Method and device for optimizing credible WLAN (Wireless Local Area Network) coverage of substation

Abstract

The invention relates to an optimization method and device for credible WLAN network coverage of a transformer substation, and belongs to the technical field of wireless communication of a power system, and the method comprises the steps: constructing a three-dimensional credible WLAN network node coverage rate model based on a ray tracing and machine learning fused node coverage range prediction method; a Sobol sequence is adopted to initialize a COOT algorithm population so as to enhance population diversity; a golden sine strategy is introduced into random motion position updating of the COOT algorithm, and a greedy strategy is introduced into random selection of individual motion modes; in the COOT algorithm optimization process, a disturbance strategy based on virtual force is introduced to carry out disturbance updating on the position of the leader; and calculating and solving a target function of the three-dimensional trusted WLAN network node coverage rate model by using the optimized COOT algorithm to obtain an optimal node position. According to the invention, the network coverage rate is improved, the deployment number of communication nodes is reduced, and the network construction cost is saved.

Description

Technical Field

[0001] The present invention relates to an optimization method and device for covering a trusted WLAN network in a substation, belonging to the technical field of wireless communication in power systems. Background Art

[0002] With the continuous development of energy digitalization and intelligence, a digital and intelligent power grid is used to support the construction of a new power system, promoting the construction of intelligent operation and maintenance systems for substations and converter stations, intelligent inspection of transmission lines, and intelligent operation and maintenance systems for distribution, and developing a development goal such as an intelligent perception system for power grid disasters. At present, within the scope of substations, the application of trusted WLAN networks in scenarios such as intelligent inspection of power transmission and transformation, intelligent safety supervision, intelligent warehouses, remote operation and maintenance of secondary systems, and physical coding is being promoted, and as many services as possible are connected to the trusted WLAN network. To better meet the new requirements and challenges encountered in the application process of the trusted WLAN network, there is a strong demand for products, and it is necessary to achieve the coordinated development of the industry and applications.

[0003] Currently, the large-scale deployment of trusted WLAN in the power system provides the interoperability, integrity, and function testing of the wireless local area network authentication and confidentiality infrastructure, and provides services for building the evaluation ability of wireless network security standards, etc. However, at present, in the substation scenario, the environment is complex and there are many devices, and there are many non-line-of-sight transmission environments, and it is difficult to effectively guarantee the coverage performance of the communication system. Therefore, it is urgent to improve the coverage depth of the trusted WLAN network in the complex substation scenario, and the coverage performance of the trusted WLAN network still needs to be optimized; in the substation, coverage optimization is an important indicator reflecting the quality of the monitoring work of the trusted WLAN network in the substation for the target area. Therefore, it is necessary to design a suitable wireless network node deployment strategy and coverage optimization algorithm according to different application requirements, while reducing the redundancy degree of nodes, supplementing the blank areas caused by the initial random deployment, and then improving the quality of network monitoring of the target area and ensuring the accuracy of data.

[0004] To sum up, in order to promote the construction of the trusted WLAN network in the substation, combined with the power service application scenario, it is urgent to study a unified standard, enhanced coverage, and economical and efficient trusted WLAN network deployment plan to provide a practical solution for digital communication and efficient operation and maintenance of equipment in the substation scenario. Summary of the Invention

[0005] To solve the above problems, the present invention proposes an optimization method and device for covering a trusted WLAN network in a substation, which can determine the optimal node layout scheme of the wireless network in a complex environment, thereby improving the network coverage rate, reducing the coverage blind area, reducing the number of deployed communication nodes, and saving the network construction cost.

[0006] The technical solution adopted by the present invention to solve its technical problems is:

[0007] In a first aspect, an optimization method for the coverage of a trusted WLAN network in a substation provided by an embodiment of the present invention includes the following steps:

[0008] Step S1, constructing a three-dimensional trusted WLAN network node coverage rate model based on a node coverage range prediction method that combines ray tracing and machine learning;

[0009] Step S2, initializing the population of the COOT (Coot Bird Optimization Algorithm) algorithm using the Sobol sequence to enhance population diversity;

[0010] Step S3, introducing the golden sine strategy in the random movement position update of the COOT algorithm and introducing the greedy strategy in the random selection of the individual movement mode;

[0011] Step S4, during the optimization process of the COOT algorithm, introducing a perturbation strategy based on virtual force to perturb and update the leader position;

[0012] Step S5, using the optimized COOT algorithm to calculate and solve the objective function of the three-dimensional trusted WLAN network node coverage rate model to obtain the optimal node position, where the objective function is:

[0013]

[0014] In the formula, F cov is the total coverage rate of the trusted WLAN network for all WLAN network communication nodes covered by communication in the target area, m is the number of small square areas into which the target area is divided, V k is the volume of the small square, V is the total volume of the target area, P(U, S j ) is the joint communication probability distribution of the trusted WLAN network communication nodes.

[0015] As a possible implementation manner of this embodiment, the step S1 includes:

[0016] Obtaining channel measurement data using a channel measurement system based on the actual scenario of the substation;

[0017] Constructing a three-dimensional scene simulation model of the substation based on the actual scenario and randomly arranging communication nodes in the target area;

[0018] Calculating the received power of the target node using the ray tracing method;

[0019] Calculating the path loss deviation value using the measured path loss value of the channel and the simulated path loss value obtained based on ray tracing;

[0020] Train a DNN neural network using the path loss deviation values at different receiver positions and the 3D position information of the communication node and the target node to accurately predict the path loss deviation value between any communication node and the target node within the target area;

[0021] Use the path loss deviation value predicted by the neural network to correct the simulated path loss value, obtain the optimized path loss value, and calculate the accurate received power at the corresponding target node;

[0022] Determine the effective coverage range of the communication node according to the received power of the target node and the set received power threshold, and establish a coverage rate model for the trusted WLAN network communication node.

