Photovoltaic power station data transmission and monitoring system based on 5G technology

Abstract

The invention relates to the technical field of communication route searching, in particular to a photovoltaic power station data transmission and monitoring system based on the 5G technology, and the system comprises a route state monitoring module which carries out the collection and statistics of the time delay value, bandwidth utilization rate and congestion degree value of each transmission link in a 5G base station service area, and generates a link state parameter set. According to the invention, the time delay value, the bandwidth utilization rate and the congestion degree value of each transmission link in the 5G base station service area are collected and counted, and the resource utilization efficiency is calculated based on the network load balance value and the energy consumption threshold value, so that the link state evaluation has dynamics and real-time performance, and the availability of different links can be described. And the links are calculated and screened based on the energy consumption load ratio to form a priority sequence of the transmission paths, so that the data traffic can be preferentially allocated to the link with the optimal transmission efficiency, and the occupation of the overall bandwidth by invalid traffic transmission is reduced.

Description

Technical Field

[0001] The present invention relates to the field of communication route search technology, and in particular to a photovoltaic power station data transmission and monitoring system based on 5G technology. Background Art

[0002] Communication route lookup is one of the core technologies in the field of computer networks and wireless communications, and is mainly used to determine the path of data transmission in complex network topologies.

[0003] Existing communication routing search technology relies on static topology structure for path selection and lacks dynamic monitoring of link status. When the network load changes, the path selection may not match the actual situation, affecting the stability of data transmission. In addition, path screening does not consider the energy consumption factor of the link, which may cause data to be transmitted preferentially through high-energy consumption links, increasing the overall energy consumption burden and reducing network operation efficiency. Therefore, improvements are needed. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose a photovoltaic power station data transmission and monitoring system based on 5G technology.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solution: A photovoltaic power station data transmission and monitoring system based on 5G technology includes:

[0006] The routing status monitoring module collects and counts the delay value, bandwidth utilization rate and congestion level value of each transmission link in the 5G base station service area to generate a link status parameter set; calculates the link status parameter set with the network load balancing value and energy consumption threshold to obtain resource utilization efficiency and generate a routing resource monitoring matrix;

[0007] A path optimization calculation module calculates the energy consumption load ratio of each link based on the routing resource monitoring matrix and generates a link weight factor set; based on the link weight factor set, selects a link according to the delay value and bandwidth utilization and generates a path optimization solution;

[0008] The network slice scheduling module, based on the path optimization scheme, classifies the photovoltaic power station control data according to the delay value to generate a service classification sequence; allocates independent network slice resources to the service classification sequence to generate a slice scheduling strategy;

[0009] The data forwarding control module allocates the data traffic under each service classification sequence to the corresponding network slice resources based on the slice scheduling strategy and generates routing control instructions.

[0010] Preferably, the steps of acquiring the link state parameter set are:

[0011] Deploy sensors on each transmission link in the 5G base station service area to monitor and record the link latency, bandwidth utilization, and congestion level in real time to obtain preliminary transmission link performance data;

[0012] Based on the preliminary transmission link performance data, outliers are removed, data is normalized and format is standardized to generate a link state parameter set.

[0013] Preferably, the steps of acquiring the routing resource monitoring matrix are:

[0014] Based on the link state parameter set, the resource utilization efficiency index of each link is calculated, and the calculation formula is:

[0015]

[0016] in, represents the resource utilization efficiency index of link i, represents the load balancing value of link i, represents the bandwidth utilization of link i, represents the energy consumption threshold of link i, represents the instantaneous transmission rate of link i, represents the congestion frequency of link i, represents the maximum throughput of link i, Represents the available bandwidth ratio of link i at the current moment;

[0017] Based on the resource utilization efficiency index, the links are sorted according to their resource utilization efficiency index, a matrix model is constructed, a mapping relationship between resource distribution and monitoring status of all links is formed, and a routing resource monitoring matrix is ​​obtained.

[0018] Preferably, the steps of acquiring the link weight factor set are:

[0019] According to the routing resource monitoring matrix, the energy consumption load ratio of each link is calculated, and the calculation formula is:

[0020]

[0021] Among them, W j represents the energy consumption load ratio of link j, K j represents the energy consumption threshold of link j, P j represents the instantaneous load value of link j, R j represents the load balancing value of link j, S j represents the data flow density of link j, T j represents the traffic change rate of link j, U j represents the instantaneous throughput capacity of link j;

[0022] Based on the energy consumption load ratio, all calculation results are integrated according to the link number, and the ratio data of all links are summarized into a set to form a link weight factor set.

[0023] Preferably, the steps of obtaining the path optimization solution are:

[0024] Extract all link data in the link weight factor set, call the delay value and bandwidth utilization of the link, build an association table, filter the links with missing or abnormal data in the association table, and obtain a link performance data set;

[0025] Based on the link performance data set, the path fitness score of each link is calculated using the following formula:

[0026]

[0027] Among them, Z k represents the path fitness score of link k, X k represents the delay value of link k, Y k represents the bandwidth utilization of link k, M k represents the maximum throughput of link k, N k represents the data flow stability of link k at the current moment, O k Represents the real-time congestion status of link k;

[0028] Based on the path fitness score, all links are sorted according to the path fitness score, and the link with the highest path fitness score is selected to form a path optimization plan.

[0029] Preferably, the steps of obtaining the service classification sequence are:

[0030] Based on the path optimization solution, the delay priority score of each data point is calculated using the following formula:

[0031]

[0032] Among them, V m represents the latency priority score of data point m, A m represents the real-time transmission delay of data point m, B m represents the maximum transmission delay of data point m, C m represents the congestion impact factor of data point m, D m represents the target link bandwidth utilization of data point m, E m represents the throughput of data point m at the current moment, F m represents the instantaneous load value of the link to which data point m belongs, G m Represents the load fluctuation amplitude of data point m, H mRepresents the load stability of data point m at the current moment;

[0033] Based on the delay priority score, all data points are graded, and according to the sorting results of the delay priority score, the data points are divided into multiple service levels to form a service classification sequence.

[0034] Preferably, the steps of obtaining the slice scheduling strategy are:

[0035] Extract each data category in the service classification sequence, call the network slice resource pool, match according to the service level priority, screen the available network slices, establish a mapping relationship between different data categories, and generate a preliminary network slice allocation plan;

[0036] Based on the preliminary network slice allocation plan, the bandwidth utilization, latency carrying capacity and load balancing value of the network slice are called, and the allocation ratio of the service classification sequence is adjusted according to the resource carrying capacity of each network slice to generate an optimized network slice allocation plan;

[0037] Based on the optimized network slice allocation scheme, network slice resources are scheduled according to the traffic changes of business data to form a slice scheduling strategy.

[0038] Preferably, the steps of obtaining the routing control instruction are:

[0039] Evaluate the network slice resource configuration in the slice scheduling strategy, extract the traffic demand of each service classification sequence, determine the matching degree between each service demand and the network slice resource, and generate a service demand and network slice matching table;

[0040] Based on the business demand and network slice matching table, the data traffic under each business classification sequence is allocated to the network slice resources to obtain a traffic allocation plan, and routing control instructions are generated according to the traffic allocation plan.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are:

[0042] In the present invention, the delay value, bandwidth utilization rate and congestion degree value of each transmission link in the service area of ​​5G base station are collected and counted, and the resource utilization efficiency is calculated based on the network load balancing value and energy consumption threshold, so that the link state evaluation has dynamic and real-time characteristics, and can characterize the availability of different links. Based on the energy consumption load ratio calculation, the link is selected to form a priority sorting of the transmission path, ensuring that the data traffic can be preferentially allocated to the link with the best transmission efficiency, reducing the occupation of the overall bandwidth by invalid traffic transmission. The optimal link is calculated by combining the delay value and the bandwidth utilization rate to improve the accuracy of path selection, so that the data can still maintain the characteristics of low latency and high throughput under the dynamic changes of the network environment. The control data of the photovoltaic power station is graded according to the delay requirements to form a hierarchical management strategy, so that data with high real-time requirements are given priority in resource allocation, ensuring the timeliness and stability of data transmission. Independently allocate network slice resources, adjust the allocation strategy in combination with the data flow type and business needs, avoid the waste of bandwidth caused by the solidification configuration of resources, and improve the utilization of network resources. The transmission path of data streams is dynamically adjusted based on the service classification sequence, so that traffic scheduling is not only based on the current network status, but also based on historical data to predict load trends, reducing link overload problems caused by burst traffic. By establishing an adaptive resource matching mechanism, different types of data streams can be allocated to the optimal link according to business characteristics, improving the balance and reliability of overall data transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0045] See also Figure 1 , the present invention provides a technical solution: a photovoltaic power station data transmission and monitoring system based on 5G technology includes:

