A method, equipment and medium for evaluating the performance of a power inspection network

Abstract

This invention relates to a method, device, and medium for evaluating the performance of a power inspection network. The method includes the following steps: dividing the power inspection network into an access network, a transmission network, and a core network; monitoring the network performance parameters of the access network, transmission network, and core network using a communication equipment performance monitoring module; inputting the preprocessed network performance parameters into a BP neural network model with optimized parameters using a genetic algorithm, and outputting a performance evaluation value within a preset numerical range. The fitness function of the genetic algorithm is obtained by weighted summation of model prediction accuracy and network resilience value based on a scenario adjustment factor. The network resilience value is determined based on fault recovery time and service degradation. Compared with existing technologies, this invention can ensure the accuracy of network performance evaluation and match the resilience requirements of the network under different application scenarios, possessing advantages such as accurate prediction and strong adaptability to different scenarios.

Description

Technical Field

[0001] This invention relates to the field of power communication networks, and in particular to a method, equipment, and medium for evaluating the performance of power inspection networks. Background Technology

[0002] Power grid inspection, as a core component ensuring the safe and stable operation of the power grid, is gradually transforming towards intelligent and digital transformation. In recent years, intelligent inspection technologies, centered on drones, intelligent robots, and IoT sensors, have been widely applied. These technologies place higher demands on the real-time performance, reliability, and coverage of communication networks. To meet the diverse needs of complex power scenarios, power inspection networks are gradually adopting a hybrid networking mode, integrating various heterogeneous communication technologies such as 5G, wired networks, and private wireless networks, combined with industrial control networks and IoT technology, to build a ubiquitous interconnected communication architecture.

[0003] However, network performance in a hybrid networking environment is affected by factors such as multi-technology collaboration mechanisms, dynamic topology changes, and heterogeneous protocol compatibility. Potential issues include bandwidth fluctuations, latency jitter, and data packet loss, which may directly impact the data transmission quality of inspection equipment, the accuracy of remote control, and the timeliness of fault early warning, thereby threatening the safe operation of the power system. Therefore, network performance evaluation is necessary.

[0004] Chinese patent CN108400895A discloses a security situation assessment algorithm based on a genetic algorithm-improved BP neural network. This algorithm constructs a network security situation assessment model and leverages the powerful self-learning ability of neural networks to apply BP neural networks to network security situation assessment. Addressing the inherent limitations of neural network algorithms, such as susceptibility to local minima and slow convergence, a genetic algorithm is introduced to optimize the weights of the BP neural network, accelerating its convergence and improving its accuracy and efficiency in network security situation assessment. This effectively solves the problems of low efficiency and uncertainty in network security situation assessment using simple neural networks. However, this method only assesses security situation and cannot cover multiple aspects of network performance assessment, especially lacking assessment of network resilience, making it difficult to cope with dynamic fault scenarios. Furthermore, the method does not specifically divide the network into layers, thus failing to determine corresponding differentiated performance indicators for each layer, affecting the effectiveness of the prediction results.

[0005] Therefore, there is currently a lack of a network performance evaluation method that can distinguish different levels of power inspection networks and specifically consider the characteristics and network resilience of different levels. Summary of the Invention

[0006] The purpose of this invention is to overcome the defects of the prior art by providing a method, equipment and medium for evaluating the performance of power inspection networks.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] According to a first aspect of the present invention, a method for evaluating the performance of a power inspection network is provided, the method comprising the following steps:

[0009] The power inspection network is divided into access network, transmission network, and core network.

[0010] The network performance parameters of the access network, transmission network, and core network are monitored using a communication equipment performance monitoring module.

[0011] After preprocessing the network performance parameters, the data is input into a BP neural network model with optimized parameters using a genetic algorithm. The model outputs a performance evaluation value within a preset range. The fitness function of the genetic algorithm is obtained by weighted summation of the model prediction accuracy and network resilience value based on a scenario adjustment factor. The network resilience value is determined based on the fault recovery time index and service degradation degree.

[0012] The access network includes a communication network, a wireless private network, and a 4G / 5G virtual private network, used to enable secure and reliable access for communication terminals; the transmission network is used to transmit access layer data to the core network; the core network includes routers and an enterprise middleware platform, wherein the routers are used to receive terminal data from the transmission network and send the data to the enterprise middleware platform, and the enterprise middleware platform is used to provide the data to the upper-layer business system for analysis, and is also responsible for management and issuing instructions to terminal devices.