[0023] As a possible implementation manner of this embodiment, the Sobol sequence is used to generate sample points approximately uniformly distributed in the multi-dimensional space, and a low-discrepancy sequence is generated by the quasi-Monte Carlo method to initialize the coot population. The position of the coot population is:

[0024] X C (i) = S N (ub - lb) + lb

[0025] In the formula, [ub, lb] is the search range of the target space, and the random number S N ∈[0, 1], X C (i) is the position of the i-th coot, so as to enhance the diversity of the initialized population distribution in the target space.

[0026] As a possible implementation manner of this embodiment, the golden sine strategy is used to introduce a sine function to dynamically adjust the search step size during the algorithm iteration process to improve the convergence speed and local search ability of the algorithm, and the greedy strategy is used to retain the current optimal solution during the local search stage to reduce the blindness during the movement process.

[0027] As a possible implementation manner of this embodiment, the virtual force-based perturbation strategy includes three virtual forces, and the position of the leader is perturbed and updated under the action of the three virtual forces to balance the local search and global optimization capabilities.

[0028] As a possible implementation manner of this embodiment, the larger the value of the objective function, the larger the coverage range of the node deployment and the more reasonable the node layout position.

[0029] As a possible implementation manner of this embodiment, the step S3 includes:

[0030] Randomly generate the positions of l coot individuals. Assume that each solution to the optimization problem corresponds to the position of the coot in the search space, X i(t) represents the spatial position of the i-th coot individual in the d-dimensional individual space at the t-th iteration, where i = 1, 2,..., l; Y i (t) is the optimal position of the coot individual i at the t-th iteration. In the (t + 1)-th iteration, the position update formula for the i-th coot individual is as follows:

[0031] X i (t + 1) = X i (t) × |sin(G 1 )| + G 2 × sin(G 1 ) × |μ × Y i (t) - ν × X i (t)|

[0032] In the formula, G 1 is a random number in [0, 2π], which determines the moving distance of the i-th coot individual in the next iteration process; G 2 is a random number in [0, π], which determines the position update direction of the i-th coot individual in the next iteration process; μ and ν are coefficients obtained by introducing the golden ratio. The golden ratio The values of μ and ν are calculated as follows:

[0033] μ = -π + (1 - π) × 2π

[0034] ν = -π + λ × 2π

[0035] After introducing the golden sine strategy, in the random movement stage, the new position update formula is as follows:

[0036] X C (i) = X i (t + 1) = X i (t) × |sin(G 1 )| + G 2 × sin(G 1 ) × |μ × Y i (t) - ν × X i (t)|

[0037] First, determine whether the position after the movement in the previous iteration has been improved, and then determine which movement method to select in this iteration. If the position after the movement of the coot individual in the previous iteration has been improved, then the movement method in this movement process is the same as that in the previous iteration; if the position after the movement of the coot individual in the previous iteration has not been improved but has become worse, then the movement method in this movement process is different from that in the previous iteration to obtain a better movement effect. The rule formula is as follows:

[0038]

[0039] Where Move(i) t is the movement mode selected by the coot during the t-th movement; ~ represents the NOT operation; random represents random selection; g(i) t is the fitness value of the coot's t-th movement.

[0040] As a possible implementation of this embodiment, the step S4 includes:

[0041] Adding virtual forces during the position update process of the coot flock leader to change the search area of the leader. The virtual forces are generated by three parts: adjacent communication nodes, uncovered grid points, and the region boundary; calculating the perturbation factor of the virtual force through the following formula for perturbing the position update formula of the leader:

[0042]

[0043] In the formula, f i is the moving distance of the node under the action of the virtual force, F i is the resultant virtual force applied to the communication node, F ij is the virtual force of adjacent communication nodes, F ik is the virtual attraction of uncovered grid points, F ib is the boundary virtual repulsion force, and Step(t) is the single virtual moving step length of the node;

[0044] Set the virtual moving step length to gradually decrease as the number of iterations increases, as shown in the following formula:

[0045]

[0046] In the formula, Stepmax and Stepmin represent the maximum and minimum moving step lengths respectively to ensure the stability of the algorithm and reduce the ineffective movement of the nodes;

[0047] Update the leader position using the improved leader movement formula. The improved leader movement formula is as follows:

[0048]

[0049] In the formula, f i is the moving distance of the node under the action of the virtual force, X L (i) represents the current leader position B 4 and B 5 are random numbers between the interval [0,1]; B 3 is a random number between the interval [-1,1]; X best is the best position that can be found, max_Iter is the maximum number of iterations, and t is the current number of iterations.

[0050] In a second aspect, an optimization device for the coverage of a trusted WLAN network in a substation provided by an embodiment of the present invention includes:

[0051] A model construction module, configured to construct a three-dimensional trusted WLAN network node coverage rate model based on a node coverage range prediction method that combines ray tracing and machine learning;

[0052] A population initialization module, configured to initialize the population of the COOT algorithm using the Sobol sequence to enhance population diversity;

[0053] A strategy introduction module, configured to introduce the golden sine strategy in the random movement position update of the COOT algorithm and introduce the greedy strategy in the random selection of the individual movement mode;

[0054] A perturbation update module, configured to introduce a perturbation strategy based on virtual force to perform perturbation update on the leader position during the optimization process of the COOT algorithm;

[0055] An objective function solving module, configured to use the optimized COOT algorithm to calculate and solve the objective function of the three-dimensional trusted WLAN network node coverage rate model to obtain the optimal node position.