[0046] The routing status monitoring module collects and counts the latency, bandwidth utilization, and congestion values ​​of each transmission link in the 5G base station service area to generate a link status parameter set; the link status parameter set is calculated with the network load balancing value and energy consumption threshold to obtain resource utilization efficiency and generate a routing resource monitoring matrix;

[0047] The path optimization calculation module calculates the energy consumption load ratio of each link based on the routing resource monitoring matrix and generates a link weight factor set; based on the link weight factor set, it selects links according to the delay value and bandwidth utilization and generates a path optimization plan;

[0048] The network slice scheduling module, based on the path optimization solution, classifies the control data of the photovoltaic power station according to the delay value and generates a service classification sequence; allocates independent network slice resources to the service classification sequence and generates a slice scheduling strategy;

[0049] The data forwarding control module allocates the data traffic under each service classification sequence to the corresponding network slice resources based on the slice scheduling strategy and generates routing control instructions.

[0050] The steps to obtain the link state parameter set are:

[0051] Deploy sensors on each transmission link in the 5G base station service area to monitor and record the link latency, bandwidth utilization, and congestion level in real time to obtain preliminary transmission link performance data;

[0052] Based on the preliminary transmission link performance data, outlier removal, data normalization and format standardization are performed to generate a link status parameter set.

[0053] Specifically, sensors are deployed on each transmission link in the service area of ​​the 5G base station to sample data 24 hours a day. For example, the delay value of the link is obtained every 10 seconds and the bandwidth utilization and congestion level at that moment are recorded. The obtained delay value is compared with the reference range of 0ms to 150ms. The reference range is a threshold set based on the comprehensive results of the delay distribution mean of about 40ms and three times the standard deviation of about 110ms obtained from statistics in the past 30 days. If some delay values ​​are found to exceed 150ms, they are marked as high delay and the high delay value is saved in a separate record. The bandwidth utilization rate is also set in the range of 0% to 85% as the daily operation benchmark. The upper limit threshold of 85% is obtained based on the recently collected bandwidth usage peak value and after leaving a certain margin. If the bandwidth utilization rate exceeds 85%, it will be cross-compared with the real-time congestion level value. When the congestion level value also reaches or exceeds 80%, it is regarded as a potential congestion state. The relevant records are marked and included in a separate queue. All transmission links are monitored cyclically at the same collection frequency, and the time series data of each link is accumulated to form a preliminary set of delay values, bandwidth utilization and congestion level values ​​to obtain preliminary transmission link performance data.

[0054] Based on the preliminary transmission link performance data, the outliers are first identified and removed. For example, the mean μ and standard deviation σ are calculated for each delay value, and the records with |X-μ|≥3σ are regarded as abnormal samples. The 3 times standard deviation threshold is a comprehensive value obtained based on historical statistical methods combined with 30-day monitoring data. If an outlier is found, the record is removed and placed in an independent list. Then, the bandwidth utilization and congestion values ​​are detected in the same way and extreme data that is too large or too small is removed. Then, the minimum-maximum normalization method is selected in the normalization stage to calculate for each value X. Where X min and X max They represent the minimum and maximum values ​​of the indicator in the filtered data respectively. For example, for bandwidth utilization, if the minimum value of the remaining data is 5% and the maximum value is 90%, the bandwidth utilization in each record is mapped to between 0 and 1. When the format is standardized, the delay value, bandwidth utilization, and congestion level value are aligned to the same field sequence, and the timestamp of each record is matched with the corresponding parameter. Finally, the processed data is merged into a new parameter list to generate a link state parameter set.

[0055] The steps to obtain the routing resource monitoring matrix are:

[0056] Based on the link state parameter set, the resource utilization efficiency index of each link is calculated using the following formula:

[0057]

[0058] in, represents the resource utilization efficiency index of link i, represents the load balancing value of link i, represents the bandwidth utilization of link i, represents the energy consumption threshold of link i, represents the instantaneous transmission rate of link i, represents the congestion frequency of link i, represents the maximum throughput of link i, Represents the available bandwidth ratio of link i at the current moment;

[0059] Based on the resource utilization efficiency index, the links are sorted according to their resource utilization efficiency index, a matrix model is constructed, a mapping relationship between resource distribution and monitoring status of all links is formed, and a routing resource monitoring matrix is ​​obtained.

[0060] Specifically, the formula is beneficial in that it comprehensively evaluates the resource allocation balance and transmission efficiency of the link in a real environment by simultaneously considering multiple factors such as the difference between bandwidth utilization and energy consumption threshold, instantaneous transmission rate, congestion frequency and maximum throughput, and available bandwidth ratio.

[0061] The acquisition steps are as follows: during the 14-day monitoring period, the real-time load data of the link is recorded every 5 minutes. By performing mean and variance analysis on each record, the average value of the load occupancy ratio in each time period is obtained after removing extreme fluctuations. These average values ​​are then compared between time periods under the same link. The average level of the load distribution difference between all time periods is defined as the load balancing degree, and the load balancing degree is divided by the maximum load potential corresponding to this link to generate a basic ratio. This ratio can be regarded as a preliminary load balancing degree, which then needs to be corrected according to the number of time periods in which extreme loads appear in historical data. By recording the frequency of extreme loads and comparing them with the pre-set threshold range of 0 to 2 times, the situations exceeding 2 times are marked and included in the correction sequence, and the corrected value is combined with the basic ratio to obtain the final load balancing value. For example, for the first link, the extreme load occurs 3 times within 14 days, and multiplied by the correction factor 0.95, we finally get About 0.86.

[0062] The acquisition steps are as follows: according to the link status parameter set obtained above, the bandwidth usage recorded during the continuous monitoring period is extracted, and the bandwidth usage of each time period is calculated by the ratio of the bandwidth usage of each time period to the bandwidth upper limit of the link, and then these proportions are combined by hour, and the daily average proportion is extracted and the average values ​​of the two cycles of 7 days and 14 days are calculated. The 14-day average value is recorded as the full-cycle bandwidth usage level, and the difference between the 7-day average value and the 14-day average value is compared. If the difference between the two is between 0.02 and 0.05, the difference is included in the correction factor to update a more accurate bandwidth utilization. If the difference is less than 0.02, the 14-day average value is directly used as the final bandwidth utilization. For example, for link 1, the average daily bandwidth usage is about 600 Mbps in a 14-day scenario with a bandwidth upper limit of 1000 Mbps, and the corresponding bandwidth utilization is 0.60. If the difference between the 7-day average and the 14-day average is less than 0.02, it is directly recorded as 0.60.

[0063] The steps to obtain are: continuously monitor the total energy consumption level of the equipment on the link, record the energy consumption values ​​of the equipment under different traffic conditions, and then calculate the critical energy consumption threshold based on the power consumption curve of the equipment. Add the threshold to a certain redundancy value to get the energy consumption threshold range, and set the range to 0.4 to 0.7 (dimensionless relative value), where 0.4 is the relative value corresponding to the average power consumption of the equipment during long-term operation under normal traffic, and 0.7 is the relative value corresponding to a peak power consumption monitored after the equipment has been running for 14 hours at maximum traffic. For a certain link, if the relative value corresponding to the average power consumption during its operation is 0.55, it is compared with the high power consumption records monitored for multiple days. If the number of high power consumption is between 1 and 3 times, 0.55 is used as the final energy consumption threshold; if the number of high power consumption exceeds 3 times, a coefficient lower than 1.0 will be introduced to correct it again, and the final result is For example, in the first link, high power consumption was found twice in multiple days of statistics, so 0.55 was directly used as

