5G base station energy-saving evaluation method based on AI intelligent control

Abstract

The invention provides a 5G base station energy-saving evaluation method based on AI intelligent control, and the method comprises the steps: recognizing the feature data corresponding to different scenes of a cell where a 5G base station is located, carrying out the classification of the feature data of all scenes based on an AI improved clustering algorithm, obtaining a plurality of data sets corresponding to different scenes, and providing a data basis for a subsequent training system efficiency evaluation index model; training a system efficiency evaluation index model to obtain a feasibility evaluation result of performing 5G base station equipment energy-saving operation on a cell in a corresponding scene, performing accurate prediction on feasibility of performing different types of software energy-saving operation in a current scene, and providing a reference basis for subsequent real execution of the software energy-saving operation; and on the basis of the feasibility evaluation result, the execution state information of the energy-saving operation of the 5G base station equipment on the cell in the corresponding scene is determined, so that an energy-saving operation scheme about the 5G base station is generated, and the energy-saving operation efficiency and reliability are improved on the premise of not influencing the key performance index of service provided by the network.

Description

Technical Field

[0001] The present invention relates to the field of 5G base stations, and in particular to a method for energy-saving evaluation of 5G base stations based on AI intelligent control. Background Art

[0002] With the advent of the 5G era, various new services are emerging one after another, and the development trend of device connections towards massive connections is irreversible, which continues to promote the explosive growth of mobile data traffic. In order to meet the development requirements of the rapid growth of 5G services and data traffic, the power consumption of 5G networks has also increased, which has further aggravated the contradiction between the data growth and high energy consumption of 5G mobile communication networks. It can be seen that energy saving and consumption reduction are important issues that the sustainable development of the future 5G mobile communication industry must face.

[0003] From the perspective of operation and maintenance, energy conservation is not just about reducing energy consumption. The purpose of energy conservation is to achieve a compromise between the energy consumption generated by network operation and the service performance provided by the network, that is, to reduce energy consumption without affecting the key performance indicators of the network's service provision. Energy-saving operations using AI technology combined with network data can help improve the energy efficiency of network terminals. According to indicators such as 5G base stations or network service changes, intelligent matching and support of energy-saving strategies and the formulation of efficient and flexible energy-saving plans are important directions for the future development of 5G networks. Summary of the invention

[0004] The purpose of the present invention is to provide a method for energy-saving evaluation of 5G base stations based on AI intelligent control, identify the characteristic data of the cell where the 5G base station is located corresponding to different scenarios, and classify the characteristic data of all scenarios based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenarios, so as to provide a data basis for the subsequent training system performance evaluation index model; the training system performance evaluation index model can obtain the feasibility evaluation results of energy-saving operations of 5G base station equipment in the cell in the corresponding scenario, accurately predict the feasibility of performing different types of software energy-saving operations in the current scenario, and provide a reference basis for the subsequent actual execution of software energy-saving operations; based on the feasibility evaluation results, the execution status information of the energy-saving operations of 5G base station equipment in the cell in the corresponding scenario is determined, so as to generate an energy-saving operation plan for the 5G base station, and improve the efficiency and reliability of energy-saving operations without affecting the key performance indicators of network service provision.

[0005] The present invention is achieved through the following technical solutions:

[0006] The method of energy-saving evaluation of 5G base stations based on AI intelligent control includes:

[0007] Perform scene recognition on the cell where the 5G base station is located to obtain feature data corresponding to different scenes of the cell; classify the feature data of all scenes based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenes;

[0008] Constructing a system performance evaluation index model, and training the system performance evaluation index model based on the data set to obtain a feasibility evaluation result of performing energy-saving operations of 5G base station equipment on the cell in the corresponding scenario;

[0009] Based on the feasibility assessment result, determine the execution status information of the energy-saving operation of the 5G base station equipment for the cell in the corresponding scenario; based on the execution status information, generate an energy-saving operation plan for the 5G base station.

[0010] Optionally, scene recognition is performed on the cell where the 5G base station is located to obtain feature data of the cell corresponding to different scenes, including:

[0011] Analyze the historical 5G network operation logs of the cell where the 5G base station is located to obtain historical operation scenario type information of the 5G base station in the cell; based on the historical original scenario type information, divide the historical 5G network operation logs into scenario operation time intervals to obtain scenario sub-logs matching each historical operation scenario type;

[0012] Data recognition is performed on each scene sub-log to obtain feature data corresponding to the corresponding scene of the cell.

[0013] Optionally, analyzing the historical 5G network operation log of the cell where the 5G base station is located to obtain historical operation scenario type information of the 5G base station in the cell, including:

[0014] The historical 5G network operation logs of the cell where the 5G base station is located are analyzed to obtain type information corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario during the historical operation of the 5G base station in the cell.