[0056] As a possible implementation manner of this embodiment, the objective function is:

[0057]

[0058] In the formula, F cov is the total coverage rate of the trusted WLAN network for all the WLAN network communication nodes covered by communication in the target area, m is the number of small square areas into which the target area is divided, V k is the volume of the small square, V is the total volume of the target area, and P(U, S j ) is the joint communication probability distribution of the trusted WLAN network communication nodes.

[0059] The beneficial effects produced by the technical solution of the embodiment of the present invention are as follows:

[0060] The present invention provides a method for predicting the coverage range of nodes based on the fusion of ray tracing and machine learning, and constructs a fusion prediction model for the coverage range of nodes applicable to substation scenarios. At the same time, the optimized Common Coot Optimization (COOT) algorithm is used to optimize the positions of trusted WLAN network nodes, and the optimal node layout scheme for wireless networks in complex environments is determined. Compared with traditional node coverage models and traditional COOT algorithms, the coverage rate of the method of the present invention has been improved, and it also has a higher convergence speed, accuracy and convergence efficiency. The optimal coverage rate can be obtained without adding extra nodes. The present invention effectively improves the network coverage rate, reduces the coverage blind area, reduces the number of deployed communication nodes, saves the network construction cost, is applicable to the WLAN network coverage optimization in the complex environment of substations, and has strong practicability and popularization value. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flowchart of an optimization method for the coverage of a trusted WLAN network in a substation shown according to an exemplary embodiment;

[0062] Figure 2 is a schematic structural diagram of an optimization device for the coverage of a trusted WLAN network in a substation shown according to an exemplary embodiment;

[0063] Figure 3 is a specific implementation flowchart of the optimization of the coverage of a trusted WLAN network in a substation shown according to an exemplary embodiment;

[0064] Figure 4 is a schematic structural diagram of a Common Coot Optimization algorithm shown according to an exemplary embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] To more clearly illustrate the technical features of the solution of the present invention, the present invention will be elaborated in detail below through specific embodiments and in conjunction with its accompanying drawings.

[0066] As Figure 1 shown, an optimization method for the coverage of a trusted WLAN network in a substation provided by an embodiment of the present invention includes the following steps:

[0067] Step S1, constructing a three-dimensional trusted WLAN network node coverage rate model based on a method for predicting the coverage range of nodes by fusing ray tracing and machine learning;

[0068] Step S2, initializing the population of the COOT algorithm using the Sobol sequence to enhance the population diversity;

[0069] Step S3, introducing the golden sine strategy in the random movement position update of the COOT algorithm, and introducing the greedy strategy in the random selection of the individual movement mode;

[0070] Step S4, during the optimization process of the COOT algorithm, a perturbation strategy based on virtual force is introduced to perturb and update the leader position;

[0071] Step S5, use the optimized COOT algorithm to calculate and solve the objective function of the three-dimensional credible WLAN network node coverage model to obtain the optimal node positions, and the objective function is:

[0072]

[0073] In the formula, F cov is the total coverage rate of the credible WLAN network for all the WLAN network communication nodes covered by communication in the target area, m is the number of small square areas into which the target area is divided, V k is the volume of the small square, V is the total volume of the target area, P(U, S j ) is the joint communication probability distribution of the credible WLAN network communication nodes.

[0074] As a possible implementation manner of this embodiment, the step S1 includes:

[0075] Obtain channel measurement data using a channel measurement system based on the actual scenario of the substation;

[0076] Construct a three-dimensional scene simulation model of the substation based on the actual scenario, and randomly arrange communication nodes in the target area;

[0077] Calculate the received power of the target node using the ray tracing method;

[0078] Calculate the path loss deviation value using the measured path loss value of the channel and the simulated path loss value obtained based on ray tracing;

[0079] Train a DNN neural network using the path loss deviation values at different receiver positions and the 3D position information of the communication node and the target node to accurately predict the path loss deviation value between any communication node and the target node in the target area;

[0080] Use the path loss deviation value predicted by the neural network to correct the simulated path loss value, obtain the optimized path loss value, and calculate the accurate received power at the corresponding target node;

[0081] Determine the effective coverage range of the communication node according to the received power of the target node and the set received power threshold, and establish a coverage rate model of the credible WLAN network communication nodes.

[0082] As a possible implementation manner of this embodiment, the Sobol sequence is used to generate sample points approximately uniformly distributed in a multi-dimensional space, generate a low-discrepancy sequence through the quasi-Monte Carlo method, and initialize the coot population. The coot population positions are:

[0083] X C (i) = S N (ub - lb) + lb

[0084] Wherein, [ub, lb] is the search range of the target space, and the random number S N ∈[0, 1], X C (i) is the position of the i-th coot, to enhance the diversity of the initial population distribution in the target space.

[0085] As a possible implementation manner of this embodiment, the golden sine strategy is used to introduce a sine function during the algorithm iteration process to dynamically adjust the search step size, improve the convergence speed of the algorithm and the local search ability, and the greedy strategy is used to retain the current optimal solution in the local search stage to reduce the blindness during the movement process.