[0064] The acquisition steps are as follows: in the busy period from 8:00 to 22:00 every day, record the transmission rate of the link every 15 minutes, select the peak value of each hour and form a sequence, then calculate the average value of the sequence as the preliminary instantaneous transmission rate, and then compare the transmission rate characteristics in the idle period from 22:00 to 8:00 the next day according to the idle period records in the late night period, and calculate the ratio relative to the peak rate during the day. If the ratio is greater than 0.5, multiply the difference by a correction coefficient and add it back to the preliminary value. If the ratio is less than 0.5, keep the preliminary value unchanged, and finally get the instantaneous transmission rate. For example, for the first link, the peak rate during the day is about 480 Mbps, and the average rate during the late night period is 300 Mbps. The ratio of late night is 300 / 480, which is about 0.625, which is greater than 0.5. Then multiply it by the correction factor 0.10 to get 0.125, and then add it back to 480 Mbps to get 480+0.125×480=540 Mbps. After reduction, we get

[0065] The acquisition steps are as follows: by continuously monitoring the queue length and packet loss rate of the link in unit time, once the queue length exceeds 1000 packets, it is considered as mild congestion, and once the packet loss rate exceeds 0.02, it is considered as moderate congestion. The total number of times these congestion periods occur within 24 hours is comprehensively counted, and the duration of the period is recorded. The product of the duration of congestion and the number of occurrences is normalized, with 0 representing no congestion and 1 representing extreme congestion, and then the preliminary value of the congestion frequency is obtained. After comparing the data period of the past 30 days, it is corrected based on whether it meets or exceeds the previously established regular interval of 0 to 10 occurrences. If the corrected congestion frequency is between 0.15 and 0.3, it is recorded as a normal congestion level. If it exceeds 0.3, an additional coefficient is added to reduce its value and make it stable. Combined with the above processing, in the 30-day statistical results of the first link, the total number of congestion times is 9 times, and each time lasts about 10 minutes on average. The result of 9×10=90 is mapped to about 0.25, and finally obtained

[0066] The steps to obtain are as follows: query the specification of the link operation equipment and cooperate with the actual test records to obtain the theoretical upper limit of its maximum throughput capacity, then conduct three throughput tests of no less than 6 hours each during the high load period, record the average throughput and peak throughput respectively, and calculate the proportion of the average peak of these three tests to the theoretical upper limit. If the proportion is less than 90%, multiply the theoretical value by a correction factor to adjust it downward. If it is higher than 90%, directly use the theoretical value as the maximum throughput capacity, and then obtain The specific setting basis of the correction coefficient is to compare the instantaneous overload conditions that occur in the test multiple times. For example, when the theoretical maximum throughput is 800Mbps and the average peak value of the three tests is only 700Mbps, the proportion is 700 / 800=0.875, which is lower than 90%. The correction coefficient 0.95 corresponding to 0.875 in the range of 0.80 to 0.90 is adjusted to obtain 800×0.95=760Mbps, which is recorded as For the first link, the average peak value of the above three tests accounts for about 88%, and the corresponding correction is

[0067] The steps for obtaining are as follows: in daily traffic detection, first count the actual bandwidth value occupied every hour, combine it with the maximum available bandwidth of the link, summarize a statistical series through the hourly ratio, and take the mean and peak ratio of the statistical series. If the difference between the mean and the peak ratio is in the range of 0.10 to 0.15, then the difference is recorded as the fluctuation correction coefficient, and the coefficient is used to correct the mean to obtain the available bandwidth ratio at the current moment. If the difference is less than 0.10, the mean is directly used as the available bandwidth ratio. If the difference is greater than 0.15, it means that extreme traffic occurs in certain time periods, which will be further split and processed by cumulative segmentation, and finally a more refined available bandwidth ratio is obtained. Referring to the first link, the actual bandwidth occupied per hour is 520 Mbps, and the peak bandwidth occupied is 580 Mbps. The difference between the two ratios is about (580-520) / 580=0.1034, which is approximately between 0.10 and 0.15. Therefore, 0.1034 is recorded as the fluctuation correction coefficient. The corrected available bandwidth ratio is about (1-0.1034)=0.8966, which is rounded to 0.90 and recorded as

[0068]

[0069] Calculation process:

[0070] The first step is to substitute the parameter values ​​into the formula:

[0071]

[0072] The second step is to calculate the numerator and denominator separately:

[0073]

[0074] ln(1+0.25 2 )=ln(1+0.0625)=ln(1.0625)≈0.0606

[0075] 1+ln(1.0625)=1.0606

[0076]

[0077] (0.86)×0.2236=0.1923

[0078] 0.1923+509.17=509.3623

[0079] 1+0.90=1.90

[0080]

[0081] 760+0.5263=760.5263

[0082] The third step is to divide the numerator by the denominator:

[0083]

[0084] The results show that under the current monitoring environment and parameter settings, the resource utilization efficiency index of the first link is about 0.67, indicating that under the combined effect of multiple factors such as load balancing, bandwidth utilization, energy consumption threshold, and congestion frequency, its overall resource allocation and transmission efficiency level is in a relatively medium range. If the same steps are performed on other links and the results are calculated in turn, the resource utilization efficiency index of each link can be compared at the overall level, providing an objective reference value for the next step of priority sorting, resource allocation, or further monitoring.

[0085] Based on the resource utilization efficiency index, after each link has been calculated and a column of final values ​​has been formed, these values ​​need to be sorted in sequence and marked with the identification information of the corresponding links, and then arranged in order from high to low or from low to high. When the resource utilization efficiency index is greater than 1.0, it can be regarded as the link has a higher load balance and transmission rate complementary characteristics. When the resource utilization efficiency index is between 0.5 and 1.0, it is marked as a medium level. If the resource utilization efficiency index of some links is monitored to be lower than 0.5, it will be included in the scope of inspection. The average value and fluctuation range of the link in the past 30 days are compared with the pre-recorded daily operation indicators. If it is found that the current value is significantly deviated from the historical comparison, for example, the difference exceeds 0.15 to 0.20, the link will be marked with a further investigation mark. After completing the above inspection, the final sorting results of all links are integrated into a visual matrix format, for example, in the matrix The columns are marked with link numbers 1 to N, and the rows are listed with the interval segments of the resource utilization efficiency index and given color classification. When the resource utilization efficiency index is between 1.5 and 2.0, it is marked as a color gradient range. If the resource utilization efficiency index is between 0.8 and 1.0, another gradient color is used to distinguish it. Such values ​​are cross-tabulated with the high-load periods of daily operation of the link. Combined with the bandwidth utilization and instantaneous transmission rate obtained previously, the stability of such links in different time periods or different seasonal changes is confirmed one by one. The congestion frequency and maximum throughput capacity ratio of each link in the past 7 days and 14 days are also listed in the matrix. These monitoring indicators are mapped to the corresponding row and column positions in the matrix and associated with the resource utilization efficiency index. Finally, a three-dimensional mapping relationship is formed, which is convenient for quickly identifying key links through this matrix in the subsequent operation monitoring process to obtain a routing resource monitoring matrix.

[0086] The steps for obtaining the link weight factor set are:

[0087] According to the routing resource monitoring matrix, the energy consumption load ratio of each link is calculated using the following formula:

[0088]

[0089] Among them, W j represents the energy consumption load ratio of link j, K j represents the energy consumption threshold of link j, P j represents the instantaneous load value of link j, R j represents the load balancing value of link j, S j represents the data flow density of link j, T j represents the traffic change rate of link j, U j represents the instantaneous throughput capacity of link j;

[0090] Based on the energy consumption load ratio, all calculation results are integrated according to the link number, and the ratio data of all links are summarized into a set to form a link weight factor set.

[0091] Specifically, the benefit of the formula is that it incorporates the link's energy consumption threshold, instantaneous load value, load balancing value, data flow density, traffic change rate and instantaneous throughput into the same calculation process, and by correlating the numerator and denominator of traffic change and energy consumption threshold, it shows a comprehensive trade-off between energy consumption and load distribution, and evaluates the link's energy consumption load level under various network conditions.

