Method and system for improving reliability of optical bypass protection function of distribution network switch

Abstract

The invention discloses a method and a system for improving the reliability of an optical bypass protection function of a distribution network switch, and relates to the technical field of power distribution network communication. Identifying key environmental factors of electromagnetic interference and signal attenuation under each frequency band according to the power distribution network data; calculating the frequency spectrum quality of each frequency band; dividing the power distribution network as a small-sized power distribution network or a non-small-sized power distribution network; for a small power distribution network, according to the spectrum quality, a genetic algorithm is adopted to carry out dynamic spectrum allocation by taking spectrum resource optimization of a global optical bypass protection function as a target; for a non-small power distribution network, according to the frequency spectrum quality, reinforcement learning is adopted to carry out dynamic frequency spectrum allocation on each optical bypass protection function. According to the embodiment of the invention, the method can be effectively applied to an extremely complex electromagnetic environment, guarantees that a centralized or distributed optical bypass protection function always carries out data transmission on the optimal frequency band, and improves the reliability of the optical bypass protection function.

Description

Technical Field

[0001] This invention relates to the field of power distribution network communication technology, and in particular to a method and system for improving the reliability of the optical bypass protection function of a distribution network switch. Background Technology

[0002] In power communication networks, optical bypass protection is a key technology to ensure the stability and security of data transmission. As the power distribution network environment becomes more complex, its reliability is directly related to the operation of the power system, data transmission efficiency, and security.

[0003] To enhance the reliability of optical bypass protection, existing technologies employ various strategies. To resist electromagnetic interference, physical isolation technology is used to reduce the impact of external interference on data transmission; to ensure data transmission continuity, redundant link design is used to achieve automatic switching to backup links in the event of a primary link failure; and to address signal attenuation issues, filtering techniques, power regulation, and other algorithmic optimization methods are employed to improve signal quality.

[0004] However, the above strategy has limitations. When faced with high-intensity electromagnetic interference and severe signal attenuation, it is difficult to guarantee the reliability of the optical bypass protection function. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for improving the reliability of optical bypass protection function of distribution network switches. This method and system can be effectively applied in extremely complex electromagnetic environments to ensure that centralized or distributed optical bypass protection function always transmits data in the optimal frequency band, thereby improving the reliability of optical bypass protection function.

[0006] To address the above technical problems, embodiments of the present invention provide a method for improving the reliability of the optical bypass protection function of a distribution network switch, comprising:

[0007] The scale of the power distribution network is obtained, and power distribution network data is collected in real time; the power distribution network data includes environmental data, electromagnetic interference intensity and signal attenuation data of each frequency band;

[0008] Based on the power distribution network data, identify the key environmental factors causing electromagnetic interference and signal attenuation in each frequency band;

[0009] Calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors.

[0010] Based on the scale of the power distribution network, the power distribution network is classified as a small power distribution network or a non-small power distribution network;

[0011] For the small power distribution network, based on the spectrum quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection.

[0012] For the aforementioned non-small-scale power distribution network, reinforcement learning is used to dynamically allocate spectrum for each optical bypass protection function based on the spectrum quality.

[0013] As an improvement to the above solution, the step of obtaining the power distribution network scale and collecting power distribution network data in real time includes:

[0014] Obtain the scale of the power distribution network; the scale of the power distribution network includes the number of topological nodes in the power distribution network;

[0015] Initial power distribution network data is obtained by collecting electromagnetic interference intensity, signal attenuation data, and environmental data at key points in real time through equipment in the power distribution network.

[0016] The initial power distribution network data is preprocessed to obtain environmental data, electromagnetic interference intensity and signal attenuation data for each frequency band, which are then used as power distribution network data.

[0017] As an improvement to the above scheme, the step of identifying key environmental factors of electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data includes:

[0018] Calculate the first correlation between the electromagnetic interference intensity and environmental data in each frequency band to obtain the first impact score of each environmental factor on electromagnetic interference.

[0019] The second correlation between the signal attenuation data and environmental data in each frequency band is calculated to obtain the second impact score of each environmental factor on signal attenuation.

[0020] The environmental factors corresponding to the first influence score being greater than a preset first score threshold or the second influence score being greater than a preset second score threshold are selected to obtain the key environmental factors.

[0021] As an improvement to the above scheme, the step of calculating the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors includes:

[0022] The electromagnetic interference intensity of each frequency band is normalized to obtain the normalized electromagnetic interference value.

[0023] Calculate the ratio of the signal attenuation data to the maximum allowable signal attenuation threshold, and normalize the ratio for each frequency band to obtain the normalized signal attenuation value.

[0024] The normalized electromagnetic interference value, the normalized signal attenuation value, and the key environmental factors are input into a pre-trained neural network model to obtain data transmission quality parameters; the data transmission quality parameters include success rate, signal-to-noise ratio, and bit error rate.