[0086] As a possible implementation manner of this embodiment, the virtual force-based perturbation strategy includes three virtual forces, and under the action of the three virtual forces, the position of the leader is perturbed and updated to balance the local search and global optimization capabilities.

[0087] As a possible implementation manner of this embodiment, the larger the value of the objective function, the larger the coverage range of the node deployment and the more reasonable the node layout position.

[0088] As a possible implementation manner of this embodiment, step S3 includes:

[0089] Randomly generate the positions of l coot individuals. Assume that each solution to the optimization problem corresponds to the position of the corresponding coot in the search space, X i (t) represents the spatial position of the i-th coot individual in the d-dimensional individual space at the t-th iteration, i = 1, 2,..., l; Y i (t) is the optimal position of the coot individual i at the t-th iteration. In the (t + 1)-th iteration, the position update formula of the i-th coot individual is as follows:

[0090] X i (t + 1) = X i (t) × |sin(G 1 )| + G 2 × sin(G 1 ) × |μ × Y i (t) - ν × X i (t)|

[0091] Wherein, G 1 is a random number in [0, 2π], which determines the moving distance of the i-th coot individual in the next iteration; G 2is a random number in [0, π], which determines the position update direction of the i-th coot individual in the next iteration process; μ and ν are coefficients obtained by introducing the golden ratio, and the golden ratio The values of μ and ν are calculated as follows:

[0092] μ = -π + (1 - π) × 2π

[0093] ν = -π + λ × 2π

[0094] After introducing the golden sine strategy, in the random movement stage, the new position update formula is as follows:

[0095] X C (i) = X i (t + 1) = X i (t) × |sin(G 1 )| + G 2 × sin(G 1 ) × |μ × Y i (t) - ν × X i (t)|

[0096] First, by comparing whether the position after the previous iteration movement has been improved, then determine which movement method to select for this iteration. If the position after the previous coot individual movement has been improved, then the movement method in this movement process is the same as the previous one; if the position after the previous coot individual movement has not been improved but becomes worse, then the movement method in this movement process is different from the previous one to obtain a better movement effect; its rule formula is as follows:

[0097]

[0098] In the formula, Move(i) t is the movement method selected by the coot at the t-th movement; ∼ represents the non-operation; random represents random selection; g(i) t is the fitness value of the coot at the t-th movement.

[0099] By introducing the golden sine strategy in the present invention, the position update of the coot individual is more reasonable, which can more effectively explore the search space and improve the solution efficiency; the present invention combines the greedy strategy, reduces the blindness in the movement process of the coot individual, and improves the convergence speed and solution accuracy.

[0100] As a possible implementation manner of this embodiment, the step S4 includes:

[0101] During the position update process of the coot flock leader, virtual forces are added to change the search area of the leader. The virtual forces are mainly generated by three parts: adjacent communication nodes, uncovered grid points, and regional boundaries. The perturbation factor of the virtual force is calculated by the following formula to perturb the position update formula of the leader:

[0102]

[0103] In the formula, f i is the moving distance of the node under the action of the virtual force, F i is the resultant virtual force applied to the communication node, F ij is the virtual force of adjacent communication nodes, F ik is the virtual attraction of uncovered grid points, F ib is the boundary virtual repulsion force, and Step(t) is the single virtual moving step length of the node;

[0104] Set the virtual moving step length to gradually decrease as the number of iterations increases, as shown in the following formula:

[0105]

[0106] In the formula, Stepmax and Stepmin respectively represent the maximum and minimum moving step lengths to ensure the stability of the algorithm and reduce the invalid movement of the node;

[0107] Update the leader position using the improved leader motion formula. The improved leader motion formula is as follows:

[0108]

[0109] In the formula, f i is the moving distance of the node under the action of the virtual force, X L (i) represents the current leader position B 4 and B 5 are random numbers between [0,1]; B 3 is a random number between [-1,1]; X best is the best position that can be found, max_Iter is the maximum number of iterations, and t is the current number of iterations.

[0110] The present invention effectively avoids the problem of the algorithm falling into a local optimal solution and improves the global optimization ability by adding virtual force perturbation during the leader position update process. The present invention avoids the oscillation of the coverage rate in the later stage of iteration and ensures the stability of the algorithm by setting the virtual moving step length to gradually decrease as the number of iterations increases. The present invention is applicable to the position optimization problem of mobile communication nodes and can significantly improve the coverage rate and communication efficiency in the monitoring area.

[0111] As Figure 2 shown, an optimization device for substation trusted WLAN network coverage provided by an embodiment of the present invention includes:

[0112] A model construction module, configured to construct a three-dimensional trusted WLAN network node coverage rate model based on a node coverage range prediction method that combines ray tracing and machine learning;

[0113] A population initialization module, configured to initialize the COOT algorithm population using the Sobol sequence to enhance population diversity;

[0114] A strategy introduction module, configured to introduce the golden sine strategy in the random movement position update of the COOT algorithm, and introduce the greedy strategy in the random selection of the individual movement mode;

[0115] A perturbation update module, configured to introduce a perturbation strategy based on virtual force to perform perturbation update on the leader position during the optimization process of the COOT algorithm;

[0116] An objective function solving module, configured to use the optimized COOT algorithm to calculate and solve the objective function of the three-dimensional trusted WLAN network node coverage rate model to obtain the optimal node position.

