Base station energy consumption optimization system based on machine learning

Abstract

The invention relates to the technical field of base station energy consumption optimization, in particular to a base station energy consumption optimization system based on machine learning, which comprises a base station energy consumption optimization platform, a base station information base, a state division unit, a demand management unit, a prediction construction unit, a prediction optimization unit and a base station management unit, according to the invention, the time period state division analysis is carried out from the flow angle of the target base station to know the state condition of each time period of the target base station, and the energy consumption information of the target base station in the current time period is further analyzed in an information progressive manner. According to the method and the device, an energy consumption optimization scheme of a current target base station is adjusted according to information feedback conditions, so that low-consumption and high-efficiency operation of the target base station in different time periods is realized, and a next prediction time period is analyzed in an information feedback and model construction prediction mode; therefore, targeted energy consumption optimization processing of the target base station is carried out according to different optimization demand levels, and long-term reliable operation of the target base station is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of base station energy consumption optimization, and in particular to a base station energy consumption optimization system based on machine learning. Background Art

[0002] With the advent of the 5G era, the construction of 5G base stations as infrastructure has developed rapidly. A base station refers to a radio transceiver station that transmits information between mobile phone terminals through a mobile communication exchange center in a certain radio coverage area; the construction of mobile communication base stations is an important part of the investment of mobile communication operators. The construction of mobile communication base stations is generally carried out around factors such as coverage, call quality, and easy maintenance; However, in the existing technology, with the expansion of network scale and the increase of user equipment, higher network performance is inevitably accompanied by higher energy consumption, and the existing base station optimization technology is often unable to make reasonable and targeted energy consumption optimization adjustments according to the time period status of the base station, which is not conducive to achieving low-consumption and high-efficiency operation of the base station, and it is impossible to predict the next energy consumption of the base station in advance, and thus it is impossible to reasonably optimize and adjust in advance, and thus it is impossible to ensure the reliable operation of the base station and the efficient use of electric energy; In view of the above technical defects, a solution is now proposed. Summary of the invention

[0003] The purpose of the present invention is to provide a base station energy consumption optimization system based on machine learning to solve the technical defects mentioned above. The present invention divides the time period status into analysis from the traffic perspective of the target base station to understand the status of each time period of the target base station, so as to analyze the energy consumption according to the time period status, which is helpful to improve the efficiency and accuracy of energy consumption management, and further evaluates and analyzes the energy consumption optimization adjustment needs of the energy consumption information of the target base station in the current time period in an information progressive manner, so as to adjust the energy consumption optimization plan of the current target base station according to the information feedback, so as to improve the low-consumption and high-efficiency operation of the target base station in different time periods, and performs energy consumption prediction optimization analysis on the next predicted time period through information feedback and model construction prediction, so as to perform targeted target base station energy consumption optimization processing according to different optimization demand levels, to ensure the long-term and reliable operation of the target base station, save the energy consumption of the target base station and improve the utilization rate of electric energy.

[0004] The object of the present invention can be achieved by the following technical solutions: A base station energy consumption optimization system based on machine learning, including a base station energy consumption optimization platform, a base station information library, a state division unit, a demand management unit, a prediction construction unit, a prediction optimization unit and a base station management unit; The base station energy consumption optimization platform is used to retrieve the flow information of the target base station from the base station information database and send the flow information to the state division unit; The state division unit is used to perform base station time period state division analysis on the received traffic information, and send the obtained idle time period and high-frequency time period to the base station energy consumption optimization platform; The demand management unit is used to collect energy consumption information corresponding to idle time periods or high-frequency time periods, and at the same time, evaluate and analyze the energy consumption information for energy consumption optimization and adjustment needs, and obtain a stable signal or an idle optimization signal or a normal signal or a high-frequency optimization signal; The prediction construction unit is used to respond to stable signals and normal signals, collect historical energy consumption information of the target base station, and perform self-check analysis on the historical energy consumption information to construct an energy consumption prediction model to obtain a feedback signal or an idle calibration signal or a high-frequency calibration signal or a dual calibration signal; The prediction and optimization unit is used to perform energy consumption prediction and optimization analysis, and compare and analyze the obtained predicted energy consumption value with the preset predicted energy consumption value threshold. If an optimization signal is generated, the predicted energy consumption value is further analyzed to obtain an energy consumption optimization evaluation index, and the obtained energy consumption optimization evaluation index is discriminated to obtain the optimization demand level YD.