[0015] Optionally, before analyzing the historical 5G network operation logs of the cell where the 5G base station is located, the data quality of the historical 5G network operation logs is determined, including:

[0016] Before analyzing the historical 5G network operation log of the cell where the 5G base station is located, the historical 5G network operation log is divided according to a preset unit time to obtain historical 5G network operation log data corresponding to multiple unit times, and the value of the unit time is 24 hours;

[0017] Scan the historical 5G network operation log data corresponding to each unit time to obtain the data volume corresponding to the abnormal data and garbled data in the historical 5G network operation log;

[0018] Obtaining a first evaluation parameter using the data amount corresponding to the abnormal data and the garbled data;

[0019] The first evaluation parameter is obtained by the following formula:

[0020]

[0021] Among them, F 01 represents the first evaluation parameter; n represents the total number corresponding to the unit time; C yi represents the amount of abnormal data corresponding to the i-th unit time; C mi represents the amount of garbled data corresponding to the i-th unit time; C yb represents the standard deviation of the amount of abnormal data corresponding to n unit time; C mb Indicates the standard deviation of the amount of garbled data corresponding to n unit time; C zi Represents the total amount of data corresponding to the i-th unit time;

[0022] Extracting the weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log;

[0023] Obtain a second evaluation parameter using a weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log;

[0024] The second evaluation parameter is obtained by the following formula:

[0025]

[0026] Among them, F 02 represents the second evaluation parameter; k represents the total number of abnormal data; g represents the total number of garbled data; w i represents the weight value of the i-th abnormal data; w b represents the standard deviation of the weight values ​​of n abnormal data; w max Indicates the maximum weight value of n abnormal data; s i Indicates the weight value of the i-th garbled data; s b Indicates the standard deviation of the weight values ​​of n garbled data; s max Indicates the maximum weight value of n garbled data;

[0027] Obtaining a data quality evaluation coefficient using the first evaluation parameter and the second evaluation parameter;

[0028] The data quality evaluation coefficient is obtained by the following formula:

[0029]

[0030] Where H represents the data quality evaluation coefficient; H0 represents the preset data quality evaluation coefficient threshold; F 01 Represents the first evaluation parameter; F 02 Represents the second evaluation parameter; F 01c Indicates the preset first evaluation parameter reference value; F 02c represents a preset second evaluation parameter reference value;

[0031] Comparing the data quality evaluation coefficient with a preset quality evaluation coefficient threshold;

[0032] When the data quality evaluation coefficient exceeds the preset quality evaluation coefficient threshold, the quality of the historical 5G network operation log is determined to be abnormal, and an abnormal quality warning of the historical 5G network operation log is issued.

[0033] Optionally, data identification is performed on each scene sub-log to obtain feature data of the cell corresponding to the corresponding scene, including:

[0034] Data identification is performed on the scenario sub-logs corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario, and the topology information, uplink and downlink measurement information, service category information, user level information and resource occupancy distribution information corresponding to the cell in the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario are obtained as the feature data.

[0035] Optionally, based on an AI-improved clustering algorithm, the feature data of all scenes are classified to obtain several data sets corresponding to different scenes, including:

[0036] Based on the K-means improved clustering algorithm, the feature data of all scenarios are initially classified and modified to obtain several data sets corresponding to multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios.

[0037] Optionally, a system performance evaluation index model is constructed, and based on the data set, the system performance evaluation index model is trained to obtain a feasibility evaluation result of performing energy-saving operation of a 5G base station device on the cell in a corresponding scenario, including:

[0038] Based on the system component structure of 5G base stations, a system performance evaluation index model is constructed;

[0039] Based on the data set, the system performance evaluation index model is trained to obtain feasibility evaluation results of performing energy-saving operations of 5G base station equipment on the cell in multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios; wherein the feasibility evaluation results include component system performance evaluation results of energy-saving operations of 5G base station equipment, component performance effectiveness evaluation results, networking effectiveness evaluation results, emergent effectiveness evaluation results and system mission task effectiveness evaluation results.

[0040] Optionally, based on the feasibility assessment result, determining execution status information of performing energy-saving operation of a 5G base station device on the cell in a corresponding scenario includes:

[0041] Based on the feasibility assessment result, determine the hardware resource allocation execution status information of the 5G base station equipment software energy-saving operation for the cell in any one of the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario.

[0042] Optionally, based on the execution status information, generating an energy-saving operation plan for the 5G base station includes:

[0043] Based on the hardware resource allocation execution status information, a software energy-saving operation plan for the 5G base station is generated; wherein the software energy-saving operation plan includes execution mode information and execution time information of the software energy-saving operation.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The method for energy-saving evaluation of 5G base stations based on AI intelligent control provided in the present application identifies the characteristic data of the cell where the 5G base station is located corresponding to different scenarios, and classifies the characteristic data of all scenarios based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenarios, thereby providing a data basis for the subsequent training system performance evaluation index model; the training system performance evaluation index model obtains the feasibility evaluation results of energy-saving operations of 5G base station equipment in the cell in the corresponding scenario, accurately predicts the feasibility of performing different types of software energy-saving operations in the current scenario, and provides a reference basis for the subsequent actual execution of software energy-saving operations; based on the feasibility evaluation results, the execution status information of the energy-saving operations of 5G base station equipment in the cell in the corresponding scenario is determined, thereby generating an energy-saving operation plan for the 5G base station, thereby improving the efficiency and reliability of energy-saving operations without affecting the key performance indicators of network service provision. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0047] Figure 1 A flowchart of the method for energy-saving evaluation of 5G base stations based on AI intelligent control provided by the present invention. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some structures related to the present application are shown in the accompanying drawings, rather than all structures. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0049] The terms "include" and "have" and any variations thereof in this application are intended to cover non-exclusive inclusions. For example, a process, method, method, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0050] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0051] See also Figure 1 As shown, an embodiment of the present application provides a method for energy-saving evaluation of a 5G base station based on AI intelligent control. The method for energy-saving evaluation of a 5G base station based on AI intelligent control includes:

[0052] Perform scene recognition on the cell where the 5G base station is located to obtain the feature data of the cell corresponding to different scenes; classify the feature data of all scenes based on the AI-improved clustering algorithm to obtain several data sets corresponding to different scenes;

[0053] Construct a system performance evaluation index model, and train the system performance evaluation index model based on the data set to obtain a feasibility evaluation result of performing energy-saving operations of 5G base station equipment in the cell in the corresponding scenario;

[0054] Based on the feasibility assessment result, determine the execution status information of the energy-saving operation of the 5G base station equipment of the cell in the corresponding scenario; based on the execution status information, generate an energy-saving operation plan for the 5G base station.

[0055] The beneficial effects of the above embodiments are as follows: the method for energy-saving evaluation of 5G base stations based on AI intelligent control identifies the characteristic data of the cell where the 5G base station is located corresponding to different scenarios, and classifies the characteristic data of all scenarios based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenarios, providing a data basis for the subsequent training system performance evaluation index model; the training system performance evaluation index model obtains the feasibility evaluation results of energy-saving operations of 5G base station equipment in the cell in the corresponding scenario, and accurately predicts the feasibility of performing different types of software energy-saving operations in the current scenario, providing a reference basis for the subsequent actual execution of software energy-saving operations; based on the feasibility evaluation results, the execution status information of the energy-saving operations of 5G base station equipment in the cell in the corresponding scenario is determined, so as to generate an energy-saving operation plan for the 5G base station, thereby improving the efficiency and reliability of energy-saving operations without affecting the key performance indicators of network service provision.

[0056] In another embodiment, scene recognition is performed on the cell where the 5G base station is located to obtain feature data corresponding to different scenes of the cell, including:

[0057] Analyze the historical 5G network operation logs of the cell where the 5G base station is located to obtain the historical operation scenario type information of the 5G base station in the cell; based on the historical original scenario type information, divide the historical 5G network operation logs into scenario operation time intervals to obtain scenario sub-logs matching each historical operation scenario type;

[0058] Data recognition is performed on each scene sub-log to obtain the feature data corresponding to the corresponding scene in the cell.

[0059] The beneficial effects of the above embodiments are that the energy-saving scheme of 5G base stations mainly includes hardware energy-saving schemes and software energy-saving schemes; among them, the hardware energy-saving scheme mainly reduces the basic power consumption of 5G base station equipment, and the software energy-saving scheme mainly reasonably allocates hardware resources from the perspective of business operation, so that 5G base stations can operate more efficiently. Specifically, for the hardware energy-saving scheme, the goal of reducing the basic energy consumption of base station equipment can be achieved by optimizing the hardware design of the equipment, using advanced production processes, and improving the integration of base station equipment. For example, the integration of key components can be improved, and FPGA components can be replaced by ASIC devices; in terms of processing technology, core components can use 10nm or 7nm processes, or even 5nm or 3nm processes, to improve the integration of single chips, which is conducive to optimizing IC design, thereby reducing the basic power consumption of base station equipment. It can be seen that the hardware energy-saving scheme can only be implemented when the 5G base station equipment is installed. If the hardware energy-saving scheme needs to be adjusted later, it can only be achieved by replacing the corresponding hardware equipment. This not only increases the cost of adjusting the hardware energy-saving scheme, but also reduces the flexibility of adjusting the hardware energy-saving scheme. It cannot guarantee that the hardware energy-saving scheme can be targeted and matched in different scenarios. Energy-saving operations, it can be seen that the hardware energy-saving scheme has limited room for energy-saving improvement for 5G base stations.

[0060] Compared with the hardware energy-saving solution, the software energy-saving solution has great advantages in terms of cost and implementation flexibility, and has a large room for energy-saving improvement for 5G base stations. For this reason, the software energy-saving solution is selected as the main way to operate 5G base stations in an energy-saving manner. In order to ensure that the 5G base station can perform accurate energy-saving operations when facing different scenarios, it is necessary to accurately identify the scenario conditions of the cell where the 5G base station is located in advance. Specifically, the historical operation scenario type information of the 5G base station in the cell is obtained from the historical 5G network operation log of the cell where the 5G base station is located, and the historical 5G network operation log is divided into scene operation time intervals to obtain scene sub-logs that match each historical operation scene type. In this way, the operation log parts corresponding to different scenarios can be accurately obtained, providing a basis for the subsequent extraction of feature data of the corresponding scenes.