[0117] As a possible implementation manner of this embodiment, the objective function is:

[0118]

[0119] In the formula, F cov is the total coverage rate of the trusted WLAN network for all WLAN network communication nodes covered by communication in the target area, m is the number of small square areas into which the target area is divided, V k is the volume of the small square, V is the total volume of the target area, P(U, S j ) is the joint communication probability distribution of the trusted WLAN network communication nodes.

[0120] As Figure 3 shown, the specific implementation process of optimizing the substation trusted WLAN network coverage by the present invention includes the following steps.

[0121] Step S101: Propose a node coverage range prediction method that combines ray tracing and machine learning, and then establish a coverage rate model of three-dimensional trusted WLAN network nodes.

[0122] The present invention proposes a node coverage range prediction method that combines ray tracing and machine learning. First, based on the actual scenario of the substation, channel measurement data is obtained using a channel measurement system, and the measured path loss value PL can be expressed as:

[0123]

[0124] where d 0 is the path loss reference distance; n PL is the path loss exponent; X σ is the shadow fading, which follows a lognormal distribution.

[0125] Secondly, a 3D scene simulation model of a substation with an area of M×N×L is constructed based on the actual scenario. Suppose n communication nodes are randomly deployed in the target area S, and the node set is represented as U = {u 1 , u 2 , …, u n}, where the position of the communication node u i is (x i , y i , z i ). Assume that the transmit power of the communication nodes in the target area is P i , and the position of the target node S j in the target area is (x j , y j , z j ). According to the position of the target node, the received power P j of the target node is calculated using the ray tracing method. P j is the superposition of the powers of all effective ray paths, which is expressed as follows:

[0126]

[0127] where N is the number of all effective ray paths; P n is the received power of the nth effective ray path, and the corresponding path loss PL S is:

[0128] PL S (dB) = P i (dBm) - P j (dBm) + G i,Max (dBi) + G j,Max (dBi) - L S (dB) (3)

[0129] where P i is the transmit power, P j is the received power of the target node, G i,Max , G j,Max are the maximum gains of the transmit antenna and the receive antenna respectively, and L S is the loss caused in the preset communication system. Subsequently, the path loss deviation value PL D is calculated using the measured path loss value of the channel at the same position in the target area and the simulated path loss value obtained based on ray tracing., which is expressed as follows: PL D =PL - PL S (4)

[0130] Finally, the path loss deviation values PL at different receiver positions obtained from the above calculations D and the 3D position information of the communication node and the target node are used to train the DNN neural network to achieve accurate prediction of the path loss deviation value between any communication node u i and the target node S j in the target area, which is expressed as follows:

[0131] PL E =f L (f L-1 (…f 2 (f 1 (u i , s j , PL D ; W (1) , b (1) ); W (2) , b (2) );…W (L) , b (L) )) (5)

[0132] where PL E is the predicted path loss deviation value; f L is the transformation function of the L-th layer of the DNN neural network; W (L) and b (L) represent the weight and bias of the L-th layer respectively; u i and S j represent the 3D positions of the communication node and the target node respectively.

[0133] The path loss deviation value accurately predicted by the above neural network is used to correct the simulated path loss value PL S obtained by the ray tracing method, and the optimized path loss value can be obtained, and the accurate received power P' j at the corresponding target node is calculated, which is expressed as follows:

[0134] P' j =P i +G i,Max +G j,Max -PL S +PL E -L S (6)

[0135] Set the received power threshold for the target node u i to be able to achieve reliable communication with the communication node S j as P th, if the communication node u i is fixed in position, by continuously adjusting the position of the target node S j to update its received power P′ j , the effective coverage range of the communication node u i can be obtained; that is, when the positions of the communication node u i and the target node S j change, if the received power P′ j of the target node S j is higher than the threshold P th , then the target node can be effectively covered by the communication node u i , that is, the communication probability is 1; conversely, the target node S j cannot be covered, that is, the communication probability is 0. Let P(u i , S j ) represent the communication probability of node u i to S j , and the mathematical expression is as follows:

[0136]

[0137] When the target is within the communication range of the node, it can be successfully covered. The same target in the target area may be covered by multiple communication nodes at the same time, then the joint communication probability distribution of the nodes is:

[0138]

[0139] Suppose the target area is divided into m small square areas, and the volume of each small square is V k . The overall coverage rate of the network is the ratio of the volume of the set of communication probabilities of all sensor nodes to the total volume V of the target area, and the calculation formula is as follows:

[0140]

[0141] In the formula, F cov is the total coverage rate of the trusted WLAN network for all points covered by communication in the target area.

[0142] Step S102: Introduce the Sobol sequence to initialize the population and enrich the diversity of the population.

[0143] In the COOT algorithm, the population represents different communication nodes. The original COOT algorithm uses a random initialization method to generate the initial solution. The random initialization method will greatly reduce the degree of uniformity of the population initialization distribution and hinder the optimization ability of the algorithm. The Sobol sequence is an efficient low-discrepancy sequence used to generate sample points with an approximately uniform distribution in a multi-dimensional space, so as to distribute the sample points as evenly as possible in the target area, which can effectively enhance the diversity of the initialized population distribution in the target space, enabling the COOT algorithm to achieve high optimization in the global area and avoiding the situation of slow convergence speed or unsatisfactory convergence accuracy.

[0144] The present invention utilizes the efficient and low-discrepancy deterministic Sobol sequence S N to complete the initialization of the population. Then the population position expression is as follows:

[0145] X C (i) = S N (ub - lb) + lb (10)

[0146] In the formula, [ub, lb] is the search range of the target space, the random number S N ∈[0, 1], and X C (i) is the position of the i-th coot.

