0265] MSE is a relative error index, and the correlation coefficient p is used to characterize the accuracy of the reconstructed curve of the intrinsic mode function after denoising and the bus load curve before denoising. The closer the correlation coefficient is to 1, the higher the overlap with the bus load curve before denoising.

[0266] Combining the noise reduction bias mean square error with the correlation coefficient, the judgment criterion is:

[0267]

[0268] Among them, α represents the influencing factor of the algorithm approximation, and its value range is between 0 and 1;

[0269] If the impact factor is increased, that is, the proportion of the mean square error of the deviation in the judgment criterion is adjusted, the judgment criterion will pay more attention to the impact of the mean square error on the degree of noise reduction.

[0270] On the other hand, this embodiment also provides a risk accurate identification system based on multi-load situation awareness, which includes:

[0271] The data acquisition module collects data through WSN nodes, builds a load risk association model, and quantitatively analyzes the load risk in the power system;

[0272] Type identification module, based on the risk results after quantitative analysis, uses the SSA algorithm to build a new load identification model to identify load types and adapt to load management in complex environments;

[0273] The load management module uses the identified load types to design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users;

[0274] The noise reduction processing module uses VMD technology to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction processing, and proposes evaluation indicators for noise reduction effects.

[0275] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0276] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0277] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0278] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0279] Example 2

[0280] The experiment was conducted in a power system network, including 10 load nodes and 15 branches. The experimental equipment included high-performance computing servers, data acquisition equipment and power system simulation software.

[0281] Software tools include Matlab, Python and power system simulation software PSCAD for data processing, algorithm implementation and result verification.

[0282] Collect load data, power data and voltage data from the actual power system and perform data preprocessing, including data cleaning, noise reduction and normalization.

[0283] Generate different load scenarios and simulate three risk situations: overload, low voltage and load loss.

[0284] The Poisson distribution is used to simulate the probability of branch withdrawal. Combined with the idea of ​​branch importance sorting and grading, the severity of overload risk, low voltage risk and load loss risk is calculated through iterative self-organizing data analysis algorithm. Construct a new load identification model: Apply the Salp Swarm Optimization Algorithm (SSA) and the multi-strategy fusion improved algorithm to divide the leader and follower roles, dynamically adjust the leader population size, and optimize the load identification accuracy and efficiency. Load management peak shifting plan optimization model: Construct a load management peak shifting plan optimization model. The objective function is to minimize the user impact index, consider the weighted coefficient of the user's electricity behavior change time, and set the branch flow constraint and node voltage constraint. Load situation awareness model for high-energy-consuming industries: The variational mode decomposition (VMD) method is used to reduce the noise of load data to improve the accuracy and reliability of the data. The experimental results are shown in Table 1.

[0285] Table 1 Experimental data table

[0286]

[0287] The risk severity of each load node in the table shows that the present invention can accurately identify the risks of different load nodes under various load conditions. The risk severity of load node 7 reaches 0.95, which is significantly higher than other nodes, indicating that this node has a higher risk in the simulated load scenario. Through an accurate load risk identification model, the present invention can effectively identify and predict potential risks in the power system and provide a reliable basis for risk assessment.

[0288] The optimization performance data show that the optimization performance of the present invention at each load node is above 85%, with the highest reaching 92%. This shows that the present invention can significantly improve the optimization performance of load identification and management by using the salp swarm optimization algorithm and the multi-strategy fusion improved algorithm. In particular, in the optimization process of high-risk nodes, by dynamically adjusting the leader population size, more efficient global search and local development are achieved, improving the optimization accuracy and efficiency.

[0289] The real-time monitoring and intelligent management of load node data performed well in the experiment. Through efficient algorithms and real-time data processing, the present invention realizes real-time monitoring and optimization management of loads, and improves the dispatching efficiency and response speed of the system. For example, the voltage amplitudes of load nodes 4 and 5 are both above 1.04, indicating that during the load optimization process, the system voltage remains within a stable range, showing the superiority of the present invention in intelligent management.

[0290] By combining a variety of advanced algorithms and data processing technologies, the present invention not only improves the accuracy and efficiency of load risk identification and optimization management, but also realizes the intelligent and real-time management of power system loads. These innovative features make the present invention highly practical and popularizable in practical applications, and can effectively solve the shortcomings of existing technologies in terms of optimization efficiency, real-time performance and intelligence.

[0291] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. The risk accurate identification method based on multi-load situation awareness is characterized by: include: Collect data through WSN nodes, build load risk association model, and quantitatively analyze load risks in the power system; Based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify load types and adapt to load management in complex environments; Using the identified load types, we design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users. VMD technology is used to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction, and evaluation indicators for noise reduction effects are proposed.