[0005] Preferably, the base station time period state division analysis process of the state division unit is as follows: The operating time period of the target base station is collected, and the operating time period of the target base station is set as the time threshold. The time threshold is divided into i sub-time periods, where i is a natural number greater than zero. The flow information of the target base station in each sub-time period is obtained, and the flow information is distinguished and processed to obtain idle time periods and high-frequency time periods.

[0006] Preferably, the energy consumption optimization adjustment of the demand management unit requires an evaluation and analysis process as follows: Obtain energy consumption information of the target base station during an idle period or a high-frequency period, the energy consumption information including the base station traffic volume and the energy consumption value, and then obtain the ratio between the base station traffic volume and the energy consumption value, and set the ratio between the base station traffic volume and the energy consumption value as the energy consumption risk ratio; The energy consumption risk ratio is compared and analyzed with the preset energy consumption risk ratio threshold. If the energy consumption risk ratio is less than the preset energy consumption risk ratio threshold, the duration corresponding to the energy consumption risk ratio being less than the preset energy consumption risk ratio threshold is obtained and set as the energy consumption regulation demand degree, and the energy consumption regulation demand degree is compared and analyzed with the preset energy consumption regulation demand degree threshold: If the energy consumption control demand is less than the preset energy consumption control demand threshold, the corresponding idle period or high-frequency period is judged to be normal; if the energy consumption control demand is greater than or equal to the preset energy consumption control demand threshold, the corresponding idle period or high-frequency period is judged to be abnormal.

[0007] Preferably, the number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are obtained, and the number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are set as idle demand values ​​and high-frequency demand values ​​respectively, and the idle demand values ​​and high-frequency demand values ​​are compared and analyzed to obtain a stable signal or an idle optimization signal or a normal signal or a high-frequency optimization signal.

[0008] Preferably, the energy consumption prediction model construction self-check analysis process of the prediction construction unit is as follows: Obtain historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, pre-process the historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, and construct an idle energy consumption prediction model and a high-frequency energy consumption prediction model through machine learning; Obtaining an idle predicted energy consumption value and a high-frequency predicted energy consumption value based on the idle energy consumption prediction model and the high-frequency energy consumption prediction model; The actual energy consumption values ​​corresponding to the idle period and the actual energy consumption values ​​corresponding to the high-frequency period of the target base station are obtained, and the predicted deviation value between the actual energy consumption value corresponding to the idle period and the idle predicted energy consumption value deviates from the preset range by the number of idle period corresponding to the number of idle period, and then the ratio between the deviation value and the total number of idle period is set as the idle prediction deviation rate, and the predicted deviation value between the actual energy consumption value corresponding to the high-frequency period and the high-frequency predicted energy consumption value deviates from the preset range by the number of high-frequency period corresponding to the number of high-frequency period, and then the ratio between the deviation value and the total number of high-frequency period is set as the high-frequency prediction deviation rate.

[0009] Preferably, the idle prediction deviation rate and the high-frequency prediction deviation rate are compared and analyzed with a preset idle prediction deviation rate threshold and a preset high-frequency prediction deviation rate threshold to obtain a feedback signal or an idle calibration signal or a high-frequency calibration signal or a dual calibration signal.

[0010] Preferably, the energy consumption prediction optimization analysis process of the prediction optimization unit is as follows: The time period corresponding to the idle period and the time period corresponding to the high-frequency period are obtained, the next predicted period is obtained, and the matching results of the predicted period with the idle period and the high-frequency period are obtained. The predicted energy consumption value of the predicted period is obtained based on the matching results, and the predicted energy consumption value is compared and analyzed with the preset predicted energy consumption value threshold to generate an optimization signal.