[0147] Step S103: Introduce the golden sine strategy and the greedy strategy to avoid the algorithm falling into the dilemma of local optimal solutions.

[0148] As Figure 4 shown, during the foraging process of coots, the entire group is divided into leaders and followers. The three movement methods for updating the positions of followers are as follows:

[0149] 1) Random movement:

[0150] First, a position Q is randomly generated according to the following formula:

[0151] Q = S N (ub - lb) + lb (11)

[0152] Then, to prevent falling into the local optimum, the position is updated:

[0153] X C (i) = X C (i) + A × B 1 ×(Q - X C (i)) (12)

[0154] In the formula, B 1 is a random number in the interval [0, 1]; A linearly decreases from 1 to 0 as the number of iterations increases, and its formula is as follows:

[0155]

[0156] Among them, max_Iter is the maximum number of iterations, and t is the current number of iterations.

[0157] 2) Chain motion:

[0158] The chain motion is realized by the average position of two coots. One coot moves towards the other coot, and the moving distance is half of the distance vector. The position update formula is as follows:

[0159]

[0160] In the formula, X C (i - 1) is the position of the (i - 1)-th coot.

[0161] 3) Leader-following motion:

[0162] The individual coot updates its own position according to the position of the leader in the group and continuously approaches the leader. The leader is selected according to the following formula:

[0163] k = 1+(i mod N L ) (15)

[0164] In the formula, k is the number of the leader; N L is the number of leaders; mod is the remainder function;

[0165] The position update formula for leader-following motion is as follows:

[0166] X C (i) = X L (k)+2×B 2 ×cos(2πB 3 )×(X L (k)-X C (i)) (16)

[0167] In the formula, X L (k) is the selected leader position; B 2 is a random number in the interval [0, 1]; B 3 is a random number in the interval [-1, 1].

[0168] In the traditional COOT algorithm, the movement of followers is randomly selected from the above three methods. This individual movement selection method is prone to problems such as slow convergence speed and falling into local optimal solutions. The present invention introduces the golden sine strategy in the position update of individual random movements. Using the sine function for iterative optimization, it can traverse all points within the effective region, and this traversal behavior is similar to the global search in optimization problems. At the same time, the golden section coefficient is introduced in the position update process, which not only speeds up the convergence speed of the algorithm but also improves the local search ability.

[0169] The update process of the solution of the golden sine strategy is its key core. First, randomly generate the positions of l coots. Assume that each solution of the optimization problem corresponds to the position of the coot in the corresponding search space, X i (t) represents the spatial position of the i-th coot individual in the d-dimensional individual space at the t-th iteration, i = 1, 2,..., l; Y i (t) is the optimal position of the coot individual i at the t-th iteration. At the (t + 1)-th iteration, the position update formula of the i-th coot individual is as follows:

[0170] X i (t + 1) = X i (t) × |sin(G 1 )| + G 2 × sin(G 1 ) × |μ × Y i (t) - ν × X i (t)| (17)

[0171] In the formula, G 1 is a random number in [0, 2π], which determines the moving distance of the i-th coot individual in the next iteration process; G 2 is a random number in [0, π], which determines the position update direction of the i-th coot individual in the next iteration process; μ and ν are coefficients obtained by introducing the golden section number. The values of the golden section numbers μ and ν are calculated as follows:

[0172] μ = -π + (1 - π) × 2π (18)

[0173] ν = -π + λ × 2π (19)

[0174] After introducing the golden sine strategy, in the random movement stage, the new position update formula is as follows:

[0175] X C (i) = X i (t + 1) = X i (t) × |sin(G 1 )| + G 2× sin(G 1 ) × |μ × Y i (t) - ν × X i (t)| (20)

[0176] Secondly, the present invention adds a greedy strategy to the process of selecting the movement mode of individual coots, so as to reduce the blindness in the movement process of individual coots. The main way to add the greedy strategy to the process of selecting the movement of individual coots is as follows: first, determine whether the position after the previous iterative movement has been improved, and then determine which movement mode to select in this iteration. If the position after the previous movement of the individual coot has been improved, then the movement mode selection in this movement process is the same as the previous one; if the position after the previous movement of the individual coot has not been improved but has become worse, then the movement mode selection in this movement process is different from the previous one to obtain a better movement effect. The rule formula is as follows:

[0177]

[0178] In the formula, Move(i) t is the movement mode selected by the coot at the t-th movement; ~ represents the NOT operation; random represents random selection; g(i) t is the fitness value of the t-th movement of the coot.

[0179] Step S104: Introduce a perturbation strategy based on virtual force to update the position of the leader to coordinate its local search and global optimization capabilities.

[0180] During the entire optimization process, the position of the leader is related to the trend of the entire population. To find the optimal position, the leader must jump out of the existing local optimal position to find the optimal position, and use the following formula to complete the update of the leader's position.