[0013] The network performance monitoring parameters of the access network are divided into local communication layer network performance monitoring parameters and remote communication layer network performance monitoring parameters, among which,

[0014] The local communication layer network performance monitoring parameters include wired access network performance monitoring parameters and wireless access network performance monitoring parameters. The wired access network performance monitoring parameters include port bandwidth utilization, xPON received optical power, and primary / backup switchover time. The wireless access network performance monitoring parameters include air interface link establishment rate, channel utilization, number of AP associated terminals, throughput, latency, packet loss rate, concurrency rate, and dual redundancy switchover time.

[0015] The network performance monitoring parameters for the remote communication layer include throughput, latency, packet loss rate, jitter, primary / backup switchover time, and routing convergence time.

[0016] The network performance monitoring parameters of the transmission network include transmitted optical power, received optical power, optical signal-to-noise ratio, bit error rate, transmission delay, delay jitter, and SDH protection switching time.

[0017] The network performance monitoring parameters of the core network include throughput, latency, latency jitter, ECMP error, CPU utilization, memory utilization, table capacity utilization, neighbor status, and route convergence time.

[0018] The fitness function of the genetic algorithm is expressed as:

[0019] ,

[0020] in, This is a scenario adjustment factor, determined based on the requirements of different network scenarios for prediction accuracy and resilience; The prediction error is for normal samples, which are network feature vectors after preprocessing of network performance parameters. For the fault recovery time index of virtual fault samples, The degree of service degradation for virtual fault samples. , , Standard fault recovery time, The fault recovery time for the virtual fault sample. For standard service rates, The service rate during a virtual fault sample failure; the method for generating the virtual fault sample is as follows: based on normal samples... ,in, To determine the number of network performance monitoring parameters, virtual fault samples are generated using the neighborhood perturbation operator. , , For network performance monitoring parameters The fault sensitivity coefficient, For fault intensity, It is a random perturbation.

[0021] The genetic algorithm performs the following steps to optimize the parameters of the BP neural network model:

[0022] The weight parameters and bias parameters in the BP neural network model are encoded with real numbers and mapped to chromosomes in the genetic algorithm.

[0023] Perform population initialization and define the fitness function;

[0024] Genetic operations are performed to update the population, including crossover, mutation, and selection. Specifically, the crossover probability is adaptively determined based on the coding of regulatory genes, and the crossover operation is performed. The coding of regulatory genes is determined based on the activity value defined by the derivative of the gene in the chromosome using the loss function of a BP neural network. A mutation operation is performed on the original gene based on the mutation intensity, which is determined based on the gene's contribution to the virtual fault samples. A selection operation is performed based on the individual fitness ranking.

[0025] The iteration terminates when the number of iterations reaches a preset value or the fitness change of the best individual in consecutive preset generations is less than a preset threshold, and the optimal chromosome is output to obtain the optimized model parameters. Otherwise, the process returns to perform genetic operations to update the population.

[0026] The crossover operation includes the following steps:

[0027] Calculate the first The activity value of each gene :

[0028] ,

[0029] in, Let be the loss function of the BP neural network on normal samples. Indicates the first The parameter values ​​of each gene, where M is the total number of genes, i.e., the total number of parameters to be optimized in the BP neural network model;

[0030] Based on the activity value, the coding of the regulatory gene is determined:

[0031] ,

[0032] in, The preset threshold, For the first The regulatory gene encoding each gene;

[0033] The crossover probability is determined based on the coding of the aforementioned regulatory genes:

[0034] ,

[0035] in, , For the preset weights, satisfy and ; Representing the father generation and father generation The regulatory gene encoding, For father and father generation The The crossover probability of a gene undergoing a crossover operation;

[0036] The mutation operation includes the following steps:

[0037] For the The genes are randomly perturbed, and the rate of change of the virtual fault sample loss function is calculated to obtain the th gene. The contribution of each gene to the virtual fault sample:

[0038] ,

[0039] in, This represents the loss function of the BP neural network on virtual fault samples. This represents a random perturbation of the gene. This is a preset value used to prevent the denominator from being 0;

[0040] Based on the contribution, the mutation intensity is calculated, and the original gene is mutated.