[0011] Preferably, after the optimization signal is generated, the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is obtained, and the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is set as the energy consumption optimization evaluation index, and the energy consumption optimization evaluation index is discriminated and processed to obtain the first-level demand, the second-level demand and the third-level demand, and the first-level demand, the second-level demand and the third-level demand are set as the optimized demand level YD, YD=1, 2, 3.

[0012] The beneficial effects of the present invention are as follows: (1) The present invention analyzes the state of each time period of the target base station by dividing the time period from the perspective of the traffic of the target base station, so as to understand the state of each time period of the target base station, so as to analyze the energy consumption according to the time period state, which is helpful to improve the efficiency and accuracy of energy consumption management, and further evaluates and analyzes the energy consumption optimization adjustment needs of the energy consumption information of the target base station in the current time period in an information progressive manner, so as to adjust the energy consumption optimization plan of the current target base station according to the information feedback, so as to improve the low-consumption and high-efficiency operation of the target base station in different time periods; (2) The present invention performs energy consumption forecast optimization analysis for the next forecast period by means of information feedback and model construction prediction, so as to carry out targeted energy consumption optimization processing of the target base station according to different optimization demand levels, thereby helping to reasonably reduce the operating energy consumption of the target base station, and at the same time helping to improve the operating efficiency of the target base station, ensuring the long-term and reliable operation of the target base station, saving the energy consumption of the target base station and improving the utilization rate of electric energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The present invention will be further described below in conjunction with the accompanying drawings; Figure 1 It is a flowchart of the system of the present invention; Figure 2 It is a local analysis reference diagram of the present invention. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] Embodiment 1: See also Figure 1 to Figure 2 As shown, the present invention is a base station energy consumption optimization system based on machine learning, including a base station energy consumption optimization platform, a base station information database, a state division unit, a demand management unit, a prediction construction unit, a prediction optimization unit and a base station management unit, the base station information database is connected to the base station energy consumption optimization platform in a one-way communication, the base station energy consumption optimization platform is connected to the state division unit in a two-way communication, the base station energy consumption optimization platform is connected to the demand management unit in a one-way communication, the demand management unit is connected to the prediction construction unit, the prediction optimization unit and the base station management unit in a one-way communication, the prediction construction unit is connected to the base station management unit and the prediction optimization unit in a one-way communication, and the prediction optimization unit is connected to the base station management unit in a one-way communication; The base station energy consumption optimization platform is used to retrieve the flow information of the target base station from the base station information database, and send the flow information to the state division unit. The state division unit is used to perform base station time period state division analysis on the received flow information to understand the state of each time period of the target base station, so as to analyze the energy consumption according to the state. The specific base station time period state division analysis process is as follows: The operating time period of the target base station is collected, and the operating time period of the target base station is set as the time threshold. The time threshold is divided into i sub-time periods, where i is a natural number greater than zero. The flow information of the target base station in each sub-time period is obtained, and the flow information is discriminated and processed: If the flow information is less than the preset flow information, the corresponding sub-time period is determined to be an idle period; If the flow information is greater than or equal to the preset flow information, the corresponding sub-time period is determined to be a high-frequency time period, and the idle time period and the high-frequency time period are sent to the base station energy consumption optimization platform; In the embodiment of the present invention, the flow information indicates the data usage; When the target base station is in an idle period or a high-frequency period, the demand management unit is used to collect energy consumption information corresponding to the idle period or the high-frequency period, and at the same time, the energy consumption optimization adjustment needs evaluation and analysis of the energy consumption information, so as to adjust the energy consumption optimization plan of the current target base station according to the information feedback, so as to improve the low-consumption and high-efficiency operation of the target base station in different time periods. The specific energy consumption optimization adjustment needs evaluation and analysis process is as follows: The energy consumption information of the target base station during the idle period or high-frequency period is obtained, and the energy consumption information includes the base station business volume and the energy consumption value, and then the ratio between the base station business volume and the energy consumption value is obtained, and the ratio between the base station business volume and the energy consumption value is set as the energy consumption risk ratio. It should be noted that the larger the value of the energy consumption risk ratio, the smaller the risk of low-consumption and high-efficiency operation of the target base station; The energy consumption risk ratio is compared and analyzed with the preset energy consumption risk ratio threshold. If the energy consumption risk ratio is less than the preset energy consumption risk ratio threshold, the duration corresponding to the energy consumption risk ratio being less than the preset energy consumption risk ratio threshold is obtained and set as the energy consumption regulation demand degree, and the energy consumption regulation demand degree is compared and analyzed with the preset energy consumption regulation demand degree threshold: If the energy consumption control demand is less than the preset energy consumption control demand threshold, it is determined that the corresponding idle period or high-frequency period is normal; If the energy consumption control demand is greater than or equal to the preset energy consumption control demand threshold, it is determined that the corresponding idle period or high-frequency period is abnormal; The number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are obtained, and the number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are set as idle demand values ​​and high-frequency demand values, respectively, and the idle demand values ​​and high-frequency demand values ​​are compared and analyzed: If the idle demand value is less than a preset idle demand value threshold, a stable signal is generated; If the idle demand value is greater than or equal to a preset idle demand value threshold, an idle optimization signal is generated; If the high frequency demand value is less than the preset high frequency demand value threshold, a normal signal is generated; If the high-frequency demand value is greater than or equal to the preset high-frequency demand value threshold, a high-frequency optimization signal is generated, and a stable signal, an idle optimization signal, a normal signal, and a high-frequency optimization signal are sent to the base station management unit. After receiving the stable signal, the idle optimization signal, the normal signal, and the high-frequency optimization signal, the base station management unit immediately performs preset warning operations corresponding to the stable signal, the idle optimization signal, the normal signal, and the high-frequency optimization signal, so as to carry out energy consumption warning for the target base station according to the information feedback, so as to adjust the energy consumption optimization plan of the target base station in different time periods, so as to improve the low-consumption and high-efficiency operation of the target base station in different time periods, and at the same time help to reasonably and targetedly manage the energy consumption of the target base station and adjust the energy consumption optimization plan in a timely manner; In an embodiment of the present invention, the preset energy consumption risk ratio threshold and the preset energy consumption control demand threshold correspond to the target base station being in an idle period or a high-frequency period. That is, when the target base station is in a high-frequency period, the preset energy consumption risk ratio threshold and the preset energy consumption control demand threshold are the set values ​​corresponding to the high-frequency period.