[0025] Based on the data transmission quality parameters, the spectral quality corresponding to each frequency band is obtained.

[0026] As an improvement to the above scheme, the pre-trained neural network model is obtained by training historical power distribution network data and preset scoring rules;

[0027] The preset scoring rules include success rate scores corresponding to different data transmission success rates, signal-to-noise ratio scores corresponding to different signal-to-noise ratios, and bit error rate scores corresponding to different bit error rates.

[0028] As an improvement to the above scheme, for the small power distribution network, based on the spectrum quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection, including:

[0029] If the power distribution network is a small power distribution network, the frequency band of each optical bypass protection function corresponding to the link is encoded into an integer sequence to obtain chromosome encoding;

[0030] Based on the spectral quality and the chromosome encoding, a total fitness function is established with the goal of optimizing the spectral resources for global optical bypass protection.

[0031] Based on the chromosome encoding, K chromosomes are randomly generated to obtain the initial population;

[0032] Crossover selection and mutation selection operations are performed on each chromosome in the initial population to obtain the offspring chromosome with the highest fitness, thus obtaining the optimal spectrum allocation method.

[0033] The optimal spectrum allocation method is used to allocate frequency bands for each optical bypass protection function.

[0034] As an improvement to the above scheme, the step of establishing a total fitness function based on the spectral quality and the chromosome encoding, with the goal of optimizing the spectral resources for global optical bypass protection, includes:

[0035] pass Establish the overall fitness function;

[0036] Where CH = [f1, f2, ..., f N ], f i ∈{1,2,……M},f i For the optical bypass protection function, link i is the link corresponding to the function; N is the total number of links; and M is the total number of available frequency bands. SQI p (i,f i ) represents the spectral quality of link i for each available frequency band; count(j) represents the number of links using frequency band j; P col This is the preset conflict penalty value.

[0037] As an improvement to the above scheme, for the non-small-scale power distribution network, based on the spectrum quality, reinforcement learning is used to dynamically allocate spectrum for each optical bypass protection function, including:

[0038] If the power distribution network is not a small-scale power distribution network, the spectrum quality is classified, and the state space is obtained based on the classification results.

[0039] Based on the available frequency bands, define the frequency band selection action to obtain the action space;

[0040] Calculate the spectral quality corresponding to the frequency band selection action and construct the reward function;

[0041] Based on the state space, the action space, and the reward function, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained by Q-learning.

[0042] The optimal spectrum allocation method is used to allocate frequency bands for each optical bypass protection function.

[0043] As an improvement to the above scheme, the step of using Q-learning to obtain the optimal spectrum allocation method corresponding to each optical bypass protection function based on the state space, the action space, and the reward function includes:

[0044] Each link corresponding to the optical bypass protection function is regarded as an intelligent agent, and the intelligent agent is initialized.

[0045] Based on the initialized agent, select and execute actions in the action space, calculate the reward function corresponding to the action to update the Q value, until the Q value converges to obtain the final action;

[0046] Based on the final action, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained.

[0047] This invention also provides a system for improving the reliability of the optical bypass protection function of a distribution network switch, comprising:

[0048] The key environmental factor identification module is used to identify key environmental factors of electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data.

[0049] The spectrum quality calculation module is used to calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors.

[0050] The power distribution network segmentation module is used to classify the power distribution network into a small power distribution network or a non-small power distribution network according to the scale of the power distribution network.

[0051] The first spectrum allocation module is used to dynamically allocate spectrum for the small power distribution network based on the spectrum quality, using a genetic algorithm with the goal of optimizing the spectrum resources for global optical bypass protection function.

[0052] The second spectrum allocation module is used to dynamically allocate spectrum for each optical bypass protection function based on the spectrum quality for the non-small power distribution network using reinforcement learning.

[0053] Compared with existing technologies, this invention discloses a method and system for improving the reliability of optical bypass protection functions in power distribution network switches. This involves acquiring the scale of the power distribution network and collecting power distribution network data in real time. The power distribution network data includes environmental data, electromagnetic interference intensity in each frequency band, and signal attenuation data. Based on the power distribution network data, key environmental factors affecting electromagnetic interference and signal attenuation in each frequency band are identified. Based on the power distribution network data and the key environmental factors, the spectral quality of each frequency band is calculated. Based on the scale of the power distribution network, it is classified as a small-scale power distribution network or a non-small-scale power distribution network. For small-scale power distribution networks, based on the spectral quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectral resources for global optical bypass protection functions. For non-small-scale power distribution networks, reinforcement learning is used to dynamically allocate spectrum for each optical bypass protection function based on the spectral quality. Using this invention, it can be effectively applied to extremely complex electromagnetic environments, ensuring that centralized or distributed optical bypass protection functions always transmit data in the optimal frequency band, thus improving the reliability of optical bypass protection functions. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the steps of a method for improving the reliability of optical bypass protection function of a distribution network switch according to an embodiment of the present invention.