[0041] ,

[0042] in, This refers to the gene after the mutation operation. The strength of the variation is determined by the contribution. This is obtained by performing segmented mapping.

[0043] According to a second aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described thereon.

[0044] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) The present invention divides the network into access network, transmission network and core network, and monitors the performance indicators of the corresponding level respectively. It can monitor the performance parameters of different components in the network in a targeted manner, thereby improving the accuracy and effectiveness of subsequent performance evaluation.

[0047] (2) Existing fitness functions often use prediction error as the sole evaluation criterion, which can easily lead to the misselection of parameter combinations that have high prediction accuracy but cannot match the actual resilience requirements of the network. The fitness function setting of this invention can ensure that the parameters of the BP neural network optimized by the genetic algorithm can not only guarantee the accuracy of network performance (such as latency and packet loss rate) prediction, but also match the resilience requirements of the network under fault conditions, making the optimized BP neural network model more suitable for actual network operation and maintenance scenarios. Furthermore, by using a dynamically adjustable scenario adjustment factor in the fitness function, the same fitness function can be adapted to different types of network performance evaluation scenarios, improving the overall applicability and robustness of the method.

[0048] (3) This invention improves the crossover and mutation processes of the genetic algorithm. The crossover operation can accurately evaluate the role of parameters in assessing the normal operation status of the power inspection network by measuring the activity value, thereby assigning a higher crossover probability to highly active genes and accelerating the convergence efficiency in normal scenarios. A certain crossover probability is retained for low-activity genes to preserve the possibility of exploring new parameter combinations and avoid getting trapped in local optima too early. The mutation operation measures the influence of a single gene on the assessment accuracy of fault scenarios by measuring the contribution value. The greater the contribution value, the more critical the gene is. Therefore, the mutation intensity after segmentation mapping is also greater. The large mutation intensity corresponding to key genes can explore better parameter combinations more efficiently and quickly improve the assessment accuracy in fault scenarios. The small mutation intensity corresponding to non-key genes avoids meaningless disturbances and retains the optimized parameter characteristics in normal scenarios, thereby improving the overall optimization efficiency of the genetic algorithm. Attached Figure Description

[0049] Figure 1 This is a flowchart of the method of the present invention;

[0050] Figure 2 This is a flowchart of the genetic algorithm of the present invention;

[0051] Figure 3 This is a flowchart of the cross-operation process of the present invention;

[0052] Figure 4 This is a flowchart of the variation operation of the present invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0054] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0055] Example 1

[0056] This embodiment provides a method for evaluating the performance of a power line inspection network, such as... Figure 1 As shown, the method includes the following steps:

[0057] S1 divides the power inspection network into access network, transmission network, and core network.

[0058] In this embodiment, the access network includes a communication network, a wireless private network, and a 4G / 5G virtual private network, used to achieve secure and reliable access for communication terminals; the transmission network is used to transmit access layer data to the core network, mainly using fiber optic transmission technologies such as SDH and OTN; the core network includes routers and an enterprise middleware platform, wherein the routers are used to receive terminal data from the transmission network and send the data to the enterprise middleware platform, and the enterprise middleware platform is used to provide the data to the upper-layer business system for analysis, and is also responsible for management and issuing instructions to terminal devices.

[0059] S2, use the communication equipment performance monitoring module to monitor the network performance parameters of the access network, transmission network and core network.

[0060] Based on the network hierarchy, network performance parameters are divided into access network performance monitoring parameters, transmission network performance monitoring parameters, and core network performance monitoring parameters.

[0061] In a preferred embodiment, the network performance monitoring parameters of the access network are divided into local communication layer network performance monitoring parameters and remote communication layer network performance monitoring parameters, wherein,

[0062] The local communication layer network performance monitoring parameters include wired access network performance monitoring parameters and wireless access network performance monitoring parameters. The wired access network performance monitoring parameters include port bandwidth utilization, xPON received optical power, and primary / backup switchover time. The wireless access network performance monitoring parameters include air interface link establishment rate, channel utilization, number of AP associated terminals, throughput, latency, packet loss rate, concurrency rate, and dual redundancy switchover time.

[0063] The network performance monitoring parameters for the remote communication layer include throughput, latency, packet loss rate, jitter, primary / backup switchover time, and routing convergence time.

[0064] Network performance monitoring parameters for transmission networks include transmitted optical power, received optical power, optical signal-to-noise ratio, bit error rate, transmission delay, delay jitter, and SDH protection switching time.