[0061] In another embodiment, before analyzing the historical 5G network operation log of the cell where the 5G base station is located, determining the data quality of the historical 5G network operation log includes:

[0062] Before analyzing the historical 5G network operation log of the cell where the 5G base station is located, the historical 5G network operation log is divided according to a preset unit time to obtain historical 5G network operation log data corresponding to multiple unit times, and the value of the unit time is 24 hours;

[0063] Scan the historical 5G network operation log data corresponding to each unit time to obtain the data volume corresponding to the abnormal data and garbled data in the historical 5G network operation log;

[0064] Obtaining a first evaluation parameter using the data amount corresponding to the abnormal data and the garbled data;

[0065] The first evaluation parameter is obtained by the following formula:

[0066]

[0067] Among them, F 01 represents the first evaluation parameter; n represents the total number corresponding to the unit time; C yi represents the amount of abnormal data corresponding to the i-th unit time; C mi Indicates the amount of garbled data corresponding to the i-th unit time; C yb Indicates the standard deviation of the amount of abnormal data corresponding to n units of time; C mb Indicates the standard deviation of the amount of garbled data corresponding to n unit time; C zi Represents the total amount of data corresponding to the i-th unit time;

[0068] Extracting the weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log;

[0069] Obtain a second evaluation parameter using a weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log;

[0070] The second evaluation parameter is obtained by the following formula:

[0071]

[0072] Among them, F 02 represents the second evaluation parameter; k represents the total number of abnormal data; g represents the total number of garbled data; w i represents the weight value of the i-th abnormal data; w b represents the standard deviation of the weight values ​​of n abnormal data; w max Indicates the maximum weight value of n abnormal data; s i Indicates the weight value of the i-th garbled data; s b Indicates the standard deviation of the weight values ​​of n garbled data; s max Indicates the maximum weight value of n garbled data;

[0073] Obtaining a data quality evaluation coefficient using the first evaluation parameter and the second evaluation parameter;

[0074] The data quality evaluation coefficient is obtained by the following formula:

[0075]

[0076] Where H represents the data quality evaluation coefficient; H0 represents the preset data quality evaluation coefficient threshold; F 01 Represents the first evaluation parameter; F 02 Represents the second evaluation parameter; F 01c Indicates the preset first evaluation parameter reference value; F 02c represents a preset second evaluation parameter reference value;

[0077] Comparing the data quality evaluation coefficient with a preset quality evaluation coefficient threshold;

[0078] When the data quality evaluation coefficient exceeds the preset quality evaluation coefficient threshold, the quality of the historical 5G network operation log is determined to be abnormal, and an abnormal quality warning of the historical 5G network operation log is issued.

[0079] The beneficial effect of the above embodiment is that by dividing the historical 5G network operation log into 24-hour units, the data situation of each time period can be carefully analyzed, which helps to discover abnormal changes in data in different time periods. By calculating the data volume, standard deviation, and weight value of abnormal data and garbled data, the data quality can be comprehensively and accurately evaluated. This multi-dimensional and multi-level data quality evaluation method improves the accuracy and reliability of the evaluation. The data quality evaluation coefficient is set and compared with the preset quality evaluation coefficient threshold. Once the data quality evaluation coefficient exceeds the threshold, the quality of the historical 5G network operation log can be determined to be abnormal, and an early warning can be issued. This early warning mechanism can detect problems in a timely manner and provide strong guarantees for subsequent data processing and analysis. The implementation of the early warning mechanism helps to discover and solve data quality problems in a timely manner, avoiding decision-making errors or business losses caused by data errors or missing data. Before analyzing the historical 5G network operation log, the data quality is first determined to ensure that the data analyzed subsequently is accurate and reliable. This helps to optimize the data processing process and improve the efficiency and accuracy of data analysis. Through data quality evaluation, problems in data collection, storage, transmission and other links can also be discovered, providing a basis for continuous improvement of data quality. Accurate and reliable data is the basis for business decision-making. By determining the data quality of historical 5G network operation logs, we can ensure that the data on which business decisions are based is of high quality, thereby improving the scientificity and accuracy of business decisions. High-quality data can also provide strong support for business optimization, troubleshooting, etc., and promote the continuous improvement and development of the business.

[0080] In summary, the technical effects of this technical solution in terms of performance indicators are mainly reflected in the accurate assessment of data quality, timely and effective abnormal warning, optimization of data processing flow, and improvement of business decision-making level. These effects work together in the entire data processing and analysis process, providing strong support for the monitoring and management of 5G network operations.