[0092] K j The acquisition steps are as follows: first, in a continuous 21-day monitoring period, extract the power consumption records of the corresponding equipment of the link every 10 minutes, and analyze these records in parallel with the load status of the link in the same time period, screen out the time periods in which the power consumption is stable at a high value for more than 2 hours (for example, the average power consumption of the same link equipment is observed to be greater than 600W), accumulate these high power consumption periods and compare them with the normal power consumption periods, and then extract the proportion of high power consumption from the comparison results of each period, and then cross-compare the proportion with the peak point of the equipment power consumption curve to clarify under what load range the equipment approaches the peak energy consumption, and then combine the high power consumption trigger points and normal power consumption levels of multiple tests to form several intervals, and count the average power consumption value in each interval, and define the power consumption value closest to the design limit of the equipment but still able to run for a long time as the relative energy consumption benchmark value, and establish an upper and lower floating range in combination with a certain proportional coefficient, and the middle position of the range is used as the energy consumption threshold K. j For example, in the statistics of a certain link device, it was found that the high power consumption period accounted for 27%, and the corresponding power consumption fluctuated between 580W and 610W. After confirming that the factory calibration limit was 700W and that it could be continuously stable around 600W, the energy consumption threshold K was set. jIt is 0.65 (here it is measured as a dimensionless relative value). Other links can obtain their respective energy consumption thresholds according to the same process.

[0093] P j The acquisition steps are as follows: when the traffic of the link is detected, the instantaneous load is recorded every 1 minute, and 1440 records within 24 hours are accumulated and segmented. The average load is segmented and counted according to the three main intervals of 0:00 to 8:00, 8:00 to 16:00, and 16:00 to 24:00. The average load values ​​of the three intervals are added to get the daily total average value. For example, if the average load of a link from 0:00 to 8:00 is 150Mbps, the average load from 8:00 to 16:00 is 220Mbps, and the average load from 16:00 to 24:00 is 180Mbps, then the daily total average value is about (150+220+180) / 3=183.33Mbps. In order to reflect the instantaneous characteristics, the peak load record of one minute level is intercepted in each interval and normalized. If the peak load exceeds the daily total average value by more than 40%, it is included in the instantaneous load statistics and continuously tracked for several hours to obtain the peak duration. Finally, the load average value within the peak duration is divided by the bandwidth upper limit to obtain a dimensionless result, which is recorded as the instantaneous load value P. j For example, if the load average value during the peak duration is 400Mbps and the bandwidth limit is 1000Mbps, we can get P j =0.40.

[0094] R j The steps for obtaining the load balance value are as follows: first query the operation records of the link for the previous 7 days and 14 days, extract the proportion of the load curve in the peak period and the valley period, and obtain the stability index of the load distribution curve by comparing the ratio of the duration of the peak period and the valley period, combined with the total bandwidth configuration of the link and the equipment specifications. Then compare the index with the range value set between 0.2 and 0.5. Records exceeding 0.5 indicate that the load accounts for a large proportion in the peak period, and records below 0.2 indicate that the load accounts for a large proportion in the valley period. If the record is between the two, it is considered that the load distribution is relatively balanced, thereby forming a preliminary judgment on the load balance value. Combined with an additional count of the number of peak and valley alternations (for example, the peak and valley alternation occurs within 3 times a day), the statistical number is weighted with the above preliminary judgment, and the final dimensionless value R is obtained by setting the step value in the range of 0 to 5 times and performing linear mapping. j For example, for a link, the peak and trough last for 6 hours each within 7 days, and the peak and trough appear twice alternately. Finally, R j =0.35.

[0095] S jThe acquisition steps are as follows: split the transmission data record of the link for 24 hours, obtain the total amount of data packets within 5 minutes each time and accumulate the number of bytes occupied to form a preliminary sequence of data flow density, and then divide the sequence into intervals according to the capacity of 0 to 1T (TB). If the capacity of the 5-minute data packet exceeds 600MB, it is recorded as a high-density record, and if it is less than 300MB, it is recorded as a low-density record. After marking each 5-minute record, the number of high-density and low-density segments within 24 hours is counted to obtain a ratio value about the data volume distribution, and then this ratio is mapped to between 0 and 1. The closer the value is to 1, the more high-density periods of data packets are, and the closer it is to 0, the more low-density periods are. If the value is in the range of 0.3 to 0.6, it means that the data flow distribution is biased to the middle. In the specific example, for example, a link has a total of 288 5-minute segments in 24 hours, of which 120 are high-density and 90 are low-density. The remaining 78 are considered to be medium density. By statistically analyzing the ratio, S j

[0096] =120 / 288=0.4167≈0.42.

[0097] T j The steps to obtain T are as follows: first, record the amount of bytes transmitted per hour and divide it by the bandwidth limit to obtain the traffic share of each hour. Then, count the change in share between two adjacent hours, accumulate these change values ​​and divide them by the total number of hours to form a preliminary measurement of the traffic change rate. Then, compare it with the benchmark change rate counted by the device in the past 14 days. If it exceeds the upper and lower floating range of the benchmark change rate (for example, the floating range of the benchmark change rate of 0.03 and 0.02), further count the number of time periods that exceed the range and include them in the correction item, and finally obtain a more accurate traffic change rate value T. j In one example, if the 24-hour traffic percentage fluctuates between 0.02 and 0.05, the average value is 0.035 and compared with the historical 14-day benchmark of 0.03. It is found that there is only one period when the fluctuation exceeds 0.02. In this case, only a slight adjustment is made during the correction, and the final result is T j =0.035.

[0098] U j The steps to obtain U are as follows: refer to the throughput capacity indicators published by the link equipment manufacturer, and combine with high-load performance tests for consecutive days, select the throughput peak values ​​measured in three different time periods: morning, noon, and evening every day, and obtain the instantaneous throughput capacity by adding a certain adjustment ratio to the average value. Then, continuously measure this instantaneous capacity. If it approaches this peak value several times within an hour, this peak value is regarded as the stable throughput level that the device can output at that moment, and it is divided by the theoretical maximum throughput value calibrated by the device to obtain the relative value, which is recorded as U jFor example, the theoretical maximum throughput of the link device is 900Mbps. After three consecutive days of actual high-load testing, the average peak is about 820Mbps, approaching 820Mbps several times an hour. The proportion of 900Mbps is 820 / 900=0.9111≈0.91, recorded as U j .

[0099] W j Represents the energy consumption-load ratio of link j. This value can be regarded as a comprehensive quantification of the relationship between the link's energy consumption threshold and multiple factors such as instantaneous load, load balancing, data flow density, and flow change rate. After calculation by the following formula, the energy consumption and load relationship of a link at the current moment can be intuitively presented.

[0100] Calculation process:

[0101] The first step is to put the sample values ​​obtained above into the formula:

[0102]

[0103] The second step is to calculate the internal terms in the numerator and denominator respectively:

[0104] (0.40+0.35)=0.75

[0105] (0.75) 2 =0.5625

[0106] |0.42-0.035|=0.385

[0107] 0.5625+0.385=0.9475

[0108] exp(-0.91)≈0.4022

[0109] 1+0.4022=1.4022

[0110]

[0111] So the main part of the denominator is:

[0112]

[0113] 0.9734+0.713=1.6864

[0114] The third step is to divide the numerator by the denominator:

[0115]

[0116] The results show that under the mutual influence of energy consumption and load, the energy consumption-to-load ratio of the example link is about 0.39. When the ratio is closer to 1, it means that a higher load and data traffic distribution can be supported under the same energy consumption threshold. When the ratio is smaller, it means that the match between the link load and energy consumption is relatively low. When sorting or filtering, the calculation results of other links can be combined for further comparison and summary.

[0117] Based on the energy consumption load ratio, after completing the numerical calculation of each link, it is necessary to arrange them one by one according to the link number and visually mark them. First, the calculation results of all links are recorded in the same structured list and the identifiers are matched with the corresponding values. When the energy consumption load ratio of a link is found to exceed 1.2, it can be regarded as a link with relatively low energy consumption and relatively high load capacity and marked with a separate color. At the same time, links with energy consumption load ratios lower than 0.3 should be extracted and compared with daily monitoring data. For example, the bandwidth utilization in the past 7 or 14 days is continuously tracked and compared with the reference load range set in the range of 0.2 to 0.5. If the actual load of the link is in the above reference range for a long time but still presents a low energy consumption load ratio, such as less than 0.3, then This indicates that there is an abnormality in the numerical setting of its energy consumption threshold or instantaneous throughput capacity, and it is necessary to re-detect the energy consumption threshold or instantaneous throughput capacity. In this case, the flow test can be repeated during high-load periods and the power consumption of the equipment can be measured with a power meter. The test is updated again according to the continuous monitoring results of no less than 2 hours each time. After the test is completed, the energy consumption threshold and instantaneous throughput capacity are re-counted and the old values ​​in the list are replaced for calculation again. By comparing the latest calculation results with the results of other links, it is observed whether there are significant changes in the ranking, such as from less than 0.3 to more than 0.5, and comparing whether the transmission data of the link in the actual operating period can be gradually increased. Finally, the final corrected energy consumption load ratio results of all links are integrated into a set and recorded to form a link weight factor set.