[0055] Figure 2 This is a schematic diagram of the structure of a system for improving the reliability of optical bypass protection function of a distribution network switch, provided by an embodiment of the present invention. Detailed Implementation

[0056] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] In the description and claims, it should be understood that the terms "first," "second," etc., used in the description and claims are only for the purpose of distinguishing the description of the same technical features, and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated, nor necessarily the order of description or chronological order. The terms are interchangeable where appropriate. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature.

[0058] This invention provides a method for improving the reliability of the optical bypass protection function of a distribution network switch. Please refer to [link to relevant documentation]. Figure 1 In this embodiment, the method for improving the reliability of the optical bypass protection function of the distribution network switch is specifically executed through steps S1 to S6:

[0059] S1. Obtain the scale of the power distribution network and collect power distribution network data in real time; the power distribution network data includes environmental data, electromagnetic interference intensity of each frequency band, and signal attenuation data.

[0060] It should be noted that the power distribution network data can be collected by sensors installed at various key points in the power distribution network. For example, electromagnetic interference sensors can be used to collect electromagnetic interference intensity, signal attenuation sensors can be used to collect signal attenuation during transmission, and environmental sensors can be used to collect environmental data such as ambient temperature and humidity. Of course, other sensors can also be used to collect data, and the specific type and model of the sensor selected does not affect the beneficial effects produced by the embodiments of the present invention.

[0061] S2. Based on the power distribution network data, identify the key environmental factors of electromagnetic interference and signal attenuation in each frequency band.

[0062] Signals of different frequency bands have different propagation characteristics and interference modes in power distribution networks, and the degree of influence of the same environmental factor on different frequency bands also varies. By identifying the key environmental factors corresponding to each frequency band, quality assessment can be effectively carried out, so as to further adaptively and dynamically allocate the spectrum, and maintenance personnel can also take targeted protective measures.

[0063] S3. Calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors.

[0064] In this embodiment of the invention, spectrum quality is evaluated using quantitative indicators by combining power distribution network data and key environmental factors. These quantitative indicators include signal-to-noise ratio, attenuation coefficient, or interference duration. Calculating spectrum quality provides an objective quality ranking for subsequent spectrum allocation, forming the basis for dynamic spectrum allocation.

[0065] S4. Based on the scale of the power distribution network, classify the power distribution network as a small-scale power distribution network or a non-small-scale power distribution network.

[0066] It should be noted that small-scale power distribution networks typically have a simple structure and few variables, while non-small-scale power distribution networks need to handle more dynamic variables. In this embodiment of the invention, classifying power distribution networks according to their scale enables the adoption of more suitable optimization algorithms for different scale distribution networks, ensuring the efficiency and accuracy of the strategy.

[0067] S5. For the small power distribution network, based on the spectrum quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection function.

[0068] S6. For the non-small-scale power distribution network, reinforcement learning is used to dynamically allocate the spectrum for each optical bypass protection function based on the spectrum quality.

[0069] In this embodiment of the invention, different dynamic spectrum allocation strategies are adaptively selected based on the different scale characteristics of the power distribution network, improving computational efficiency and strategy effectiveness. The optical bypass protection function of small power distribution networks is treated as centralized, and dynamic spectrum allocation is performed through a genetic algorithm; the optical bypass protection function of non-small power distribution networks is treated as distributed, and dynamic spectrum allocation is performed through reinforcement learning, so that frequency bands with higher spectral quality index are preferentially selected for data transmission, and protection signals are allocated to high-priority frequency bands.

[0070] As a preferred implementation, step S1, obtaining the power distribution network scale and collecting power distribution network data in real time, includes:

[0071] Obtain the scale of the power distribution network; the scale of the power distribution network includes the number of topological nodes in the power distribution network;

[0072] Initial power distribution network data is obtained by collecting electromagnetic interference intensity, signal attenuation data, and environmental data at key points in real time through equipment in the power distribution network.

[0073] The initial power distribution network data is preprocessed to obtain environmental data, electromagnetic interference intensity and signal attenuation data for each frequency band, which are then used as power distribution network data.

[0074] It should be noted that, in this embodiment of the invention, the scale of the power distribution network is represented by the number of topological nodes in the power distribution network. Therefore, in some preferred embodiments, step S4, classifying the power distribution network as a small-scale or non-small-scale power distribution network based on its scale, includes: if the number of topological nodes in the power distribution network is less than a preset threshold, then the power distribution network is considered a small-scale power distribution network; otherwise, it is considered a non-small-scale power distribution network. In other preferred embodiments, the scale of the power distribution network can also be represented by its coverage area.