[0016] Embodiment 2: When stable signals and normal signals are generated, the prediction construction unit is used to respond to the stable signals and normal signals, collect historical energy consumption information of the target base station, and perform self-check analysis on the historical energy consumption information to build an energy consumption prediction model, which helps to predict the next energy consumption situation through model prediction so as to make reasonable optimization and adjustment. The specific energy consumption prediction model construction self-check analysis process is as follows: Obtain historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, pre-process the historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, and construct an idle energy consumption prediction model and a high-frequency energy consumption prediction model through machine learning; Obtaining an idle predicted energy consumption value and a high-frequency predicted energy consumption value based on the idle energy consumption prediction model and the high-frequency energy consumption prediction model; In the embodiment of the present invention, the preprocessing includes data cleaning, data screening, etc.; the historical energy consumption information indicates the historical energy consumption value corresponding to the idle period; The actual energy consumption value corresponding to the idle period and the actual energy consumption value corresponding to the high-frequency period of the target base station are obtained, and the predicted deviation value between the actual energy consumption value corresponding to the idle period and the idle predicted energy consumption value deviates from the preset range by the number of idle periods corresponding to the number of idle periods, and then the ratio between the deviation value and the total number of idle periods is set as the idle predicted deviation rate, and the predicted deviation value between the actual energy consumption value corresponding to the high-frequency period and the high-frequency predicted energy consumption value deviates from the preset range by the number of high-frequency periods corresponding to the number of high-frequency periods, and then the ratio between the deviation value and the total number of high-frequency periods is set as the high-frequency predicted deviation rate; Compare and analyze the idle prediction deviation rate and the high-frequency prediction deviation rate with the preset idle prediction deviation rate threshold and the preset high-frequency prediction deviation rate threshold: If the idle prediction deviation rate is less than a preset idle prediction deviation rate threshold, and the high frequency prediction deviation rate is less than a preset high frequency prediction deviation rate threshold, a feedback signal is generated; If the idle prediction deviation rate is greater than or equal to a preset idle prediction deviation rate threshold, and the high frequency prediction deviation rate is less than a preset high frequency prediction deviation rate threshold, an idle calibration signal is generated; If the idle prediction deviation rate is less than a preset idle prediction deviation rate threshold, and the high frequency prediction deviation rate is greater than or equal to a preset high frequency prediction deviation rate threshold, a high frequency calibration signal is generated; If the idle prediction deviation rate is greater than or equal to the preset idle prediction deviation rate threshold, and the high-frequency prediction deviation rate is greater than or equal to the preset high-frequency prediction deviation rate threshold, a dual calibration signal is generated, and the feedback signal, the idle calibration signal, the high-frequency calibration signal, and the dual calibration signal are sent to the base station management unit. After receiving the feedback signal, the idle calibration signal, the high-frequency calibration signal, and the dual calibration signal, the base station management unit immediately performs preset warning operations corresponding to the feedback signal, the idle calibration signal, the high-frequency calibration signal, and the dual calibration signal, so as to optimize and calibrate the idle energy consumption prediction model and the high-frequency energy consumption prediction model, so as to improve the prediction accuracy of the idle energy consumption prediction model and the high-frequency energy consumption prediction model; When generating stable signals, normal signals and feedback signals, the prediction optimization unit is used to perform energy consumption prediction optimization analysis, so as to optimize and adjust the target base station in advance by means of energy consumption prediction, so as to perform targeted energy consumption optimization processing of the target base station according to different optimization demand levels, thereby helping to reasonably reduce the operating energy consumption of the target base station, and at the same time helping to improve the operating efficiency of the target base station. The specific energy consumption prediction optimization analysis process is as follows: Get the time period corresponding to the idle period and the time period corresponding to