2. The risk accurate identification method based on multi-load situation awareness according to claim 1 is characterized by: The data collected through WSN nodes includes: the WSN node coverage model describes that a number of WSN nodes are deployed in a two-dimensional plane area S = L × W, there are a number of obstacles in the area, and the entire area to be monitored is discretized into L × W pixels; Assume that the WSN node set N = {n1,n2,…,n k , k represents the number of WSN nodes, and the obstacle set O = {o 1，i o 2，i …,o j，i }, j represents the number of obstacles, i represents the type of obstacles; let the center of the power system in the two-dimensional plane be the origin of the coordinate system, let (x k ,y k ) is the WSN node n k The coordinates of the point to be detected are (x g ,y g ), wireless sensor node n k The probability of sensing the g-point is expressed as: 0<r e <r α=d(n k ,g)-(rr e Among them, d(n k ,g) represents WSN node n k The Euclidean distance from the target point g, r e It represents the uncertainty measure of the perception ability of WSN nodes in an obstacle environment, and λ and β represent the perception factors; Calculate WSN node n k The Euclidean distance d(n k ,g)The formula is expressed as: Among them, (x g ,y g ) represents the coordinates of the target point g; The joint perception probability of n WSN nodes to the same target point is obtained by superposition calculation. Node coverage calculation: Calculate the coverage of the entire area; Since a target point g can be covered by multiple WSN nodes at the same time, its joint perception probability P(g) is defined as: Where p(N,g) represents a single WSN node n k The perceived probability of point g; The coverage formula of the entire area is expressed as: Wherein, L represents the length of the area to be monitored, and W represents the width of the area to be monitored; In complex environments, there are usually obstacles in the monitored area. If the WSN node deployment location falls into the obstacle area, the node needs to be moved and redeployed. In order to avoid the clustering of WSN node deployment caused by selecting inappropriate constraints, which leads to a decrease in regional coverage, a virtual repulsive field strategy based on artificial potential field is designed to redeploy the WSN node position. Establish an obstacle model for WSN node deployment. The actual obstacle is contained in a circle, and the size of the obstacle determines the radius r of the circle. o The size of the circle is also the area where the repulsive field exists. The distance between the WSN node and the obstacle is simplified to the distance from the center of the circle o i If the node falls into the circle, it will be affected by the repulsive force of the obstacle and move along the repulsive direction by a distance corresponding to the repulsive force, so that the WSN node can avoid the obstacle. The formula for the repulsive force from an obstacle is: Among them, F no Represents the impact of obstacles on WSN node n i The repulsive force, r o The radius of the circle d(n i ,o iceriter ) represents WSN node n i and obstacle center o iceriter The distance between i represents the position vector of the i-th WSN node, o iceriter Represents the center position vector of the obstacle; When there is a deployment area boundary around the obstacle, in order to prevent the WSN node from moving beyond the boundary, the node moving toward the boundary will move in the opposite direction of the repulsive force. The magnitude of the repulsive force is defined as: F’ no =2r o -F no The improved SSA algorithm enables nodes to avoid obstacles and cover more areas, improving the coverage and efficiency of load management.

3. The risk accurate identification method based on multi-load situation awareness according to claim 2 is characterized by: Construct a load risk association model to quantitatively analyze the load risks in the power system. Since overload risk, low voltage risk, and load loss risk are closely related, quantitative correlation analysis is performed. The importance ranking and classification of power system branches are applied in the analysis process, and an iterative self-organizing data analysis algorithm is used; Overload risk is the combination of the possibility and severity of the active power of other non-fault branches exceeding their rated values ​​after a power system accident. The formula is: Risk(P L ∣E i )=P(E i )∫f(P L ∣E i ,L)×Sev(P L )dP L Among them, P(E i ) represents the probability of branch i being decommissioned, which generally follows the Poisson distribution; P L represents the probability distribution of the system being in a specific operating state; S represents the severity of overload; S ij represents the overload severity of the i-th branch in the j-th state; V represents the low voltage severity, which depends on the voltage of the node after the accident; The severity of overload depends on the power flow distribution of all other branches after the accident. The severity function is expressed as: Among them, Sev is the abbreviation of Severity, which means the severity function. Indicates actual power P and rated power P e The ratio of P represents the actual power, P e Indicates rated power; Load loss risk is the combination of the possibility and severity of load loss at a system load node after an accident. The formula is: Where m represents the number of load nodes, P qi Indicates accident E i The load lost by the i-th load node, Risk(P q ∣E i ) indicates that in event E i Occurrence load P q The risk of P(E i ) represents event E i The probability of occurrence, Sev(P qi ) represents power P qi The corresponding severity.