[0075] In some preferred embodiments, the initial power distribution network data is preprocessed, including filtering, amplification, formatting, and outlier detection operations, to ensure the accuracy and validity of the data.

[0076] As a preferred implementation, step S2 involves identifying key environmental factors affecting electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data, including:

[0077] Calculate the first correlation between the electromagnetic interference intensity and environmental data in each frequency band to obtain the first impact score of each environmental factor on electromagnetic interference.

[0078] The second correlation between the signal attenuation data and environmental data in each frequency band is calculated to obtain the second impact score of each environmental factor on signal attenuation.

[0079] The environmental factors corresponding to the first influence score being greater than a preset first score threshold or the second influence score being greater than a preset second score threshold are selected to obtain the key environmental factors.

[0080] In some preferred embodiments, step S2 above can be executed using a deep learning algorithm. The deep learning algorithm learns and analyzes the correlation between electromagnetic interference intensity, signal attenuation, and various environmental factors in each frequency band, thereby identifying environmental factors whose impact on electromagnetic interference intensity or signal attenuation in each frequency band is greater than a preset intensity, and these are considered key environmental factors. The training samples for this deep learning algorithm are the actual measured values ​​of interference intensity and signal attenuation in each frequency band under changes in a single environmental factor.

[0081] As a preferred implementation, step S3, calculating the spectral quality of each frequency band based on the power distribution network data and the key environmental factors, is executed through steps S31-S34:

[0082] S31. Normalize the electromagnetic interference intensity of each frequency band to obtain normalized electromagnetic interference values.

[0083] In some preferred embodiments, the electromagnetic interference intensity I is represented by the reciprocal of the signal-to-noise ratio (SNR):

[0084] I = 1 / SNR.

[0085] In some other preferred embodiments, the electromagnetic interference intensity I is represented by the interference power density:

[0086] I = P interference -P noisefloor ;

[0087] Among them, P interference P represents the power of the interference signal. noisefloor This represents the power of the inherent noise.

[0088] After obtaining the expression for the electromagnetic interference intensity, the intensity is then mapped to the range [0,1] using a maximum-minimum approach. Understandably, the stronger the electromagnetic interference, the closer the normalized electromagnetic interference value will be to 1.

[0089] S32. Calculate the ratio of the signal attenuation data to the maximum allowable signal attenuation threshold, and normalize the ratio for each frequency band to obtain the normalized signal attenuation value.

[0090] In this embodiment of the invention, when the ratio of the signal attenuation data to the maximum allowable signal attenuation threshold is greater than 1, that is, when the signal attenuation data is greater than the maximum allowable signal attenuation threshold, the normalized signal attenuation value is defined as 1.

[0091] S33. Input the normalized electromagnetic interference value, the normalized signal attenuation value, and the key environmental factors into the pre-trained neural network model to obtain data transmission quality parameters; the data transmission quality parameters include success rate, signal-to-noise ratio, and bit error rate.

[0092] S34. Based on the data transmission quality parameters, obtain the spectral quality corresponding to each frequency band.

[0093] Furthermore, preferably, the pre-trained neural network model is obtained by training historical power distribution network data and preset scoring rules;

[0094] The preset scoring rules include success rate scores corresponding to different data transmission success rates, signal-to-noise ratio scores corresponding to different signal-to-noise ratios, and bit error rate scores corresponding to different bit error rates.

[0095] For example, the scoring rules include: data transmission success rate scores corresponding to the first data transmission success rate range, the second data transmission success rate range, and the third data transmission success rate range, respectively; signal-to-noise ratio (SNR) scores corresponding to the first SNR range, the second SNR range, and the third SNR range, respectively; and bit error rate (BER) scores corresponding to the first BER range, the second BER range, and the third BER range, respectively.

[0096] It should be noted that the data transmission quality parameters calculated by the pre-trained neural network in this embodiment of the invention correspond to a scenario where the power distribution network equipment is free from faults and the power distribution network protocol configuration is error-free.

[0097] In some preferred embodiments, the spectral quality SQI of step S34 p Represented as:

[0098]

[0099] Where k1 is the first parameter coefficient of the normalized electromagnetic interference value I, k2 is the second parameter coefficient of the normalized electromagnetic interference value A, and k3 is the key environmental factor E. p The third parameter coefficient; k1, k2 and k3 are all obtained by fitting historical power distribution network data during the training process of the pre-trained neural network.

[0100] As a preferred implementation, step S5, for the small power distribution network, based on the spectrum quality, uses a genetic algorithm to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection, specifically executed through steps S51-S54:

[0101] S51. If the power distribution network is a small power distribution network, the frequency band of each optical bypass protection function corresponding to the link is encoded into an integer sequence to obtain the chromosome code.