the high-frequency period, get the next predicted period, and get the matching result between the predicted period and the idle period and the high-frequency period. Get the predicted energy consumption value of the predicted period based on the matching result, and compare and analyze the predicted energy consumption value with the preset predicted energy consumption value threshold: If the predicted energy consumption value is less than or equal to the preset predicted energy consumption value threshold, no signal is generated; If the predicted energy consumption value is greater than a preset predicted energy consumption value threshold, an optimization signal is generated; After the optimization signal is generated, the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is obtained, and the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is set as the energy consumption optimization evaluation index, and the energy consumption optimization evaluation index is judged: If the energy consumption optimization evaluation index is greater than the maximum value in the preset energy consumption optimization evaluation index range, it is determined to be a first-level demand; If the energy consumption optimization evaluation index falls within the preset energy consumption optimization evaluation index range, it is determined to be a secondary demand; If the energy consumption optimization evaluation index is less than the minimum value in the preset energy consumption optimization evaluation index range, it is determined to be a third-level demand, wherein the optimization degrees corresponding to the first-level demand, the second-level demand and the third-level demand are successively reduced, and the first-level demand, the second-level demand and the third-level demand are set as the optimization demand level YD, YD=1, 2, 3, that is, the optimization demand level YD=1 indicates the first-level demand, the optimization demand level YD=2 indicates the second-level demand, and the optimization demand level YD=3 indicates the third-level demand. The current optimization demand level YD is sent to the base station management unit. After receiving the current optimization demand level YD, the base station management unit immediately makes a preset warning operation corresponding to the current optimization demand level YD, so as to perform targeted target base station energy consumption optimization processing according to different optimization demand levels, thereby helping to reasonably reduce the operating energy consumption of the target base station, and at the same time helping to improve the operating efficiency of the target base station, ensuring the long-term and reliable operation of the target base station, saving the energy consumption of the target base station and improving the utilization rate of electric energy; In the embodiment of the present invention, the matching result of the prediction period with the idle period and the high-frequency period is: if the prediction period is in the time period corresponding to the idle period, the idle energy consumption prediction model is matched; If the predicted period is in the time period corresponding to the high-frequency period, the high-frequency energy consumption prediction model is matched; If the prediction period is both in the time period corresponding to the idle period and in the time period corresponding to the high-frequency period, the prediction period is divided into the idle period and the high-frequency period, and the corresponding idle period and high-frequency period are respectively substituted into the idle energy consumption prediction model and the high-frequency energy consumption prediction model for analysis, thereby obtaining the predicted energy consumption value of the prediction period; To summarize, the present invention divides the time period status into analysis from the traffic perspective of the target base station to understand the status of each time period of the target base station, so as to analyze the energy consumption according to the time period status, which is helpful to improve the efficiency and accuracy of energy consumption management, and further evaluates and analyzes the energy consumption optimization and adjustment needs of the energy consumption information of the target base station in the current time period in an information progressive manner, so as to adjust the energy consumption optimization plan of the current target base station according to the information feedback, so as to improve the low-consumption and high-efficiency operation of the target base station in different time periods, and performs energy consumption prediction optimization analysis on the next predicted time period through information feedback and model construction prediction, so as to perform targeted target base station energy consumption optimization processing according to different optimization demand levels, thereby helping to reasonably reduce the operating energy consumption of the target base station, and at the same time help to improve the operating efficiency of the target base station, ensure the long-term and reliable operation of the target base station, save the energy consumption of the target base station and improve the utilization rate of electric energy.