[0065] The network performance monitoring parameters of the core network include throughput, latency, latency jitter, ECMP error, CPU utilization, memory utilization, table capacity utilization, neighbor status, and route convergence time.

[0066] S3 preprocesses the network performance parameters and inputs them into the BP neural network model with optimized parameters using a genetic algorithm, outputting performance evaluation values ​​within a preset range.

[0067] First, the network topology of the BP neural network model is designed; this invention does not limit its specific structure. In one preferred embodiment, the BP neural network model employs multiple hidden layers, with the initial number of nodes in the hidden layers determined by an empirical formula. The activation function of the output layer is Linear, outputting continuous performance evaluation scores, and the loss function is measured by Mean Squared Error (MSE). In another embodiment, the output layer of the model can also use a softmax activation function to output different classification levels of network performance; this can be set according to the needs of the actual application scenario.

[0068] For the BP neural network model, the first Layer There are neurons, and their weighted input values ​​are:

[0069] ,

[0070] Its activation output value is:

[0071] ,

[0072] in, For the first Layer The neuron connects to the previous layer. The weights between neurons; For the first Layer Bias of each neuron; For activation functions; For the first Layer The output of each neuron.

[0073] During model training, forward propagation begins, calculating the predicted output through the hidden and output layers. The error is then calculated by comparing this predicted output with the actual target output. In the backpropagation phase, based on the error, gradient descent is used to propagate the error layer by layer from the output layer to the hidden layers, continuously adjusting the weights and biases to reduce the error. This process involves the application of the chain rule, updating the weights and biases of each neuron by calculating gradients.

[0074] like Figure 2 As shown, the genetic algorithm performs the following steps to optimize the parameters of the BP neural network model:

[0075] Step 1) Encode the weight parameters and bias parameters in the BP neural network model with real numbers and map them to chromosomes in the genetic algorithm.

[0076] Step 2) Initialize the population and define the fitness function.

[0077] In this embodiment, the fitness function of the genetic algorithm is obtained by weighting and summing the model prediction accuracy and network resilience value based on the scene adjustment factor. The network resilience value is determined based on the fault recovery time index and service degradation degree.

[0078] ,

[0079] in, This is a scenario-adjusting factor, determined based on the different network scenarios' requirements for prediction accuracy and resilience. For example, in industrial control network scenarios, it's necessary to reduce the weighting of prediction accuracy and increase the weighting of resilience. For civilian broadband network scenarios, it is necessary to increase the weighting of prediction accuracy and decrease the weighting of resilience. ; The prediction error is for normal samples, which are network feature vectors after preprocessing of network performance parameters. For the fault recovery time index of virtual fault samples, The degree of service degradation for virtual fault samples. , , Standard fault recovery time, The fault recovery time for the virtual fault sample. For standard service rates, The service rate is the value when a virtual fault sample fails. This invention quantifies fault recovery time and service degradation as two indicators, which are then incorporated into the fitness function to improve the consideration of network resilience in the parameter optimization results.

[0080] In this embodiment, the method for generating virtual fault samples is as follows: for normal samples ,in, To determine the number of network performance monitoring parameters, virtual fault samples are generated using the neighborhood perturbation operator. , , For network performance monitoring parameters The fault sensitivity coefficient, The fault intensity is represented by a value of [0.2, 0.5]. To obey N Random perturbations with a (0,1) distribution.

[0081] Step 3) Perform genetic operations to update the population, including crossover, mutation, and selection.

[0082] Step 4) When the number of iterations reaches the preset value or the fitness change of the best individual in consecutive preset generations is less than the preset threshold, the iteration is terminated, the optimal chromosome is output, and the optimized model parameters are obtained. Otherwise, return to perform genetic operations to update the population.

[0083] Example 2

[0084] This embodiment, based on Embodiment 1, provides an improved scheme for crossover and mutation operations in a genetic algorithm. Specifically, the crossover probability is adaptively determined based on the coding of regulatory genes, and the crossover operation is performed. The coding of regulatory genes is determined based on the activity value defined by the derivative of the loss function of the gene in the chromosome using a BP neural network. The mutation operation is performed on the original gene based on the mutation intensity, which is determined based on the contribution of the gene to the virtual fault sample. The selection operation is performed based on the fitness ranking of individuals.