[0081] In another embodiment, the historical 5G network operation log of the cell where the 5G base station is located is analyzed to obtain the historical operation scenario type information of the 5G base station in the cell, including:

[0082] The historical 5G network operation logs of the cell where the 5G base station is located are analyzed to obtain type information corresponding to multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios during the historical operation of the 5G base station in the cell.

[0083] The beneficial effect of the above-mentioned embodiment is that the software energy-saving solution can flexibly adjust energy saving for different types of scenarios. In order to fully obtain the characteristic data of the cell where the 5G base station is located for different types of scenarios, it is necessary to pre-identify and determine the scene type information that may exist in the cell where the 5G base station is located. Specifically, the historical 5G network operation log of the cell where the 5G base station is located is analyzed to obtain the type information corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario of the 5G base station in the historical operation process of the cell. Among them, the multi-carrier deployment scenario is when the traffic volume of the carrier is lower than the preset threshold, the carrier can be shut down by the network management while ensuring the quality of user services; at this time, the AAU / RRU equipment can reduce the power consumption of the equipment to achieve the effect of equipment energy saving; when the network management configures the restart of the AAU / RRU equipment, the AAU / RRU equipment and the neighboring cells should be restored to the configuration before shutdown. The symbol shutdown scenario refers to the process of concentrating resources in specific time slots for scheduling when the network is under low load, shutting down the transmission power of time slots with no business, and reducing the total power consumption of the power amplifier module by discontinuous business transmission. When the symbol shutdown function is turned on, when there is no user data to be sent in the downlink symbol, the base station equipment can achieve energy saving by turning off the transmission power of the power amplifier module in the RF part. When the symbol shutdown function is turned on, it is necessary to transmit the reference signals of the necessary control channels, such as CSI-RS, SSS, PSS and PBCH in the 5G system, while maintaining the original transmission cycle and power without affecting the 5G network coverage. The power amplifier module can resume normal working state according to the increase in base station business volume to ensure that the 5G network performance is not affected. The channel shutdown scenario refers to the scenario where the 5G base station can shut down the transmission power of some RF channels in scenarios with low business load, such as idle time at night and non-capacity cell scenarios, to achieve energy saving effects; when the base station business load increases to a certain value, the base station turns on the closed RF channel and restores the multi-channel transmission state; enabling the channel shutdown solution will affect the system capacity, and the reduction in total transmission power may change the downlink antenna beam shape, affecting the original coverage and certain business performance. The cell shutdown scenario refers to shutting down low-traffic cells or cells with no traffic according to the traffic volume in the coverage area to achieve energy saving; for example, when the traffic load of the 5G cell in the capacity layer is low, the 5G cell is shut down and the 4G LTE cell providing the coverage layer is retained to reduce the energy consumption of 5G equipment. The deep sleep scenario refers to powering off most of the components of the AAU equipment (except the basic transmission link) when the base station traffic volume is very low. The deep sleep energy-saving function only retains the transmission link between the BBU and AAU, and supports the AAU device to enter / exit deep sleep mode, thereby achieving energy saving to the greatest extent; when AI is used for network energy saving, the accuracy of its prediction results, including network load, will directly affect the performance of the overall network.Accurate load prediction can timely determine the load trend of different nodes in the network, locate sites that can be deactivated, or implement optimal resource allocation, so that network performance is not affected and energy consumption is saved; however, inaccurate prediction will lead to incorrect site deactivation, causing a ping-pong effect in the site activation state, resulting in user call drops, affecting cell coverage, and worsening overall network performance.

[0084] In another embodiment, data recognition is performed on each scene sub-log to obtain feature data corresponding to the scene of the cell, including:

[0085] Data identification is performed on the scenario sub-logs corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario, and the topology information, uplink and downlink measurement information, service category information, user level information and resource occupancy distribution information corresponding to the cell in the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario are obtained as the feature data.

[0086] The beneficial effect of the above embodiment is that the software energy-saving solution needs to adjust different types of parameters in different scenarios. In order to ensure efficient and accurate software energy-saving operations in different scenarios, it is necessary to perform data identification on the scenario sub-logs corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario, and obtain the topology information, uplink and downlink measurement information, service category information, user level information and resource occupancy distribution information corresponding to the cell in the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario, respectively. This is used as the feature data, so that the model training is carried out for all scenarios based on the feature data in the future, so that the trained model can be compatible with the scenarios that the cell where the 5G base station is located may face.

[0087] In another embodiment, based on an AI-improved clustering algorithm, the feature data of all scenes are classified to obtain several data sets corresponding to different scenes, including:

[0088] Based on the improved K-means clustering algorithm, the feature data of all scenarios are initially classified and modified to obtain several data sets corresponding to multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios.