[0118] The steps to obtain the path optimization solution are:

[0119] Extract all link data in the link weight factor set, call the link delay value and bandwidth utilization, build an association table, filter out links with missing or abnormal data in the association table, and obtain a link performance data set;

[0120] Based on the link performance data set, the path fitness score of each link is calculated using the following formula:

[0121]

[0122] Among them, Z k represents the path fitness score of link k, X k represents the delay value of link k, Yk represents the bandwidth utilization of link k, M k represents the maximum throughput of link k, N k represents the data flow stability of link k at the current moment, O k Represents the real-time congestion status of link k;

[0123] Based on the path fitness score, all links are sorted according to the path fitness score, and the link with the highest path fitness score is selected to form a path optimization plan.

[0124] Specifically, all link data in the link weight factor set are extracted and the link delay value and bandwidth utilization are called. These values ​​are read one by one and compared with the link number. At the same time, a record entry is created for each link to mark the delay value range and bandwidth utilization distribution of the link. During the comparison process, the delay and bandwidth utilization thresholds extracted in advance are verified. For example, the delay value is compared with the range of 0ms to 200ms. When the delay value is greater than 200ms, the link is marked as a candidate for investigation. If the bandwidth utilization is greater than 0.85, the bandwidth availability fluctuation during this period is compared again. Records greater than 0.85 are included in the potential anomaly category for continuous tracking. After merging such annotation information of all links, these objects are examined one by one to see if there is a serious deviation in the delay value or the bandwidth utilization is much higher than the normal level. If some records have high delay and high bandwidth occupancy or data loss at the same time, they are marked as data missing or abnormal. Finally, all link entries that meet the requirements and are relatively stable within each range are screened out to obtain a link performance data set.

[0125] The benefit of the formula is that it incorporates the link's delay value, bandwidth utilization, maximum throughput, current data flow stability, and real-time congestion status into the same calculation process, and comprehensively evaluates the link's path adaptability through quantitative evaluation of multiple angles such as delay, bandwidth, throughput, and congestion.

[0126] X k The acquisition steps are as follows: During the 7 consecutive days of formal operation of the link, the round-trip delay is measured every 10 seconds, and these measured values ​​are stored as a time series. Then the average round-trip delay per hour is calculated, and the highest and lowest round-trip delay points per hour are counted. These data are compared with the distribution of the past 7 days. If the difference in delay between different time periods is less than 25ms, it is included in the intermediate stable range statistics, otherwise it is recorded as a peak or trough delay segment. Finally, multiple time periods are weighted and integrated to obtain the average round-trip delay for the whole day, and converted into a millisecond value X k For example, if the average daily delay of a link is about 30ms, it may rise to 60ms during peak hours and drop to 20ms during trough hours. After comprehensive weighting, we can get Xk =35ms.

[0127] Y k The steps to obtain y are as follows: in the same 7-day period, by recording the bandwidth occupancy rate per hour, the actual bandwidth value used per hour is divided by the maximum available bandwidth upper limit of the link, and then a bandwidth utilization sequence is obtained. If the bandwidth utilization rate appears more frequently in the range of 0.50 to 0.80, it is considered to be in the normal range. Otherwise, it is recorded as a possible high or low situation. In order to make the result more precise, it is necessary to additionally mark the peak and trough periods of the day. If the peak occupancy rate exceeds 0.90 for most of the time, an additional correction coefficient needs to be introduced to further smooth the sequence. Finally, a dimensionless value Y can be obtained after calculation. k For example, if the maximum bandwidth of a link is 800Mbps and the actual average bandwidth usage during the day is about 600Mbps, the bandwidth utilization rate is about 0.75. If the peak usage reaches 0.90 and the time period does not exceed 20%, then the final Y k =0.75.

[0128] M k The steps to obtain M are as follows: first query the maximum throughput performance marked by the manufacturer of the link equipment, and combine it with a continuous 2-hour test under high load, during which the throughput peak is recorded every 5 minutes. When the recorded peak value is close to the nominal maximum value for many times, the nominal value can be regarded as valid. If the peak value is generally lower than the nominal value by more than 10%, data correction can be introduced. By comparing the throughput data under high load test with the actual online state, the proportion of the average peak value to the nominal value is extracted, and the maximum throughput parameter is updated accordingly to form a dimensionless value M. k For example, the manufacturer's nominal maximum throughput is 900Mbps, and the average peak value is 810Mbps after testing, which is about 10% away from 900Mbps. If the difference remains within 10% for many days, then the M k It is recorded as 0.90, which means that the maximum throughput can reach about 90% of the nominal value.

[0129] N k The acquisition steps are as follows: record the data flow stability every hour during the 7-day monitoring period, first count the changes in the number of data packets per hour, if the frequency of data packet changes within 1 hour is greater than the fluctuation threshold of 10 times per minute, it is considered an unstable segment, and count the proportion of unstable segments in a whole day. If this proportion is between 10% and 20%, it is considered to be a medium stable level, otherwise these unstable segments need to be split and analyzed again. After aggregating data from multiple days, calculate the average duration ratio of stable segments in all monitoring periods, and then obtain a dimensionless data flow stability value N. kFor example, it is found that the unstable data flow period accounts for about 15% of the 5 days, and about 18% of the other 2 days. The weighted total proportion is 16%, corresponding to N k =0.84 (1 minus 0.16 gives 0.84, which means that the overall stability is 84% ​​of the time).

[0130] O k The acquisition steps are as follows: when real-time congestion monitoring is performed on the link, the packet loss rate and queue length are detected every 10 seconds, and the situation where the packet loss rate is greater than 2% or the queue length exceeds 1000 packets is determined as mild congestion, and the number of occurrences and duration are recorded. By summarizing the monitoring period of each day, the ratio of the total congestion duration to the total duration of the day is calculated as the basic congestion ratio, and then the peak frequency of congestion (multiple consecutive occurrences in one day) is combined to make corrections. If the corrected value is between 0.10 and 0.20, it is considered as moderate congestion, and the ratio is converted into a dimensionless value between 0 and 1. k For example, if the average congestion ratio for three consecutive days is 0.15, and the congestion peak occurs twice a day, each time lasting more than 5 minutes, then 0.15 is multiplied by the additional fluctuation coefficient of 0.90 to get about 0.135, which is recorded as O k =0.135.

[0131] Calculation process:

[0132] The first step is to obtain the parameters of the example link as described above:

[0133] X 1 =35,Y 1 =0.75,M 1 =0.90,N 1 =0.84,O 1 =0.135

[0134] The second step is to substitute the formula:

[0135]

[0136] The third step is to calculate the items:

[0137]

[0138] |0.90-0.84|=0.06

[0139]

[0140] 1+0.135=1.135

[0141]

[0142] ln(1+0.8816)=ln(1.8816)≈0.6313

[0143] Add them together:

[0144] 46.6667+0.2449+0.6313=47.5429

[0145] Then take the reciprocal:

[0146]

[0147] The result shows that in this example, the path fitness score of the link is about 0.021, which means that the comprehensive score of the link is relatively low when the latency is high, the bandwidth utilization is medium, the maximum throughput availability and data flow stability are good, and the congestion is light. When compared with the scores of other links, the feasibility of the link in path selection can be more intuitively evaluated. If the score of a link is significantly greater than 0.05, it means that it is relatively more adaptable in terms of factors such as latency and bandwidth.