[0181]

[0182] In the formula, X L (i) represents the current position of the leader B 4 and B 5 is a random number between [0, 1]; B 3 is a random number between [-1, 1]; X best is the best position that can be found; D is obtained from the following formula:

[0183]

[0184] The basic coot optimization algorithm needs to select the optimal position by calculating the fitness after each iteration. If the position update formula is not perturbed, the algorithm is prone to falling into a local optimum. In the present invention, a virtual force is added during the position update process of the coot flock leader, which can change the search area of the leader and balance the capabilities of local search and global optimization, thereby obtaining the optimal search result. The virtual force is mainly generated by three parts: adjacent communication nodes, uncovered grid points, and regional boundaries. Under the influence of the virtual force, the mobile communication nodes in the monitoring area may be repositioned. After each perturbation of the virtual force, the communication nodes will eventually move to the optimal position along the direction of the resultant force. The perturbation factor of the virtual force is calculated by the following formula:

[0185]

[0186] In the formula, f i is the moving distance of the node under the action of the virtual force. F i is the resultant virtual force applied to the communication node, F ij is the virtual force of adjacent communication nodes, F ik is the virtual attraction of uncovered grid points, F ib is the boundary virtual repulsion. Step(t) is the single virtual moving step length of the node.

[0187] To avoid the oscillation of the coverage rate in the later stage of iteration, the virtual moving step length is set to gradually decrease as the number of iterations t increases, as shown in the following formula:

[0188]

[0189] This ensures the stability of the algorithm and reduces the ineffective movement of the nodes. Stepmax and Stepmin represent the maximum and minimum moving step lengths.

[0190] The improved leader motion formula is as follows:

[0191]

[0192] In the formula, f i is the moving distance of the node under the action of the virtual force.

[0193] Step S105: Use the COOT algorithm to optimize the position of the sink node and output the optimal node position.

[0194] After establishing the coverage rate model of the three-dimensional trusted WLAN network nodes, using F cov as the objective function, use an optimized coot algorithm-based method to calculate and solve the objective function. Using F COVLet \(f\) be the objective function. The larger the value of the objective function, the larger the coverage area of the node deployment and the more reasonable the node layout position.

[0195] For the problem of coverage optimization of three-dimensional trusted WLAN network communication nodes in a substation scenario, the present invention proposes a method for fusing and predicting the coverage area of nodes based on ray tracing and machine learning. The DNN neural network is trained by using the 3D position information of the nodes and the path loss deviation value between the ray tracing simulation data and the measured channel data, so as to accurately predict the path loss to calculate the received power of the target node, and then a prediction model for the node coverage area applicable to the substation scenario is constructed. At the same time, the present invention also proposes an improved coot algorithm (COOT) to guide the deployment scheme of trusted WLAN network communication nodes in the substation with the coverage rate as the optimization target. Aiming at the problems such as low degree of uniformity of the initial population distribution in the traditional coot algorithm, the present invention introduces the Sobol sequence for population initialization to enrich the diversity of the population, and introduces the golden sine strategy in the random movement of individuals to accelerate the convergence speed of the algorithm. Secondly, a greedy strategy is added to the selection of the movement mode of the followers to reduce the blindness in the movement process. Finally, a virtual force perturbation strategy is introduced in the algorithm iteration process to perturb and update the position of the leader, so as to coordinate the local search and global optimization capabilities, and thus obtain the optimal search result. The present invention can ensure full coverage of trusted WLAN network communication nodes in the complex environment of the substation while reducing the repeated coverage rate of the nodes, and effectively solves the problems of weak coverage and blind coverage of nodes in the complex environment.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that it is still possible to modify the specific implementation manners of the present invention or make equivalent replacements without departing from the spirit and scope of the present invention. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for optimizing the coverage of a substation trusted WLAN network, characterized in that: The steps include: Step S1, constructing a three-dimensional trusted WLAN network node coverage model based on a node coverage prediction method fused with ray tracing and machine learning; Step S2, using the Sobol sequence to initialize the COOT algorithm population to enhance population diversity; Step S3, introducing the golden sine strategy in the random motion position update of the COOT algorithm, and introducing the greedy strategy in the random selection of individual motion modes; Step S4, during the optimization process of the COOT algorithm, a disturbance strategy based on virtual force is introduced to perform disturbance update on the leader position; Step S5, using the optimized COOT algorithm to calculate and solve the objective function of the three-dimensional trusted WLAN network node coverage model to obtain the optimal node position, the objective function is: In the formula, F cov is the total coverage rate of the trusted WLAN network to all the communication nodes in the target area, m is the number of small square areas into which the target area is divided, and V k is the volume of the small block, V is the total volume of the target area, P(U,S j ) is the joint communication probability distribution of the trusted WLAN network communication nodes.

2. The method for optimizing the trusted WLAN network coverage of a substation according to claim 1, characterized in that: The step S1 comprises: Based on the actual scenario of the substation, the channel measurement system is used to obtain channel measurement data; Build a three-dimensional substation simulation model based on actual scenarios and randomly arrange communication nodes in the target area; The received power of the target node is calculated using the ray tracing method; The path loss deviation value is calculated using the measured path loss value of the channel and the simulated path loss value obtained based on ray tracing; The DNN neural network is trained using the path loss deviation values ​​at different receiving end locations and the 3D position information of the communication node and the target node to accurately predict the path loss deviation value between any communication node and the target node in the target area. The simulated path loss value is corrected using the path loss deviation value predicted by the neural network to obtain the optimized path loss value, and the accurate received power at the corresponding target node is calculated; The effective coverage range of the communication node is determined according to the received power of the target node and the set received power threshold, and a coverage model of the trusted WLAN network communication node is established.