[0085] like Figure 3 As shown, the crossover operation includes the following steps:

[0086] A1, calculate the... The activity value of each gene :

[0087] ,

[0088] in, Let be the loss function of the BP neural network on normal samples. Indicates the first The parameter values ​​of each gene, where M is the total number of genes, i.e., the total number of parameters to be optimized in the BP neural network model;

[0089] A2, Based on the activity value, determine the encoding of the regulatory gene:

[0090] ,

[0091] in, The preset threshold, For the first The regulatory gene encoding each gene;

[0092] A3, Determine the crossover probability based on the coding of the aforementioned regulatory genes:

[0093] ,

[0094] in, , For the preset weights, satisfy and ; Representing the father generation and father generation The regulatory gene encoding, For father and father generation The The crossover probability of a gene undergoing a crossover operation.

[0095] like Figure 4 As shown, the mutation operation includes the following steps:

[0096] B1, for the first The genes are randomly perturbed, and the rate of change of the virtual fault sample loss function is calculated to obtain the th gene. The contribution of each gene to the virtual fault sample:

[0097] ,

[0098] in, This represents the loss function of the BP neural network on virtual fault samples. This represents a random perturbation of the gene. This is a preset value used to prevent the denominator from being 0;

[0099] B2, Calculate the mutation intensity based on the contribution, and perform mutation operations on the original gene:

[0100] ,

[0101] in, This refers to the gene after the mutation operation. The strength of the variation is determined by the contribution. This is obtained by performing segmented mapping.

[0102] Example 3

[0103] This embodiment provides a method for preprocessing network performance parameters, based on embodiment 1.

[0104] This step involves decomposing the network performance parameters related to decision-making into multiple levels, then performing quantitative and qualitative analysis on them, using the analytic hierarchy process (AHP) to determine the weight of each parameter, and finally multiplying the weights by the initial network performance parameters as the input to the BP neural network model.

[0105] Further steps include:

[0106] 1) For the same level indicators in the three network levels, perform pairwise comparisons to determine their relative importance. The specific determination method is shown in Table 1.

[0107] Table 1

[0108]

[0109] 2) Construct a judgment matrix based on the determined relative importance. :

[0110] ,

[0111] Among them, the judgment matrix Elements in the middle have the following properties:

[0112] .

[0113] 3) Based on the judgment matrix ,get The largest eigenvalue and its corresponding eigenvectors :

[0114] = ,

[0115] For eigenvectors After normalization, the resulting values ​​are the ranking weights of the relative importance of each network performance indicator within the same level.

[0116] 4) By calculating the consistency index Consistency ratio Check the consistency of the judgment matrix to ensure that the judgment results are reasonable.

[0117] ,

[0118] In the formula, To determine the matrix The order of;

[0119] ,

[0120] In the formula, The random consistency index is an average consistency index calculated from a large number of random matrices, and its values ​​are shown in Table 2 below.

[0121] Table 2

[0122]

[0123] like If the value is less than 0.1, the matrix is ​​considered to have acceptable consistency.

[0124] Example 4

[0125] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0126] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0127] The processing unit executes the various methods and processes described above, such as methods S1 to S3. For example, in some embodiments, methods S1 to S3 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S3 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S3 by any other suitable means (e.g., by means of firmware).