[0089] The beneficial effect of the above embodiment is that in order to accurately and effectively classify the feature data of all scenes, the K-means improved clustering algorithm can be used to process the feature data. Specifically, the following steps are included:

[0090] Step 1: Randomly select a center u among the data points of the feature datai ;

[0091] Step 2: For each data point x that has not yet been selected, calculate That is, the distance between x and the closest center that has been selected, where j is the number of centers;

[0092] Step 3: Use the weighted probability distribution to randomly select a new data point as the new center, where the probability of the selected data point x is the same as is proportional to;

[0093] Step 4, repeat the above steps 2 and 3 until K centers are selected, i.e. j = K;

[0094] Step 5: Based on the selected initial centers, continue to use standard K-means clustering.

[0095] Through the above process, the feature data of all scenarios are initially classified and modified to obtain several data sets corresponding to multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios, so as to ensure that each classified data set matches the corresponding scenario.

[0096] In another embodiment, a system performance evaluation index model is constructed, and based on the data set, the system performance evaluation index model is trained to obtain a feasibility evaluation result of performing energy-saving operation of 5G base station equipment on the cell in the corresponding scenario, including:

[0097] Based on the system component structure of 5G base stations, a system performance evaluation index model is constructed;

[0098] Based on the data set, the system performance evaluation index model is trained to obtain the feasibility evaluation results of the energy-saving operation of 5G base station equipment in the cell in multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios; wherein the feasibility evaluation results include component system performance evaluation results, component performance effectiveness evaluation results, networking effectiveness evaluation results, emergent effectiveness evaluation results and system mission task effectiveness evaluation results of the energy-saving operation of 5G base station equipment.

[0099] The beneficial effect of the above embodiment is that in order to comprehensively predict the feasibility and effectiveness of software energy-saving operations in all scenarios when the 5G base station faces different scenarios, a corresponding system performance evaluation index model can be constructed through a neural network model based on the system component structure of the 5G base station (such as the hardware and software component structure of the 5G base station, etc.), thereby providing an effective and reliable model basis for the 5G base station to perform software energy-saving operations in different scenarios. Specifically, based on the data set, the system performance evaluation index model is trained to obtain the feasibility evaluation results of the energy-saving operation of the 5G base station equipment in the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario of the cell, and obtain different evaluation indicators such as component system performance evaluation results, component performance effectiveness evaluation results, networking effectiveness evaluation results, emergent effectiveness evaluation results and system mission task effectiveness evaluation results to comprehensively evaluate the software energy-saving operation. Among them, the component system performance evaluation results (Measure of Component Performance, MOCP) and the component performance effectiveness evaluation results (Measure of Component Effectiveness, MOCE) are for any component system S i For the corresponding component performance evaluation results MOCE i ,

[0100]

[0101] In the above formula, MOCE is the component performance evaluation result i The component performance evaluation results corresponding to the j-th dimension;

[0102] further, In the above formula, f i,j () is the preset first function, is the component system performance evaluation result corresponding to the kth dimension of the component system performance evaluation result.

[0103] The results of the network effectiveness evaluation (Measure of SOS Net.work Effectiveness, MONE) include complex network modeling of the system and the construction of network effectiveness indicators based on complex system network modeling, namely

[0104] MONE=(G SOS , M)

[0105] In the above formula, G SOS is the system network model, M = {Metric1, Metric2, ..., Metric n} is a set of network performance indicators, and Metric i =f i (G SOS ), f i () is the preset second function.

[0106] The Measure of SOS Emergence Effectiveness (MOEE) measures the overall emergence of the system, with particular emphasis on the overall characteristics that emerge in terms of structure, function, and behavior during the system evolution process, such as the robustness and fragility of the architecture, the new capabilities generated by the functional coupling of component systems, and the adaptive and synchronous behavior of the system.

[0107] The system mission effectiveness evaluation results (Measure of SOS Task Effectiveness, MOTE), its indicators mainly include: shutdown effect, false alarm rate, shutdown success rate, shutdown success / false alarm ratio, shutdown delay, and shutdown time.

[0108] In another embodiment, based on the feasibility evaluation result, determining the execution state information of performing the energy-saving operation of the 5G base station device on the cell in the corresponding scenario includes:

[0109] Based on the feasibility assessment result, determine the hardware resource allocation execution status information of the 5G base station equipment software energy-saving operation for the cell in any one of the multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios.

[0110] The beneficial effects of the above embodiments are based on the feasibility assessment results to determine the hardware resource allocation execution status information of the 5G base station equipment software energy-saving operation for the cell in any of the multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios. In this way, with the measurement of the system's mission completion as the driving force, the overall emergent capability assessment of the system as the focus, and the "networked" structure evaluation of the system as the key, a networked system capability assessment framework is established, and reference indicators for system capability assessment are given.

[0111] In another embodiment, based on the execution status information, generating an energy-saving operation plan for the 5G base station includes:

[0112] Based on the hardware resource allocation execution status information, a software energy-saving operation plan for the 5G base station is generated; wherein the software energy-saving operation plan includes execution mode information and execution time information of the software energy-saving operation.

[0113] The beneficial effect of the above embodiment is that based on the hardware resource allocation execution status information, a software energy-saving operation plan for the 5G base station is generated, so that an energy-saving operation plan for the 5G base station can be generated, thereby improving the energy-saving operation efficiency and reliability without affecting the key performance indicators of the network providing services.