[0102] S52. Based on the spectral quality and the chromosome encoding, establish a total fitness function with the goal of optimizing the spectral resources for global optical bypass protection;

[0103] S53. Based on the chromosome encoding, randomly generate K chromosomes to obtain the initial population;

[0104] S54. Perform crossover selection and mutation selection operations on each chromosome in the initial population to obtain the offspring chromosome with the highest fitness, and obtain the optimal spectrum allocation method.

[0105] S55. Using the optimal spectrum allocation method, frequency bands are allocated for each optical bypass protection function.

[0106] In some preferred embodiments, in step S51, the frequency band of each optical bypass protection function link in the small power distribution network is encoded into an integer sequence. When there are N optical bypass protection links and M available frequency bands in the power distribution network, the sequence encoding is represented as follows:

[0107] CH = [f1, f2, ..., f N ], f i ∈{1,2,……M};

[0108] Among them, f i The frequency band number assigned to the link corresponding to the i-th optical bypass protection function.

[0109] Further, preferably, the overall fitness function Fit, which aims at optimizing the spectrum resources of the global optical bypass protection function, as described in step S52, is expressed as:

[0110]

[0111] Among them, SQI p(i,f i ) represents the spectral quality of link i for each available frequency band; count(j) represents the number of links using frequency band j; P col This is the preset conflict penalty value.

[0112] Preferably, when performing step S53, constraint correction is set to ensure that the initial population is conflict-free. Specifically, if there are conflicting frequency bands during initialization, the frequency bands of the conflicting links are randomly adjusted to idle positions; under the above constraint conditions, K chromosomes are randomly generated to obtain the initial population.

[0113] Step S54: Perform crossover and mutation selection operations on each chromosome in the initial population to obtain the offspring chromosome with the highest fitness, thus obtaining the optimal spectrum allocation method, including:

[0114] A tournament selection strategy is used to randomly select T individuals from the initial population. The individual with the highest fitness among the T individuals is selected as the parent. The random selection process is repeated until the number of parents reaches a preset threshold.

[0115] Using a single-point crossover method, crossover points are randomly selected, and genes after the crossover points of two parent chromosomes are exchanged to obtain the first offspring chromosome;

[0116] Using a random mutation method, a gene in a chromosome is randomly selected with a preset probability and replaced with another frequency band to obtain the second offspring chromosome;

[0117] The optimal spectrum allocation method is obtained by selecting the offspring chromosome with the highest fitness from the first and second offspring chromosomes.

[0118] As a preferred implementation, step S6, for the non-small-scale power distribution network, dynamically allocates the spectrum for each optical bypass protection function using reinforcement learning based on the spectrum quality, and is executed through steps S61-S65:

[0119] S61. If the power distribution network is not a small-scale power distribution network, the spectrum quality is classified, and the state space is obtained based on the classification result.

[0120] S62. Define the frequency band selection action based on the available frequency bands to obtain the action space;

[0121] S63. Calculate the spectral quality corresponding to the frequency band selection action and construct the reward function;

[0122] S64. Based on the state space, the action space, and the reward function, Q-learning is used to solve for the optimal spectrum allocation method corresponding to each optical bypass protection function.

[0123] S65. Using the optimal spectrum allocation method, frequency bands are allocated for each optical bypass protection function.

[0124] In a preferred embodiment of the present invention, when performing step S61, the spectrum quality is divided into four levels: “excellent”, “good”, “medium”, and “poor” according to the value of the spectrum quality. The four levels constitute a state space. When the spectrum of the optical bypass protection function is redistributed, the state space may change.

[0125] Preferably, the frequency band selection action is defined as A = {a1, a2, ..., a...} M}, where a i This indicates a switch to the i-th frequency band. More preferably, action constraints can also be defined, for example, reserving a "spectrum protection zone" for critical protection signals, preventing other signals from selecting that specific frequency band.

[0126] In some preferred embodiments, the reward function described in step S63 provides a corresponding reward based on the spectral quality of the selected frequency band; the higher the spectral quality, the higher the reward. Similarly, for actions that select low-quality frequency bands or cause transmission failures, a negative reward is given as a penalty. The reward function is expressed as follows:

[0127] R(t) = SOI p (t)+Penalty(t);

[0128]

[0129] Among them, SOI p (t) represents the spectral quality at time t; η is the preset penalty coefficient.

[0130] Further, preferably, step S64, based on the state space, the action space, and the reward function, uses Q-learning to obtain the optimal spectrum allocation method corresponding to each optical bypass protection function, including:

[0131] Each link corresponding to the optical bypass protection function is regarded as an intelligent agent, and the intelligent agent is initialized.

[0132] Based on the initialized agent, select and execute actions in the action space, calculate the reward function corresponding to the action to update the Q value, until the Q value converges to obtain the final action;

[0133] Based on the final action, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained.