[0128] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0129] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0130] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0131] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for evaluating the performance of a power line inspection network, characterized in that, The method includes the following steps: The power inspection network is divided into access network, transmission network, and core network. The network performance parameters of the access network, transmission network, and core network are monitored using a communication equipment performance monitoring module. After preprocessing the network performance parameters, the data is input into a BP neural network model with optimized parameters using a genetic algorithm. The model outputs a performance evaluation value within a preset range. The fitness function of the genetic algorithm is obtained by weighted summation of the model prediction accuracy and network resilience value based on a scenario adjustment factor. The network resilience value is determined based on the fault recovery time index and the service degradation degree. The fitness function of the genetic algorithm is expressed as: ， in, This is a scenario adjustment factor, determined based on the requirements of different network scenarios for prediction accuracy and resilience; The prediction error is for normal samples, which are network feature vectors after preprocessing of network performance parameters. For the fault recovery time index of virtual fault samples, The degree of service degradation for virtual fault samples. , , Standard fault recovery time, The fault recovery time for the virtual fault sample. For standard service rates, The service rate during a virtual fault sample failure; the method for generating the virtual fault sample is as follows: based on normal samples... ,in, To determine the number of network performance monitoring parameters, virtual fault samples are generated using the neighborhood perturbation operator. , , For network performance monitoring parameters The fault sensitivity coefficient, For fault intensity, For random perturbations; The genetic algorithm performs the following steps to optimize the parameters of the BP neural network model: The weight parameters and bias parameters in the BP neural network model are encoded with real numbers and mapped to chromosomes in the genetic algorithm. Perform population initialization and define the fitness function; Genetic operations are performed to update the population, including crossover, mutation, and selection. Specifically, the crossover probability is adaptively determined based on the coding of regulatory genes, and the crossover operation is performed. The coding of regulatory genes is determined based on the activity value defined by the derivative of the gene in the chromosome using the loss function of a BP neural network. A mutation operation is performed on the original gene based on the mutation intensity, which is determined based on the gene's contribution to the virtual fault samples. A selection operation is performed based on the individual fitness ranking. The iteration terminates when the number of iterations reaches a preset value or the fitness change of the best individual in consecutive preset generations is less than a preset threshold, and the optimal chromosome is output to obtain the optimized model parameters. Otherwise, the process returns to perform genetic operations to update the population.

2. The power inspection network performance evaluation method according to claim 1, characterized in that, The access network includes a communication network, a wireless private network, and a 4G / 5G virtual private network, used to enable secure and reliable access for communication terminals; the transmission network is used to transmit access layer data to the core network; the core network includes routers and an enterprise middleware platform, wherein the routers are used to receive terminal data from the transmission network and send the data to the enterprise middleware platform, and the enterprise middleware platform is used to provide the data to the upper-layer business system for analysis, and is also responsible for management and issuing instructions to terminal devices.

3. The power inspection network performance evaluation method according to claim 1, characterized in that, The network performance monitoring parameters of the access network are divided into local communication layer network performance monitoring parameters and remote communication layer network performance monitoring parameters, among which, The local communication layer network performance monitoring parameters include wired access network performance monitoring parameters and wireless access network performance monitoring parameters. The wired access network performance monitoring parameters include port bandwidth utilization, xPON received optical power, and primary / backup switchover time. The wireless access network performance monitoring parameters include air interface link establishment rate, channel utilization, number of AP associated terminals, throughput, latency, packet loss rate, concurrency rate, and dual redundancy switchover time. The network performance monitoring parameters for the remote communication layer include throughput, latency, packet loss rate, jitter, primary / backup switchover time, and routing convergence time.

4. The power inspection network performance evaluation method according to claim 1, characterized in that, The network performance monitoring parameters of the transmission network include transmitted optical power, received optical power, optical signal-to-noise ratio, bit error rate, transmission delay, delay jitter, and SDH protection switching time.

5. The power inspection network performance evaluation method according to claim 1, characterized in that, The network performance monitoring parameters of the core network include throughput, latency, latency jitter, ECMP error, CPU utilization, memory utilization, table capacity utilization, neighbor status, and route convergence time.

6. The power inspection network performance evaluation method according to claim 1, characterized in that, The crossover operation includes the following steps: Calculate the first The activity value of each gene : ， in, Let be the loss function of the BP neural network on normal samples. Indicates the first The parameter values ​​of each gene, where M is the total number of genes, i.e., the total number of parameters to be optimized in the BP neural network model; Based on the activity value, the coding of the regulatory gene is determined: ， in, The preset threshold, For the first The regulatory gene encoding each gene; The crossover probability is determined based on the coding of the aforementioned regulatory genes: ， in, , For the preset weights, satisfy and ; Representing the father generation and father generation The regulatory gene encoding, For father and father generation The The crossover probability of a gene undergoing a crossover operation; The mutation operation includes the following steps: For the first The genes are randomly perturbed, and the rate of change of the virtual fault sample loss function is calculated to obtain the th gene. The contribution of each gene to the virtual fault sample: ， in, This represents the loss function of the BP neural network on virtual fault samples. This represents a random perturbation of the gene. This is a preset value used to prevent the denominator from being 0; Based on the contribution, the mutation intensity is calculated, and the original gene is mutated. ， in, This refers to the gene after the mutation operation. The strength of the variation is determined by the contribution. This is obtained by performing segmented mapping.

7. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 6.

Images