4. The risk accurate identification method based on multi-load situation awareness according to claim 3 is characterized by: Based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify the load type. Load management in a complex environment also includes dividing the population into leaders and followers. The role of the leader is to lead the followers to find food. Let the food be G, and the position matrix of the salp group in the d-dimensional space is set to Sn×d, where the element si,j represents the position of the i-th salp in the j-dimensional search space, and n represents the salp population capacity; The leader in the SSA algorithm updates its position based on the food position. The leader's position update formula is expressed as: Among them, u j represents the upper limit of the jth dimension, G j Indicates the position of the leader, s i,j represents the current position of the jth individual, r1 represents the upper limit of the jth dimension, l j Indicates the lower limit value of the j-th dimension; r1, r2, and r3 all represent adjustment coefficients. The value thresholds of r1 and r2 are both between 0 and 1. k represents the number of iterations. max Indicates the maximum number of iterations; The follower moves along with the leader, and the follower's position update formula is expressed as: Among them, s i,j represents the position of the jth individual at the i-th iteration; The MSISSA algorithm dynamically and adaptively adjusts the size of the leader population. It introduces a nonlinear update mechanism for the proportion of leader individuals based on a fuzzy inference system. The two input variables of the fuzzy inference system that defines the leader proportion factor are population diversity. and algorithm iteration stage k, where population diversity It represents the Euclidean distance between each salp individual and the current optimal solution individual, and the formula is expressed as: in, represents the average Euclidean distance metric, N represents the population size, D represents the dimension of the problem, and x i,j (t) represents the value of the jth dimension of the ith individual at time t, x best,j (t) represents the value of the best individual in the jth dimension at time t; Iteration stage k represents the ratio of the current number of iterations of the algorithm to the maximum number of iterations, which is defined as: k=t / T max Among them, T max Indicates the maximum number of iterations of the algorithm.

5. The risk accurate identification method based on multi-load situation awareness according to claim 4 is characterized by: Using the identified load types, an optimization model for load management peak shifting plan is designed to minimize the impact of the peak shifting plan in load management on users, including taking the minimum impact index of users of the peak shifting plan in load management as the objective function and constructing an optimization model for load management peak shifting plan; The formula is: Among them, f represents the impact index of users in the peak shifting plan in load management, N represents the total number of users, T i represents the change time of the electricity consumption behavior of off-peak user i, W i represents the daily electricity consumption of off-peak user i, g(T i ) represents the weighted coefficient of the time of change of the user's electricity consumption behavior. The longer the change time, the greater the weighted coefficient. Considering that the impact of electricity consumption behavior in advance on users is greater than that in postponement, the following weighted method is adopted, and the formula is expressed as: Branch flow constraint. In the power system, due to the constraints of wire diameter and material, the power transmitted by each branch is capped. Branch flow constraint means that the branch transmission power shall not exceed its transmission limit power. The formula of branch flow constraint is expressed as: S j ≤S j,max Among them, S j : represents the load of the jth branch, S j,max represents the maximum load or power of the jth branch; Node voltage constraint, the node voltage is an important indicator to characterize the stability and power quality of the power system. In order to ensure that the system node fluctuation value is within a certain range during load management, the upper and lower limits of the node voltage must be constrained. The formula for the node voltage constraint is expressed as: IN min ≤U i ≤U max Among them, U i represents the voltage amplitude of node i, U min Indicates the minimum value of the voltage amplitude at node i, U max represents the maximum value of the voltage amplitude at node i; The power network equation with node voltage and branch current as variables, in the rectangular coordinate system, when the branch conductivity to the ground is ignored, the power network can be described as a mixed form of branch current and node voltage equations, and the formula is expressed as: For node i, the formula is: Where L represents the number of branches l = l, 2, ..., L; N represents the number of nodes, i, j = l, 2, ..., N; lei represents that branch ι is associated with node i; represents the real part of the current in branch l, Represents the imaginary part of the current in branch l; U i represents the voltage amplitude at node i, θ i represents the phase angle of the voltage at node i; R ij represents the resistance of branch l, X ij represents the reactance of branch l; B l represents the 1 / 2 susceptance of branch l to ground; p i represents the active power injected by the node, q i Represents the reactive power injected by the node; For node j, there is an equation of the same form as the above two equations; set up represents the real part of the node injected current, let represents the imaginary part of the current injected into the WSN node; To get B i0 =∑ 1∈i The real and imaginary parts of the WSN node voltage of B1 are expressed as: Among them, x i represents the reactance associated with node i, y i represents the conductance associated with node i; s i,j represents the position of the jth individual at the i-th iteration; s i-1,j represents the position of the jth individual at the i-1th iteration; The node voltage expression is expressed by the branch current and the node injection power. The generator node is usually defined as a PV node; The formula for the boundary condition of the power flow without solution is expressed as: IN min ≤U i ≤U max When the amplitude of the node injection current is in a circle with O as the center, When it is outside a circle with a radius of , the equation has high and low pressure solutions, and the system is solvable by flow. Only when the equality sign holds, the equation has a unique solution, which is on the circle. When the conditions are met, the equation has no real solution, the nodes are located inside the circle, and the system is unsolvable by flow.