[0134] Preferably, the update rule for the Q value is expressed as follows:

[0135] Q(s,a)←Q(s,a)+λ[R(t)+ρ·max a 'Q(s ’ a ’ )-Q(s,a)];

[0136] Where λ is the learning rate, ρ is the discount factor, and s ’ The new state to be entered after performing action a; when selecting an action, the action is selected with random probability.

[0137] For example, each optical bypass protection function corresponding to a link is considered as a reference, and a Q-table is initialized for each agent; then, action selection interaction is performed; the reward value is calculated; state transition and Q-table update are performed; the above steps are repeated for iterative looping until the Q-table converges or the maximum number of iterations is reached, and the optimal frequency band allocation for each agent is obtained.

[0138] The method for improving the reliability of optical bypass protection function of distribution network switch provided by the embodiments of the present invention can be effectively applied in extremely complex electromagnetic environments, ensuring that centralized or distributed optical bypass protection function always transmits data on the optimal frequency band, thereby improving the reliability of optical bypass protection function.

[0139] This invention provides a system for improving the reliability of the optical bypass protection function of a distribution network switch. Please refer to [link to relevant documentation]. Figure 2 The system for improving the reliability of the optical bypass protection function of the distribution network switch includes a data acquisition module 11, a key environmental factor identification module 12, a spectrum quality calculation module 13, a distribution network division module 14, a first spectrum allocation module 15, and a second spectrum allocation module 16, wherein:

[0140] The data acquisition module 11 is used to acquire the scale of the power distribution network and collect power distribution network data in real time; the power distribution network data includes environmental data, electromagnetic interference intensity of each frequency band, and signal attenuation data;

[0141] The key environmental factor identification module 12 is used to identify key environmental factors of electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data.

[0142] The spectrum quality calculation module 13 is used to calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors.

[0143] The power distribution network division module 14 is used to classify the power distribution network into a small power distribution network or a non-small power distribution network according to the scale of the power distribution network.

[0144] The first spectrum allocation module 15 is used to dynamically allocate spectrum for the small power distribution network based on the spectrum quality using a genetic algorithm with the goal of optimizing the spectrum resources for global optical bypass protection function.

[0145] The second spectrum allocation module 16 is used to dynamically allocate spectrum for each optical bypass protection function based on the spectrum quality for the non-small power distribution network using reinforcement learning.

[0146] In a preferred embodiment, the data acquisition module 11 is specifically used for:

[0147] Obtain the scale of the power distribution network; the scale of the power distribution network includes the number of topological nodes in the power distribution network;

[0148] Initial power distribution network data is obtained by collecting electromagnetic interference intensity, signal attenuation data, and environmental data at key points in real time through equipment in the power distribution network.

[0149] The initial power distribution network data is preprocessed to obtain environmental data, electromagnetic interference intensity and signal attenuation data for each frequency band, which are then used as power distribution network data.

[0150] In a preferred embodiment, the key environmental factor identification module 12 is specifically used for:

[0151] Calculate the first correlation between the electromagnetic interference intensity and environmental data in each frequency band to obtain the first impact score of each environmental factor on electromagnetic interference.

[0152] The second correlation between the signal attenuation data and environmental data in each frequency band is calculated to obtain the second impact score of each environmental factor on signal attenuation.

[0153] The environmental factors corresponding to the first influence score being greater than a preset first score threshold or the second influence score being greater than a preset second score threshold are selected to obtain the key environmental factors.

[0154] In a preferred embodiment, the spectrum quality calculation module 13 includes:

[0155] An electromagnetic interference normalization unit is used to normalize the electromagnetic interference intensity of each frequency band to obtain a normalized electromagnetic interference value.

[0156] The signal attenuation normalization unit is used to calculate the ratio of the signal attenuation data to the maximum allowable signal attenuation threshold, and to normalize the ratio for each frequency band to obtain the normalized signal attenuation value.

[0157] The data transmission quality parameter calculation unit is used to input the normalized electromagnetic interference value, the normalized signal attenuation value, and the key environmental factors into a pre-trained neural network model to obtain data transmission quality parameters; the data transmission quality parameters include success rate, signal-to-noise ratio, and bit error rate;

[0158] The spectrum quality calculation unit includes obtaining the spectrum quality corresponding to each frequency band based on the data transmission quality parameters.

[0159] Furthermore, preferably, the pre-trained neural network model is obtained by training historical power distribution network data and preset scoring rules;

[0160] The preset scoring rules include success rate scores corresponding to different data transmission success rates, signal-to-noise ratio scores corresponding to different signal-to-noise ratios, and bit error rate scores corresponding to different bit error rates.