[0148] Based on the path fitness score, after all links have completed the score calculation, a path fitness score list corresponding to multiple links will be obtained. In this list, the score value is first matched with the identification information of each link and entered, and then the scores are arranged in order from high to low and the partition threshold is set to distinguish the range of high scores and medium and low scores. For example, a score greater than 0.05 is considered a high score in the priority selection column, and a score between 0.02 and 0.05 is considered a medium optional range. Links with a score lower than 0.02 can be listed as alternatives or temporarily suspended. In the process of determining the score threshold, the delay stability and bandwidth usage of different score segments will be compared with historical monitoring data. If a link is found to have a low score, If the latency has been significantly higher than 80ms for the past 14 days or the bandwidth utilization has exceeded 0.90 multiple times in different time periods, it will be marked as a note item. These note items will be specifically extracted and combined with the previously formed score list for comparison to confirm whether the low score is due to bandwidth anomalies or extreme traffic fluctuations. When all links are assigned to their respective score segments, the link with the highest score will be re-verified. For example, whether the latency of the link has been maintained at around 40ms and the bandwidth utilization has not exceeded 0.85 in the past 7 days, and the throughput records during the peak hours in the morning, noon and evening will be summarized and checked. Finally, if there are no abnormalities in all aspects, the link with the highest score will be selected as the priority channel to form a path optimization plan.

[0149] The steps to obtain the service classification sequence are:

[0150] Based on the path optimization solution, the delay priority score of each data point is calculated using the following formula:

[0151]

[0152] Among them, V m represents the latency priority score of data point m, A m represents the real-time transmission delay of data point m, B m represents the maximum transmission delay of data point m, C m represents the congestion impact factor of data point m, D m represents the target link bandwidth utilization of data point m, E m represents the throughput of data point m at the current moment, F m represents the instantaneous load value of the link to which data point m belongs, G m Represents the load fluctuation amplitude of data point m, H m Represents the load stability of data point m at the current moment;

[0153] Based on the delay priority score, all data points are graded, and according to the sorting results of the delay priority score, the data points are divided into multiple service levels to form a service classification sequence.

[0154] Specifically, the formula is beneficial in that it incorporates real-time transmission delay, maximum transmission delay, congestion impact factor, target link bandwidth utilization, throughput capacity at the current moment, instantaneous load value of the link, load fluctuation amplitude and load stability. By performing comprehensive calculations on these relatively independent and interrelated indicators, the delay priority of each data point can be quantified.

[0155] A m The acquisition steps are as follows: within a continuous monitoring period of 7 days, the round-trip transmission delay of the current data point is recorded every 10 seconds to obtain a delay sequence in milliseconds. After removing the extreme abnormal records above 500ms, the hourly delay mean and delay peak are selected. By comparing the proportion of peak hours and low-peak hours throughout the day, combined with the common empirical thresholds of 200ms being considered normal and stable and 200ms to 400ms being considered acceptable in industrial network environments, if some records appear more frequently between 400ms and 500ms, these records are counted separately and compared with the distribution of other time periods. Finally, the average delay for the whole day or the whole cycle is calculated and corrected in combination with the changes to form the real-time transmission delay. The value range is usually around 30ms to 300ms. Finally, this average value is used as A. m For example, after 7 days of statistics on a data point, the effective delay value obtained is between 40ms and 280ms. The average value is about 120ms, which is not much different from other time periods. m Take it as 120.

[0156] B mThe steps for obtaining B are as follows: when evaluating the maximum delay limit that the data point can withstand, it is necessary to first query the delay limit description marked by the network equipment supplier, and combine it with the maximum delay test of the link where the data point is located. Usually, each link is monitored at least 3 times around the clock, and the delay peak is recorded at 10-second intervals each time. If multiple tests show that the delay peak can be kept within 500ms, 500ms is recorded as the preliminary value of the maximum delay; if there are individual time periods in some test cycles that obviously exceed 500ms, the correction coefficient is calculated according to the excess amplitude when summarizing. For example, if the delay is around 550ms for 3 consecutive times, it is increased by 5% to 525ms on the basis of 500ms, and finally a more reasonable maximum transmission delay is obtained, which is recorded as B m For example, the peak latency distribution of a data point in three tests is concentrated in the range of 495ms to 520ms. After comprehensive correction, m Set to 520.

[0157] C m The acquisition steps are as follows: the congestion impact factor needs to be quantified by combining data such as packet loss rate and queue length during the monitoring period. A 10-second sampling interval is set for each observation period to collect information on all possible congestion periods within 1 hour. The sampling points with packet loss rate exceeding 2% and queue length exceeding 1000 data packets are recorded. Then, the proportion of the total duration of congestion to the total duration of the hour is calculated. After accumulating the hourly proportion within 24 hours or more days, an original congestion ratio is obtained, and then linear mapping is performed according to the interval from 0 to 1. The mapping value is multiplied by a correction factor formulated according to the common load conditions of industrial networks. For example, after 5 consecutive days of observation, it is found that the total congestion period accounts for about 10%. Compared with the interval set between 0.0 and 0.3, 0.10 is converted into 0.33, and then combined with the correction coefficient 0.90 to obtain 0.297, which can finally be recorded as C m =0.297.

[0158] D m The acquisition steps are as follows: The target link bandwidth utilization is to measure the bandwidth occupancy of the data point under the current link. It is necessary to first obtain the upper limit of the available bandwidth of the link, which can usually be combined with the maximum available bandwidth monitored previously. Then, the actual bandwidth usage is recorded every 5 minutes for a period of time (for example, 7 days), and the ratio of the bandwidth usage in each period to the maximum available bandwidth is calculated. After generating the time series, the average value is taken as the preliminary ratio. If it is found that some periods exceed the mean value by more than 30% and last for more than 1 hour, this part of the value is recorded separately and the average value is corrected in a small range to obtain the final dimensionless bandwidth utilization value, which is recorded as D mFor example, the maximum available bandwidth of a link where a data point is located is 700Mbps. After monitoring, the average usage is about 420Mbps, and the initial value is 0.60. If the high usage period lasts for 1 hour, the proportion is small, then maintain 0.60 unchanged, and finally D m =0.60.

[0159] E m The steps to obtain E are as follows: the throughput capacity at the current moment needs to be extracted from the throughput upper limit supported by the link device and the real-time load status. Usually, the link is tested for high load for several consecutive days, and three to five peak hours are selected every day for recording. The actual throughput is compared with the nominal throughput of the device. If the comparison result continues to remain above 85% of the nominal value, the ratio of the average peak value of the actual test to the nominal value is regarded as the throughput capacity coefficient, and then converted into a dimensionless value as E m For example, if the nominal throughput is 800Mbps, and the average peak value in the 5-day test can reach 700Mbps, the proportion is about 700 / 800 = 0.875. If the proportion is stable between 85% and 90%, then E m =about 0.88.

[0160] F m The steps for obtaining are as follows: the instantaneous load value of the link needs to be integrated with its load conditions in the peak and valley intervals during the day, and the hourly occupancy rate is obtained by statistically analyzing the traffic proportion every hour and combining it with the bandwidth limit. These occupancy rates are then weighted to reflect the impact of the peak period. If the peak period lasts for more than 3 hours and the occupancy rate is significantly higher than the average by more than 30%, the peak period data is extracted and corrected separately and then merged with the rest of the period to obtain a dimensionless value F that represents the instantaneous load occupancy of the link. m For example, if we record the load factor every hour from 0:00 to 24:00 in a day, and finally find that the average load factor for the whole day is about 0.50, and the occupancy rate during the peak period reaches 0.75 and lasts for 4 hours, then we can separate this period from the average value for the whole day and make corrections, and get F after statistics. m =0.55.

[0161] G m The steps to obtain the load fluctuation amplitude are as follows: The load fluctuation amplitude needs to analyze the degree of load change within each hour. In the sampling data of each hour, if the difference in occupancy between two adjacent sampling points is greater than 10%, it is counted as a fluctuation. Then the number of fluctuations within an hour is counted and divided by the total number of samplings within the hour to form a preliminary fluctuation ratio, which is then compared with the historical average fluctuation level of the past 7 days. The part that is significantly higher than the average level is marked separately, and a linear mapping is performed on it through a pre-set range of 0 to 0.5 to obtain a fluctuation amplitude value G. mIf the fluctuation exceeds 0.5 in a certain period, an additional correction factor is multiplied by 0.95 or a similar method. Complete example: 30 samples are taken in one hour, and 8 of them have a load difference of more than 10%, 8 / 30 = 0.2667, and then refer to the average fluctuation level of the past 7 days of about 0.20. If the difference is large, it is corrected to about 0.30, recorded as G m =0.30.