[0017] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0018] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula that is close to the actual value. The coefficients in the formula are set by technical personnel in this field according to actual conditions. The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with the technical field within the technical scope disclosed by the present invention, according to the technical solution and the inventive concept of the present invention, make equivalent replacement or change, which should be covered within the protection scope of the present invention.

Claims

1. A base station energy consumption optimization system based on machine learning, characterized in that: It includes a base station energy consumption optimization platform, a base station information database, a state division unit, a demand management unit, a prediction construction unit, a prediction optimization unit and a base station management unit; The base station energy consumption optimization platform is used to retrieve the flow information of the target base station from the base station information database and send the flow information to the state division unit; The state division unit is used to perform base station time period state division analysis on the received traffic information, and send the obtained idle time period and high-frequency time period to the base station energy consumption optimization platform; The demand management unit is used to collect energy consumption information corresponding to idle time periods or high-frequency time periods, and at the same time, evaluate and analyze the energy consumption information for energy consumption optimization and adjustment needs, and obtain a stable signal or an idle optimization signal or a normal signal or a high-frequency optimization signal; The prediction construction unit is used to respond to stable signals and normal signals, collect historical energy consumption information of the target base station, and perform self-check analysis on the historical energy consumption information to construct an energy consumption prediction model to obtain a feedback signal or an idle calibration signal or a high-frequency calibration signal or a dual calibration signal; The prediction and optimization unit is used to perform energy consumption prediction and optimization analysis, and compare and analyze the obtained predicted energy consumption value with the preset predicted energy consumption value threshold. If an optimization signal is generated, the predicted energy consumption value is further analyzed to obtain an energy consumption optimization evaluation index, and the obtained energy consumption optimization evaluation index is discriminated to obtain the optimization demand level YD.

2. The base station energy consumption optimization system based on machine learning according to claim 1, characterized in that: The base station time period state division analysis process of the state division unit is as follows: The operating time period of the target base station is collected, and the operating time period of the target base station is set as the time threshold. The time threshold is divided into i sub-time periods, where i is a natural number greater than zero. The flow information of the target base station in each sub-time period is obtained, and the flow information is distinguished and processed to obtain idle time periods and high-frequency time periods.