[0161] In a preferred embodiment, the first spectrum allocation module 15 includes:

[0162] The encoding unit is used to encode the frequency band of each optical bypass protection function corresponding to the link into an integer sequence to obtain chromosome encoding if the power distribution network is a small power distribution network;

[0163] The function establishment unit is used to establish a total fitness function with the goal of optimizing the spectrum resources of the global optical bypass protection function, based on the spectrum quality and the chromosome encoding.

[0164] The population initialization unit is used to randomly generate K chromosomes according to the chromosome encoding to obtain an initial population;

[0165] The selection unit is used to perform crossover selection and mutation selection operations on each chromosome in the initial population to obtain the offspring chromosome with the highest fitness and to obtain the optimal spectrum allocation method.

[0166] The first allocation unit is used to allocate frequency bands for each optical bypass protection function using the optimal spectrum allocation method.

[0167] Further, preferably, the function establishment unit is specifically used for:

[0168] pass Establish the overall fitness function;

[0169] Where CH = [f1, f2, ..., f N ], f i ∈{1,2,……M},f i For the optical bypass protection function, link i is the link corresponding to the function; N is the total number of links; and M is the total number of available frequency bands. SQI p (i,f i ) represents the spectral quality of link i for each available frequency band; count(j) represents the number of links using frequency band j; P col This is the preset conflict penalty value.

[0170] In a preferred embodiment, the second spectrum allocation module 16 includes:

[0171] A state space construction unit is used to classify the spectrum quality if the power distribution network is not a small-scale power distribution network, and obtain the state space based on the classification result.

[0172] The action space construction unit is used to select actions based on available frequency bands to obtain the action space.

[0173] The reward function construction unit is used to calculate the spectral quality corresponding to the frequency band selection action and construct the reward function.

[0174] The solving unit is used to obtain the optimal spectrum allocation method corresponding to each optical bypass protection function by using Q-learning based on the state space, the action space and the reward function;

[0175] The second allocation unit is used to allocate frequency bands for each optical bypass protection function using the optimal spectrum allocation method.

[0176] Further, preferably, the solving unit is specifically used for:

[0177] Each link corresponding to the optical bypass protection function is regarded as an intelligent agent, and the intelligent agent is initialized.

[0178] Based on the initialized agent, select and execute actions in the action space, calculate the reward function corresponding to the action to update the Q value, until the Q value converges to obtain the final action;

[0179] Based on the final action, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained.

[0180] The system for improving the reliability of optical bypass protection function of distribution network switch provided by the embodiments of the present invention can be effectively applied in extremely complex electromagnetic environments, ensuring that centralized or distributed optical bypass protection function always transmits data on the optimal frequency band, thereby improving the reliability of optical bypass protection function.

[0181] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0182] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for improving the reliability of optical bypass protection function of distribution network switches, characterized in that, include: The scale of the power distribution network is obtained, and power distribution network data is collected in real time; the power distribution network data includes environmental data, electromagnetic interference intensity and signal attenuation data of each frequency band; Based on the power distribution network data, identify the key environmental factors causing electromagnetic interference and signal attenuation in each frequency band; Calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors. Based on the scale of the power distribution network, the power distribution network is classified as a small power distribution network or a non-small power distribution network; For the small power distribution network, based on the spectrum quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection. For the aforementioned non-small-scale power distribution network, reinforcement learning is used to dynamically allocate spectrum for each optical bypass protection function based on the spectrum quality.

2. The method for improving the reliability of optical bypass protection function of distribution network switches as described in claim 1, characterized in that, The process of obtaining the scale of the power distribution network and collecting power distribution network data in real time includes: Obtain the scale of the power distribution network; the scale of the power distribution network includes the number of topological nodes in the power distribution network; Initial power distribution network data is obtained by collecting electromagnetic interference intensity, signal attenuation data, and environmental data at key points in real time through equipment in the power distribution network. The initial power distribution network data is preprocessed to obtain environmental data, electromagnetic interference intensity and signal attenuation data for each frequency band, which are then used as power distribution network data.

3. The method for improving the reliability of optical bypass protection function of distribution network switches as described in claim 1, characterized in that, The identification of key environmental factors affecting electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data includes: Calculate the first correlation between the electromagnetic interference intensity and environmental data in each frequency band to obtain the first impact score of each environmental factor on electromagnetic interference. The second correlation between the signal attenuation data and environmental data in each frequency band is calculated to obtain the second impact score of each environmental factor on signal attenuation. The environmental factors corresponding to the first influence score being greater than a preset first score threshold or the second influence score being greater than a preset second score threshold are selected to obtain the key environmental factors.