3. The method for optimizing the trusted WLAN network coverage of a substation according to claim 1, characterized in that: The Sobol sequence is used to generate sample points that are approximately evenly distributed in a multidimensional space. A low-deviation sequence is generated by a quasi-Monte Carlo method to initialize the white-bone bird population. The position of the white-bone bird population is: X C (i)=S N (ub-lb)+lb Where [ub,lb] is the search range of the target space, and the random number S is N ∈[0,1],X C (i) is the position of the i-th white-bone bird, which is used to enhance the diversity of the initial population distribution in the target space.

4. The method for optimizing the trusted WLAN network coverage of a substation according to claim 1, characterized in that: The golden sine strategy is used to introduce a sine function to dynamically adjust the search step size during the algorithm iteration process, thereby improving the convergence speed and local search capability of the algorithm. The greedy strategy is used to retain the current optimal solution during the local search phase and reduce blindness during the movement process.

5. The method for optimizing the trusted WLAN network coverage of a substation according to claim 1, characterized in that: The virtual force-based perturbation strategy includes three virtual forces, under the action of which the position of the leader is perturbed and updated to balance the capabilities of local search and global optimization.

6. The method for optimizing the trusted WLAN network coverage of a substation according to claim 1, characterized in that: The larger the value of the objective function is, the larger the coverage of the node deployment is and the more reasonable the node layout is.

7. The method for optimizing the trusted WLAN network coverage of a substation according to any one of claims 1 to 6, characterized in that: The step S3 comprises: Randomly generate the positions of l white-bone birds. Assume that each solution of the optimization problem corresponds to the position of the white-bone bird in the search space. i (t) represents the spatial position of the i-th white-bone bird individual in the t-th iteration in the d-dimensional individual space, i = 1, 2, ..., l; Y i (t) is the optimal position of the i-th white-bone bird in the t-th iteration. In the t+1 iteration, the position update formula of the i-th white-bone bird is as follows: X i (t+1)=X i (t)×|sin(G1)|+G2×sin(G1)×|μ×Y i (t)-ν×X i (t)| Where G1 is a random number in [0,2π], which determines the moving distance of the i-th white-bone bird in the next iteration; G2 is a random number in [0,π], which determines the position update direction of the i-th white-bone bird in the next iteration; μ and ν are coefficients obtained by introducing the golden section number. The values ​​of μ and ν are calculated as follows: μ=-π+(1-π)×2π ν=-π+λ×2π After the introduction of the golden sine strategy, in the random motion phase, the new position update formula is as follows: X C (i)=X i (t+1)=X i (t)×|sin(G1)|+G2×sin(G1)×|μ×Y i (t)-ν×X i (t)| First, we compare whether the position after the previous iteration has been improved. If the position of the white-bone bird after the previous movement has been improved, the movement mode of this movement process is the same as the previous one; if the position of the white-bone bird after the previous movement has not been improved but has become worse, the movement mode of this movement process is different from the previous one to obtain a better movement effect. The rule formula is as follows: Where, Move(i) t is the movement mode chosen by the white-bone bird during its t-th movement; ~ represents a negation operation; random represents a random selection; g(i) t is the fitness value of the t-th movement of the white-bone bird.

8. The method for optimizing the trusted WLAN network coverage of a substation according to any one of claims 1 to 6, characterized in that: The step S4 comprises: In the process of updating the position of the leader of the white-crowned bird flock, a virtual force is added to change the leader's search area. The virtual force is generated by three parts: adjacent communication nodes, uncovered grid points, and area boundaries. The perturbation factor of the virtual force is calculated by the following formula to perturb the leader's position update formula: In the formula, f i is the distance the node moves under the action of the virtual force, F i is the resultant virtual force applied on the communication node, F ij is the virtual force of the adjacent communication nodes, F ik is the virtual attraction of uncovered grid points, F ib is the virtual repulsive force of the boundary, Step(t) is the single virtual moving step of the node; The virtual moving step size is set to gradually decrease as the number of iterations increases, as shown in the following formula: In the formula, Stepmax and Stepmin represent the maximum and minimum moving step lengths respectively, to ensure the stability of the algorithm and reduce the invalid movement of nodes; The leader position is updated using the improved leader motion formula, which is as follows: In the formula, f i is the moving distance of the node under the action of virtual force, X L (i) indicates that the current leader positions B4 and B5 are random numbers between the interval [0,1]; B3 is a random number between the interval [-1,1]; X best is the best position that can be found, max_Iter is the maximum number of iterations, and t is the current number of iterations.

9. A device for optimizing the coverage of a substation trusted WLAN network, characterized in that: include: A model building module is used to build a three-dimensional trusted WLAN network node coverage model based on a node coverage prediction method that integrates ray tracing and machine learning; The population initialization module is used to initialize the COOT algorithm population using the Sobol sequence to enhance population diversity; Strategy introduction module, used to introduce the golden sine strategy in the random motion position update of the COOT algorithm, and introduce the greedy strategy in the random selection of individual motion modes; The perturbation update module is used to introduce a perturbation strategy based on virtual force to perturb the leader position during the COOT algorithm optimization process; The objective function solving module is used to calculate and solve the objective function of the three-dimensional trusted WLAN network node coverage model using the optimized COOT algorithm to obtain the optimal node position.

10. The device for optimizing the trusted WLAN network coverage of a substation according to claim 9, characterized in that: The objective function is: In the formula, F cov is the total coverage rate of the trusted WLAN network to all the communication nodes in the target area, m is the number of small square areas into which the target area is divided, and V k is the volume of the small block, V is the total volume of the target area, P(U,S j ) is the joint communication probability distribution of the trusted WLAN network communication nodes.

Images