3. The base station energy consumption optimization system based on machine learning according to claim 2, characterized in that: The energy consumption optimization adjustment of the demand management unit requires evaluation and analysis process as follows: Obtain energy consumption information of the target base station during an idle period or a high-frequency period, the energy consumption information including the base station traffic volume and the energy consumption value, and then obtain the ratio between the base station traffic volume and the energy consumption value, and set the ratio between the base station traffic volume and the energy consumption value as the energy consumption risk ratio; The energy consumption risk ratio is compared and analyzed with the preset energy consumption risk ratio threshold. If the energy consumption risk ratio is less than the preset energy consumption risk ratio threshold, the duration corresponding to the energy consumption risk ratio being less than the preset energy consumption risk ratio threshold is obtained and set as the energy consumption regulation demand degree, and the energy consumption regulation demand degree is compared and analyzed with the preset energy consumption regulation demand degree threshold: If the energy consumption control demand is less than the preset energy consumption control demand threshold, it is determined that the corresponding idle period or high-frequency period is normal; If the energy consumption regulation demand is greater than or equal to the preset energy consumption regulation demand threshold, it is determined that the corresponding idle period or high-frequency period is abnormal.

4. The base station energy consumption optimization system based on machine learning according to claim 3 is characterized in that: The number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are obtained, and the number of abnormal idle time periods or high-frequency time periods and the number of normal idle time periods or high-frequency time periods are set as idle demand values ​​and high-frequency demand values ​​respectively, and the idle demand values ​​and the high-frequency demand values ​​are compared and analyzed to obtain a stable signal or an idle optimization signal or a normal signal or a high-frequency optimization signal.

5. The base station energy consumption optimization system based on machine learning according to claim 4, characterized in that: The energy consumption prediction model construction self-check analysis process of the prediction construction unit is as follows: Obtain historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, pre-process the historical energy consumption information corresponding to idle periods and high-frequency periods of the target base station, and construct an idle energy consumption prediction model and a high-frequency energy consumption prediction model through machine learning; Obtaining an idle predicted energy consumption value and a high-frequency predicted energy consumption value based on an idle energy consumption prediction model and a high-frequency energy consumption prediction model; The actual energy consumption values ​​corresponding to the idle period and the actual energy consumption values ​​corresponding to the high-frequency period of the target base station are obtained, and the predicted deviation value between the actual energy consumption value corresponding to the idle period and the idle predicted energy consumption value deviates from the preset range by the number of idle period corresponding to the number of idle period, and then the ratio between the deviation value and the total number of idle period is set as the idle prediction deviation rate, and the predicted deviation value between the actual energy consumption value corresponding to the high-frequency period and the high-frequency predicted energy consumption value deviates from the preset range by the number of high-frequency period corresponding to the number of high-frequency period, and then the ratio between the deviation value and the total number of high-frequency period is set as the high-frequency prediction deviation rate.

6. The base station energy consumption optimization system based on machine learning according to claim 5, characterized in that: The idle prediction deviation rate and the high-frequency prediction deviation rate are compared and analyzed with the preset idle prediction deviation rate threshold and the preset high-frequency prediction deviation rate threshold to obtain a feedback signal or an idle calibration signal or a high-frequency calibration signal or a dual calibration signal.

7. The base station energy consumption optimization system based on machine learning according to claim 6, characterized in that: The energy consumption prediction optimization analysis process of the prediction optimization unit is as follows: The time period corresponding to the idle period and the time period corresponding to the high-frequency period are obtained, the next predicted period is obtained, and the matching results of the predicted period with the idle period and the high-frequency period are obtained. The predicted energy consumption value of the predicted period is obtained based on the matching results, and the predicted energy consumption value is compared and analyzed with the preset predicted energy consumption value threshold to generate an optimization signal.

8. The base station energy consumption optimization system based on machine learning according to claim 7, characterized in that: After the optimization signal is generated, the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is obtained, and the part of the predicted energy consumption value that is greater than the preset predicted energy consumption value threshold is set as the energy consumption optimization evaluation index, and the energy consumption optimization evaluation index is discriminated and processed to obtain the first-level demand, the second-level demand and the third-level demand, and the first-level demand, the second-level demand and the third-level demand are set as the optimized demand level YD, YD=1, 2, 3.

Images