4. The method for improving the reliability of optical bypass protection function of distribution network switch as described in claim 1, characterized in that, The calculation of the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors includes: The electromagnetic interference intensity of each frequency band is normalized to obtain the normalized electromagnetic interference value. Calculate the ratio of the signal attenuation data to the maximum allowable signal attenuation threshold, and normalize the ratio for each frequency band to obtain the normalized signal attenuation value. The normalized electromagnetic interference value, the normalized signal attenuation value, and the key environmental factors are input into a pre-trained neural network model to obtain data transmission quality parameters; the data transmission quality parameters include success rate, signal-to-noise ratio, and bit error rate. Based on the data transmission quality parameters, the spectral quality corresponding to each frequency band is obtained.

5. A method for improving the reliability of optical bypass protection function of a distribution network switch as described in claim 4, characterized in that, The pre-trained neural network model is obtained by training historical power distribution network data and preset scoring rules; The preset scoring rules include success rate scores corresponding to different data transmission success rates, signal-to-noise ratio scores corresponding to different signal-to-noise ratios, and bit error rate scores corresponding to different bit error rates.

6. The method for improving the reliability of optical bypass protection function of distribution network switches as described in claim 1, characterized in that, For the small-scale power distribution network, based on the spectrum quality, a genetic algorithm is used to dynamically allocate spectrum with the goal of optimizing the spectrum resources for global optical bypass protection, including: If the power distribution network is a small power distribution network, the frequency band of each optical bypass protection function corresponding to the link is encoded into an integer sequence to obtain chromosome encoding; Based on the spectral quality and the chromosome encoding, a total fitness function is established with the goal of optimizing the spectral resources for global optical bypass protection. Based on the chromosome encoding, K chromosomes are randomly generated to obtain the initial population; Crossover selection and mutation selection operations are performed on each chromosome in the initial population to obtain the offspring chromosome with the highest fitness, thus obtaining the optimal spectrum allocation method. The optimal spectrum allocation method is used to allocate frequency bands for each optical bypass protection function.

7. A method for improving the reliability of optical bypass protection function of a distribution network switch as described in claim 6, characterized in that, The step of establishing a total fitness function based on the spectral quality and the chromosome encoding, with the goal of optimizing spectral resources for global optical bypass protection, includes: pass Establish the overall fitness function; Where CH = [f1, f2, ..., f N ], f i ∈{1,2,……M},f i For the optical bypass protection function, link i is the link corresponding to the function; N is the total number of links; and M is the total number of available frequency bands. SQI p (i,f i ) represents the spectral quality of link i for each available frequency band; count(j) represents the number of links using frequency band j; P col This is the preset conflict penalty value.

8. A method for improving the reliability of optical bypass protection function of a distribution network switch as described in claim 1, characterized in that, For the non-small-scale power distribution network, based on the spectrum quality, reinforcement learning is used to dynamically allocate spectrum for each optical bypass protection function, including: If the power distribution network is not a small-scale power distribution network, the spectrum quality is classified, and the state space is obtained based on the classification results. Based on the available frequency bands, define the frequency band selection action to obtain the action space; Calculate the spectral quality corresponding to the frequency band selection action and construct the reward function; Based on the state space, the action space, and the reward function, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained by Q-learning. The optimal spectrum allocation method is used to allocate frequency bands for each optical bypass protection function.

9. A method for improving the reliability of optical bypass protection function of a distribution network switch as described in claim 8, characterized in that, The step of obtaining the optimal spectrum allocation method corresponding to each optical bypass protection function using Q-learning based on the state space, the action space, and the reward function includes: Each link corresponding to the optical bypass protection function is regarded as an intelligent agent, and the intelligent agent is initialized. Based on the initialized agent, select and execute actions in the action space, calculate the reward function corresponding to the action to update the Q value, until the Q value converges to obtain the final action; Based on the final action, the optimal spectrum allocation method corresponding to each optical bypass protection function is obtained.

10. A system for improving the reliability of optical bypass protection function of distribution network switches, characterized in that, include: The data acquisition module is used to obtain the scale of the power distribution network and collect power distribution network data in real time; the power distribution network data includes environmental data, electromagnetic interference intensity of each frequency band, and signal attenuation data; The key environmental factor identification module is used to identify key environmental factors of electromagnetic interference and signal attenuation in each frequency band based on the power distribution network data. The spectrum quality calculation module is used to calculate the spectrum quality of each frequency band based on the power distribution network data and the key environmental factors. The power distribution network segmentation module is used to classify the power distribution network into a small power distribution network or a non-small power distribution network according to the scale of the power distribution network. The first spectrum allocation module is used to dynamically allocate spectrum for the small power distribution network based on the spectrum quality, using a genetic algorithm with the goal of optimizing the spectrum resources for global optical bypass protection function. The second spectrum allocation module is used to dynamically allocate spectrum for each optical bypass protection function based on the spectrum quality for the non-small power distribution network using reinforcement learning.

Images