6. The risk accurate identification method based on multi-load situation awareness according to claim 5 is characterized by: Based on the risk results after quantitative analysis, a new load identification model is constructed using the SSA algorithm to identify the load type and adapt to load management in complex environments. For the optimization adjustment strategy with no solution for the power flow, the minimum node load adjustment is used as the objective function. The formula is expressed as follows: Among them, p α represents the active power boundary value of a node, q α Represents the reactive power boundary value of a node, p b represents the initial load value given by the node, q b represents the initial value of the node load, x i represents the reactance of node i, y i represents the conductance B of node i i0 represents the reference reactive power of node i, q i represents the reactive power of node i, p i represents the active power of node i; The minimum Euclidean distance is used as the optimization target to obtain the load adjustment direction with the minimum adjustment amount. The inequality constraint condition is: The results are expressed as: The optimization model with the minimum node load adjustment as the objective function and constraint condition forms the Lagrangian equation L(s,t). The Kuhn-Tucker conditional formula of L(s,t) is expressed as: Among them, α and β represent Lagrangian factors; The formula for the change direction of the minimum load margin of node b is expressed as: The minimum load margin is expressed by the Euclidean distance formula: Where s represents the active and reactive power of node a, t represents the real and imaginary parts of the current associated with node a, and η b represents the minimum load margin of node b, p a represents the active power of node a, p b represents the active power of node b, q a represents the reactive power of node a, q b Represents the reactive power of node b.

7. The risk accurate identification method based on multi-load situation awareness according to claim 6 is characterized by: The VMD technology is used to build a load situation awareness model for high-energy-consuming industries to perform load data denoising, and the evaluation indicators of the denoising effect are proposed, including the use of variational mode decomposition to achieve the denoising process of the bus load, which consists of two parts: construction and decomposition. In order to find the inherent mode function uk(t) with the smallest bandwidth for the original signal X(t), and to make the sum of the inherent mode functions equal to the original function, let δ(t) represent the Dirichlet function, * is the convolution symbol, and the established constraints are: Among them, u k (t) represents the kth signal or control variable, represents an oscillating signal; After assuming the constraints of the intrinsic mode function, the Lagrange multiplier method is used to unconstrain the constrained problem and the following function expression is constructed: Initialize the object to be solved iteratively On this basis, using iteration The function is iteratively updated under ω ≥ 0 ω k and The VMD noise reduction evaluation index is the reconstructed signal formula after the noisy signal X(t) is decomposed by variational mode: Among them, M(t) represents the remainder, and n represents the IMF quantity; Assume that the noisy signal is expressed as: x=(x1,x2,…,x m ) The mean square error of the noise reduction deviation is defined as: Define the correlation as: Define standard deviation δ A ,δ B for MSE is a relative error index, and the correlation coefficient p is used to characterize the accuracy of the reconstructed curve of the intrinsic mode function after denoising and the bus load curve before denoising. The closer the correlation coefficient is to 1, the higher the overlap with the bus load curve before denoising. Combining the noise reduction bias mean square error with the correlation coefficient, the judgment criterion is: Among them, α represents the influencing factor of the algorithm approximation, and its value range is between 0 and 1; If the impact factor is increased, that is, the proportion of the mean square error of the deviation in the judgment criterion is adjusted, the judgment criterion will pay more attention to the impact of the mean square error on the degree of noise reduction.

8. A risk accurate identification system based on multi-load situation awareness using the method according to any one of claims 1 to 7, characterized in that: The data acquisition module collects data through WSN nodes, builds a load risk association model, and quantitatively analyzes the load risk in the power system; Type identification module, based on the risk results after quantitative analysis, uses the SSA algorithm to build a new load identification model to identify load types and adapt to load management in complex environments; The load management module uses the identified load types to design a load management peak shifting plan optimization model to minimize the impact of the peak shifting plan in load management on users; The noise reduction processing module uses VMD technology to build a load situation awareness model for high-energy-consuming industries to perform load data noise reduction processing, and proposes evaluation indicators for noise reduction effects.

9. Computer equipment, including: Memory and processor; The memory stores a computer program, which is characterized in that when the processor executes the computer program, the steps of the risk accurate identification method based on multi-load situation awareness are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for accurately identifying risks based on multi-load situational awareness are implemented.

Images