[0114] The method for energy-saving evaluation of 5G base stations based on AI intelligent control of the present application can be implemented through the following embodiments:

[0115] In 5G networks, the locations of AI model-related modules can be configured in a variety of ways. The deployment locations of AI model training modules, reasoning modules, etc. need to be determined based on specific use cases. According to the computational complexity and latency requirements of different use cases, the AI ​​training module and reasoning module of the AI-based base station energy-saving technology can be configured on the radio access network (RAN) side or the network management (OAM) side. A typical deployment method is: the AI ​​training module is deployed in OAM, and the AI ​​reasoning module is deployed in CU; the CU sends action instructions to its internal units, DU and neighboring stations based on the reasoning results; the CU, DU, terminal and neighboring stations feedback network performance to OAM, and OAM updates the AI ​​profile based on these feedbacks and sends it to the CU. The detailed process is as follows:

[0116] (1) OAM performs offline AI training based on historical network data to obtain an AI model for network energy saving, and sends the AI ​​model for network energy saving to CU;

[0117] (2) DU reports AI capabilities to CU, such as whether DU supports symbol switching and supported RF power range;

[0118] (3) Based on the DU's network energy-saving AI capabilities, the CU sends energy-saving collection configurations to the DU and UE to collect network data required for energy-saving AI reasoning;

[0119] (4) DU and UE report the data required for network energy-saving AI reasoning;

[0120] (5) Based on the strategies and AI models issued by OAM, CU uses the data reported by DU and UE to perform AI reasoning and output the reasoning results. The reasoning results can be cell shutdown, symbol shutdown, adjustment of one or some SSB burst cycles, SSB power adjustment, etc.

[0121] (6) The CU sends the corresponding actions of the reasoning results to the DU (which may include the neighboring DU), and the modification of the network configuration may also include the reconfiguration of system information, etc.

[0122] (7) After the network configuration is modified, the DU and UE may feedback new network performance to the CU, such as coverage holes and UE wireless link failures. If the AI ​​module in the CU has a reinforcement learning function, it will use this network performance feedback information to compare with expectations and iteratively optimize the AI ​​model. At the same time, OAM can also receive this type of network performance feedback information at a certain period and perform reinforcement learning on the AI ​​model to further improve the prediction accuracy of the AI ​​model.

[0123] (8) After CU updates the AI ​​model through reinforcement learning, it can provide feedback of the updated AI model to OAM.

[0124] In general, the method of 5G base station energy-saving evaluation based on AI intelligent control identifies the characteristic data of the cell where the 5G base station is located corresponding to different scenarios, and classifies the characteristic data of all scenarios based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenarios, providing a data basis for the subsequent training system performance evaluation index model; the training system performance evaluation index model obtains the feasibility evaluation results of the energy-saving operation of 5G base station equipment in the cell in the corresponding scenario, and accurately predicts the feasibility of performing different types of software energy-saving operations in the current scenario, providing a reference basis for the subsequent actual execution of software energy-saving operations; based on the feasibility evaluation results, the execution status information of the energy-saving operation of 5G base station equipment in the cell in the corresponding scenario is determined, so as to generate an energy-saving operation plan for the 5G base station, thereby improving the efficiency and reliability of energy-saving operations without affecting the key performance indicators of network service provision.

[0125] The above is only a specific implementation of the present invention, and any other improvements made based on the concept of the present invention are considered to be within the protection scope of the present invention.

Claims

1. A method for energy-saving evaluation of 5G base stations based on AI intelligent control, characterized in that: include: Perform scene recognition on the cell where the 5G base station is located to obtain feature data corresponding to different scenes of the cell; classify the feature data of all scenes based on the AI ​​improved clustering algorithm to obtain several data sets corresponding to different scenes; Constructing a system performance evaluation index model, and training the system performance evaluation index model based on the data set to obtain a feasibility evaluation result of performing energy-saving operations of 5G base station equipment on the cell in the corresponding scenario; Based on the feasibility assessment result, determine execution status information of performing energy-saving operation of a 5G base station device on the cell in a corresponding scenario; Based on the execution status information, an energy-saving operation plan for the 5G base station is generated.

2. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 1, characterized in that: Perform scene recognition on the cell where the 5G base station is located to obtain feature data corresponding to different scenes of the cell, including: Analyze the historical 5G network operation logs of the cell where the 5G base station is located to obtain historical operation scenario type information of the 5G base station in the cell; based on the historical original scenario type information, divide the historical 5G network operation logs into scenario operation time intervals to obtain scenario sub-logs matching each historical operation scenario type; Data recognition is performed on each scene sub-log to obtain feature data corresponding to the corresponding scene of the cell.

3. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 1, characterized in that: Before analyzing the historical 5G network operation logs of the cell where the 5G base station is located, the data quality of the historical 5G network operation logs is determined, including: Before analyzing the historical 5G network operation log of the cell where the 5G base station is located, the historical 5G network operation log is divided according to a preset unit time to obtain historical 5G network operation log data corresponding to multiple unit times, and the value of the unit time is 24 hours; the historical 5G network operation log data corresponding to each unit time is scanned to obtain the data volume corresponding to the abnormal data and garbled data in the historical 5G network operation log; Obtaining a first evaluation parameter using the data amount corresponding to the abnormal data and the garbled data; The first evaluation parameter is obtained by the following formula: Among them, F 01 represents the first evaluation parameter; n represents the total number corresponding to the unit time; C yi represents the amount of abnormal data corresponding to the i-th unit time; C mi represents the amount of garbled data corresponding to the i-th unit time; C yb Indicates the standard deviation of the amount of abnormal data corresponding to n units of time; C mb Indicates the standard deviation of the amount of garbled data corresponding to n unit time; C zi Represents the total amount of data corresponding to the i-th unit time; Extracting the weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log; Obtain a second evaluation parameter using a weight value corresponding to each abnormal data and garbled data in the historical 5G network operation log; The second evaluation parameter is obtained by the following formula: Among them, F 02 represents the second evaluation parameter; k represents the total number of abnormal data; g represents the total number of garbled data; w i represents the weight value of the i-th abnormal data; w b represents the standard deviation of the weight values ​​of n abnormal data; w max Indicates the maximum weight value of n abnormal data; s i Indicates the weight value of the i-th garbled data; s b Indicates the standard deviation of the weight values ​​of n garbled data; s max Indicates the maximum weight value of n garbled data; Obtaining a data quality evaluation coefficient using the first evaluation parameter and the second evaluation parameter; The data quality evaluation coefficient is obtained by the following formula: Where H represents the data quality evaluation coefficient; H0 represents the preset data quality evaluation coefficient threshold; F 01 Represents the first evaluation parameter; F 02 Represents the second evaluation parameter; F 01c Indicates the preset first evaluation parameter reference value; F 02c Indicates a preset second evaluation parameter reference value; Comparing the data quality evaluation coefficient with a preset quality evaluation coefficient threshold; When the data quality evaluation coefficient exceeds the preset quality evaluation coefficient threshold, the quality of the historical 5G network operation log is determined to be abnormal, and an abnormal quality warning of the historical 5G network operation log is issued.

4. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 2, characterized in that: Analyze the historical 5G network operation logs of the cell where the 5G base station is located to obtain the historical operation scenario type information of the 5G base station in the cell, including: The historical 5G network operation logs of the cell where the 5G base station is located are analyzed to obtain type information corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario during the historical operation of the 5G base station in the cell.

5. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 4, characterized in that: Data recognition is performed on each scene sub-log to obtain feature data corresponding to the scene of the cell, including: Data identification is performed on the scenario sub-logs corresponding to the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario, and the topology information, uplink and downlink measurement information, service category information, user level information and resource occupancy distribution information corresponding to the cell in the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario are obtained as the feature data.

6. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 5, characterized in that: Based on the AI-improved clustering algorithm, the feature data of all scenes are classified to obtain several data sets corresponding to different scenes, including: Based on the K-means improved clustering algorithm, the feature data of all scenarios are initially classified and modified to obtain several data sets corresponding to multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios.

7. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 1, characterized in that: Constructing a system performance evaluation index model, and training the system performance evaluation index model based on the data set to obtain a feasibility evaluation result of performing energy-saving operation of 5G base station equipment on the cell in the corresponding scenario, including: Based on the system component structure of 5G base stations, a system performance evaluation index model is constructed; Based on the data set, the system performance evaluation index model is trained to obtain feasibility evaluation results of performing energy-saving operations of 5G base station equipment on the cell in multi-carrier deployment scenarios, symbol shutdown scenarios, channel shutdown scenarios, cell shutdown scenarios and deep sleep scenarios; wherein the feasibility evaluation results include component system performance evaluation results of energy-saving operations of 5G base station equipment, component performance effectiveness evaluation results, networking effectiveness evaluation results, emergent effectiveness evaluation results and system mission task effectiveness evaluation results.

8. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 1, characterized in that: Based on the feasibility assessment result, determining execution status information of performing energy-saving operation of a 5G base station device on the cell in a corresponding scenario includes: Based on the feasibility assessment result, determine the hardware resource allocation execution status information of the 5G base station equipment software energy-saving operation for the cell in any one of the multi-carrier deployment scenario, symbol shutdown scenario, channel shutdown scenario, cell shutdown scenario and deep sleep scenario.

9. The method for energy-saving evaluation of 5G base stations based on AI intelligent control according to claim 8, characterized in that: Based on the execution status information, generating an energy-saving operation plan for the 5G base station includes: Based on the hardware resource allocation execution status information, a software energy-saving operation plan for the 5G base station is generated; wherein the software energy-saving operation plan includes execution mode information and execution time information of the software energy-saving operation.

Images