[0162] H m The steps to obtain the load stability are as follows: the load stability corresponds to the fluctuation amplitude. It is necessary to calculate the variance or standard deviation of each load record within a period of time. If the standard deviation within this period of time is low, it means that the load distribution is relatively stable. Then, the dimensionless stability value H can be obtained by subtracting the ratio of the standard deviation occupancy rate from 1. m , usually all sampling points in the past 24 hours are processed. For example, the load standard deviation in this period is 0.12, which is divided by the set maximum allowable standard deviation of 0.50 to get 0.24, and then 1 minus 0.24 = 0.76, recorded as H m =0.76. If a larger range of fluctuations is encountered, segmented corrections can also be made between the upper and lower limits of the maximum allowable standard deviation.

[0163] Calculation process:

[0164] In the first step, assume that the parameters of a data point m have been determined according to the above acquisition steps:

[0165] A m =120,B m =520,C m =0.297,D m =0.60,E m =0.88,F m =0.55,G m

[0166] =0.30,H m =0.76

[0167] The second step is to bring the parameters into the formula one by one:

[0168]

[0169] The third step is to calculate the local terms first:

[0170] ln(1+120)=ln(121)≈4.7958

[0171] 520 2 =270400

[0172]

[0173] exp(-0.88)≈0.4149

[0174] 1+0.4149=1.4149

[0175]

[0176] 0.30-0.76=-0.46

[0177] |-0.46|=0.46

[0178] Add the numerator parts:

[0179] 4.7958+520.3848+0.4242=525.6048

[0180] Denominator:

[0181] 0.55+0.46=1.01

[0182] Finally take the ratio:

[0183]

[0184] The result shows that at this sample data point, the latency priority score is about 520.40, which means that this data point is at a relatively high priority level in the overall structure of latency-related factors. If someone selects a numerical range, such as more than 300 is considered high priority, between 100 and 300 is medium priority, and less than 100 is normal priority, then this data point has a high priority score due to its high latency, moderate bandwidth utilization, and the combined effect of other factors.

[0185] Based on the delay priority score, after all data points have completed the calculation and the corresponding score sequence is initially obtained, it is necessary to read the score values of each data point one by one. First, pair the scores with the identification information of the corresponding data points and sort them from high to low. During the sorting process, scores exceeding 200 are marked as the high-priority range and separated by a separate color or mark. Scores in the range of 50 to 200 are marked as the intermediate service range. If the score is below 50, the data point is listed in the low-priority range. When determining the score boundary value, the monitoring results of three days or more can be referred to compare the delay distribution and bandwidth occupancy in different time periods. For example, first set the score range of 50 to 200 as the basic service level. If it is found during monitoring that the scores of some data points fluctuate around 50 and the bandwidth occupancy rate is often higher than 0.90, it is necessary to perform repeated sampling on them to check for instantaneous high-delay phenomena. When it is confirmed that the situations of these data points are relatively stable, these scores are recorded in the same list and corresponding marks are added to the records with congestion doubts. Finally, after all scores are classified, further screening will be carried out on the data points within the high-priority range to check whether their delays and throughput capabilities remain consistent within 24 hours. If no additional abnormalities are found, these data points will be included in the high-level services. Then, the data point identification and score range information will be summarized separately in all the divided levels, thereby performing multiple-level classification on the data points and forming a service classification sequence.

[0186] The steps to obtain the slice scheduling strategy are as follows:

[0187] Extract each data category in the service classification sequence, call the network slice resource pool, match according to the priority of the service level, screen available network slices, and establish a mapping relationship for different data categories to generate a preliminary network slice allocation plan;

[0188] Based on the preliminary network slice allocation plan, call the bandwidth utilization rate, delay bearing capacity, and load balancing value of the network slice, and adjust the allocation ratio of the service classification sequence according to the resource bearing capacity of each network slice to generate an optimized network slice allocation plan;

[0189] Based on the optimized network slice allocation plan, schedule the network slice resources according to the traffic change situation of the service data to form a slice scheduling strategy.

[0190] Specifically, to extract each data category in the service classification sequence, first read the priority level and data volume range of each data category from the previously obtained service classification sequence, and decompose and extract these data categories with reference to the pre-agreed matching criteria. For example, for the high-priority category, its average bandwidth occupancy rate and delay fluctuation data per unit time will be extracted, and the medium and low-priority categories will extract the data volume statistics in regular time periods and low-load time periods respectively. After the decomposition is completed, the network slice resource pool is called and compared in sequence according to the priority of the service level. For example, the high-priority category is screened according to the delay stability coefficient exceeding 0.70 and the bandwidth occupancy ratio threshold of 0.80. If it meets the requirements, it is matched to the slice with a larger relative bandwidth margin for subsequent registration. At the same time, the slices that do not meet the requirements are The category of bandwidth or delay requirements is searched again and the bandwidth available value and delay limit standard of the next slice are compared by parameter calibration. When the match is completed, the corresponding business category and the selected slice information are sorted and recorded one by one in the same mapping list. Then, the successful matching results are merged in descending order according to the priority of the data category to form a preliminary mapping relationship. In this process, if it is found that a certain data category has not found suitable bandwidth or delay support in all network slices, it will be marked as temporarily reserved. Later, it will be checked in coordination whether the remaining capacity of the slice resource pool can be expanded. Finally, the slice number and estimated bandwidth allocation ratio corresponding to each business category can be seen in the preliminary mapping relationship, thereby generating a preliminary network slice allocation plan.

[0191] Based on the preliminary network slice allocation plan, first read the correspondence between the service category and the slice formed in the mapping list before, and call the three indicators of the network slice's bandwidth utilization, delay carrying capacity and load balancing value for cross-examination. The bandwidth utilization can be in the range of 0.50 to 0.85 as the normal usage range. If it exceeds 0.85, the slice will be marked as resource-constrained in the subsequent review. The delay carrying capacity is judged against the range of 20ms to 80ms. If the delay carrying capacity of a slice is above 80ms, it is classified as a high-delay mark. The setting of the load balancing value needs to be corrected according to the previously statistical hourly traffic distribution curve and the occupancy during peak hours. For example, if the occupancy rate is higher than 0.90 multiple times in a day, the load balancing value will be The value is lowered to below 0.40. If there are only one or two occupancy peaks and the duration does not exceed one hour, it can be maintained at around 0.60. By matching these values ​​with the needs of existing business categories, it can be found that some high-priority categories will put greater pressure on the slice for low latency requirements, and some medium-priority categories only require a stable load balancing value to meet the needs. Based on this information, the allocation ratio of the business classification sequence can be fine-tuned. For example, the amount of data in some high-priority categories can be reduced by 10% and a certain amount of bandwidth can be freed up for those medium-priority categories. For low-priority categories, if it is found that the remaining bandwidth or load balancing value of the slice can still accommodate it, it will continue to be allocated according to the original ratio. Finally, the above adjustment results are integrated and recorded as the optimized network slice allocation plan.

[0192] Based on the optimized network slice allocation scheme, we first check the recent traffic changes of business data one by one. For example, we count the bandwidth usage curve and latency fluctuation range for high-priority data points in the past two hours, and compare the curve with the pre-defined bandwidth usage reference range of 0.60 to 0.90. When it is found that there is a period of more than 0.90 for 20 consecutive minutes, the corresponding slice is marked as overloaded. At the same time, the instantaneous throughput and average load level of medium-priority data points are inspected. If the load balancing value of a slice is once lower than 0.40, it means that latency accumulation may occur during peak hours. For these situations, a portion of the slice can be recorded after the recording is completed. High-priority traffic can be temporarily switched to slices with relatively light loads. The transmission period of some medium-priority data can also be postponed by half an hour to the off-peak period. If it is detected that low-priority data points still occupy a large bandwidth late at night, their average occupancy is calculated and compared with the pre-set bandwidth matching range. If it exceeds 0.80, a small-scale scheduling is performed within the slice. When this scheduling operation is confirmed to be feasible and will not cause conflicts with other data categories, the switch is executed and the traffic trend is observed for at least 15 minutes. After confirming that the traffic curve gradually falls back, the completion result of this slice scheduling is saved in the record. After multiple operations of the same type, a slice scheduling strategy is finally formed.

[0193] The steps for obtaining the routing control instruction are as follows:

[0194] Evaluate the network slice resource configuration in the slice scheduling policy, extract the traffic requirements of each service classification sequence, determine the matching degree between each service requirement and the network slice resources, and generate a service requirement and network slice matching table;

[0195] Based on the service requirement and network slice matching table, allocate the data traffic under each service classification sequence to the network slice resources to obtain a traffic allocation plan, and generate a routing control instruction according to the traffic allocation plan.

[0196] Specifically, to evaluate the network slice resource configuration in the slice scheduling policy, first disassemble and extract the previously obtained network slice configuration information and service classification sequence, and retrieve the traffic requirements of each service classification sequence in the past 24 hours one by one, including the peak bandwidth occupancy period and delay distribution. Then compare these data with the current available bandwidth, delay bearing margin, and load change records in the corresponding network slice. During the comparison process, it is necessary to conduct itemized statistics on the data scales included in all service classification sequences. For example, calculate the total traffic value of high-priority services during an 8-hour peak period and compare it with the pre-set available bandwidth ratio range of 0.60 to 0.90. If it is found that it exceeds 0.90, it is classified as a situation that may cause load tension. At the same time, for medium-priority and low-priority services, they should also be judged against the available bandwidth ratio ranges of 0.20 to 0.60 and 0.10 to 0.30 respectively. When the demand of any service continuously reaches the upper limit of these intervals for 30 minutes, record its tendency to cause congestion. Refer to this record to mark potential overlapping conflicts in the network slice resource configuration table. When multiple services are running in parallel, check whether there are periods exceeding the predetermined standard according to the specific traffic values. For example, if it is set that the transmission volume of more than 5GB of traffic within one hour requires additional marking. Here, 5GB is a range formulated based on the peak throughput of the device and daily observations, with a certain redundancy reserved. If it is detected that the above conditions are met, add a conflict identifier to the matching information. Finally, integrate and match the traffic requirement statistics and network slice resource bearing information in the above process, obtain and summarize the corresponding relationships between all service requirements and network slice resources, and record the matching degree among them to generate a service requirement and network slice matching table.

[0197] Based on the service demand and network slice matching table, the marked service classification sequence is read in turn and data traffic is allocated for the corresponding network slice resources. The peak traffic of high-priority services is first calculated according to the traffic value and available bandwidth ratio indicated above. If a high-priority service has a bandwidth demand of about 0.70 and keeps the delay below 40ms within two hours, it is confirmed that the service occupancy ratio meets the allocation requirements. If the medium-priority service has a delay demand of about 80ms, it is placed in a slice with a certain availability and a load balancing level of not less than 0.50. For low-priority services, if their traffic demand is small and can be transmitted in a concentrated manner during the late night, It is allocated to slices with more idle time. During this process, for each type of business, it is necessary to check whether its peak load and daily fluctuation range are close to the preset threshold. For example, if the fluctuation value exceeds 0.30 multiple times within an hour, a correction coefficient will be introduced to reduce the allocation ratio by 5% to 10%, and it will be marked in combination with the allocation record. When the data traffic under all business classification sequences obtains the corresponding slice resources, the traffic allocation plan can be obtained. Subsequently, according to the traffic allocation plan, routing control instructions can be generated in sequence and the link path associated with each slice can be recorded. After completing this step, each business classification sequence has corresponding slice resource allocation and has executable routing control instructions.

Claims

1. Photovoltaic power station data transmission and monitoring system based on 5G technology, characterized by: The system comprises: The routing status monitoring module collects and counts the delay value, bandwidth utilization rate and congestion level value of each transmission link in the 5G base station service area to generate a link status parameter set; calculates the link status parameter set with the network load balancing value and energy consumption threshold to obtain resource utilization efficiency and generate a routing resource monitoring matrix; A path optimization calculation module calculates the energy consumption load ratio of each link based on the routing resource monitoring matrix and generates a link weight factor set; based on the link weight factor set, selects a link according to the delay value and bandwidth utilization and generates a path optimization solution; The network slice scheduling module, based on the path optimization scheme, classifies the photovoltaic power station control data according to the delay value to generate a service classification sequence; allocates independent network slice resources to the service classification sequence to generate a slice scheduling strategy; The data forwarding control module allocates the data traffic under each service classification sequence to the corresponding network slice resources based on the slice scheduling strategy and generates routing control instructions.

2. According to the 5G technology-based photovoltaic power station data transmission and monitoring system of claim 1, it is characterized in that: The steps of obtaining the link state parameter set are: Deploy sensors on each transmission link in the 5G base station service area to monitor and record the link latency, bandwidth utilization, and congestion level in real time to obtain preliminary transmission link performance data; Based on the preliminary transmission link performance data, outliers are removed, data is normalized and format is standardized to generate a link state parameter set.

3. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps of obtaining the routing resource monitoring matrix are: Based on the link state parameter set, the resource utilization efficiency index of each link is calculated, and the calculation formula is: in, represents the resource utilization efficiency index of link i, represents the load balancing value of link i, represents the bandwidth utilization of link i, represents the energy consumption threshold of link i, represents the instantaneous transmission rate of link i, represents the congestion frequency of link i, represents the maximum throughput of link i, Represents the available bandwidth ratio of link i at the current moment; Based on the resource utilization efficiency index, the links are sorted according to their resource utilization efficiency index, a matrix model is constructed, a mapping relationship between resource distribution and monitoring status of all links is formed, and a routing resource monitoring matrix is ​​obtained.

4. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps of obtaining the link weight factor set are: According to the routing resource monitoring matrix, the energy consumption load ratio of each link is calculated, and the calculation formula is: Among them, W j represents the energy consumption load ratio of link j, K j represents the energy consumption threshold of link j, P j represents the instantaneous load value of link j, R j represents the load balancing value of link j, S j represents the data flow density of link j, T j represents the traffic change rate of link j, U j represents the instantaneous throughput capacity of link j; Based on the energy consumption load ratio, all calculation results are integrated according to the link number, and the ratio data of all links are summarized into a set to form a link weight factor set.

5. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps for obtaining the path optimization solution are: Extract all link data in the link weight factor set, call the delay value and bandwidth utilization of the link, build an association table, filter the links with missing or abnormal data in the association table, and obtain a link performance data set; Based on the link performance data set, the path fitness score of each link is calculated using the following formula: Among them, Z k represents the path fitness score of link k, X k represents the delay value of link k, Y k represents the bandwidth utilization of link k, M k represents the maximum throughput of link k, N k represents the data flow stability of link k at the current moment, O k Represents the real-time congestion status of link k; Based on the path fitness score, all links are sorted according to the path fitness score, and the link with the highest path fitness score is selected to form a path optimization plan.

6. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps for obtaining the service classification sequence are as follows: Based on the path optimization solution, the delay priority score of each data point is calculated using the following formula: Among them, V m represents the latency priority score of data point m, A m represents the real-time transmission delay of data point m, B m represents the maximum transmission delay of data point m, C m represents the congestion impact factor of data point m, D m represents the target link bandwidth utilization of data point m, E m represents the throughput of data point m at the current moment, F m represents the instantaneous load value of the link to which data point m belongs, G m Represents the load fluctuation amplitude of data point m, H m Represents the load stability of data point m at the current moment; Based on the delay priority score, all data points are graded, and according to the sorting results of the delay priority score, the data points are divided into multiple service levels to form a service classification sequence.

7. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps for obtaining the slice scheduling strategy are as follows: Extract each data category in the service classification sequence, call the network slice resource pool, match according to the service level priority, screen the available network slices, establish a mapping relationship between different data categories, and generate a preliminary network slice allocation plan; Based on the preliminary network slice allocation plan, the bandwidth utilization, latency carrying capacity and load balancing value of the network slice are called, and the allocation ratio of the service classification sequence is adjusted according to the resource carrying capacity of each network slice to generate an optimized network slice allocation plan; Based on the optimized network slice allocation scheme, network slice resources are scheduled according to the traffic changes of business data to form a slice scheduling strategy.

8. The photovoltaic power station data transmission and monitoring system based on 5G technology according to claim 1 is characterized in that: The steps of obtaining the routing control instruction are as follows: Evaluate the network slice resource configuration in the slice scheduling strategy, extract the traffic demand of each service classification sequence, determine the matching degree between each service demand and the network slice resource, and generate a service demand and network slice matching table; Based on the business demand and network slice matching table, the data traffic under each business classification sequence is allocated to the network slice resources to obtain a traffic allocation plan, and routing control instructions are generated according to the traffic allocation plan.

Images