A cell energy saving control method, device, equipment and readable storage medium

Abstract

The present invention discloses a cell energy-saving control method, device, equipment and readable storage medium, which relates to the field of communication technology, so as to be able to more accurately configure energy-saving periods to tap energy-saving potential and reduce the impact on the network. The method includes: obtaining historical data; using historical data to obtain the changes in business data of the cell business data in the time period to be predicted; determining the stable time period and sensitive time period of the time period to be predicted; according to the stable time period and the sensitive time period, obtaining the first energy-saving strategy for the time period to be predicted; predicting the stable time period to obtain the second energy-saving strategy for the stable time period, and predicting the sensitive time period to obtain the third energy-saving strategy for the sensitive time period; based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable time period and the sensitive time period. The embodiment of the present invention can more accurately configure the energy-saving period to tap energy-saving potential and reduce the impact on the network.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a cell energy-saving control method, device, equipment and readable storage medium. Background Art

[0002] Currently, the generally recognized energy-saving technologies for wireless networks are: cell dormancy, carrier dormancy, channel dormancy and symbol dormancy. The principle is to shut down some wireless resources when the network load is relatively light, and use the remaining resources to provide services to achieve the purpose of energy saving.

[0003] There are two energy-saving strategies in the prior art: one is to control energy saving based on a fixed time period strategy, by setting the start and end time of energy saving, and start energy saving within the time period. The other is an energy-saving method based on cell traffic prediction, by setting energy-saving thresholds, to control the working state of the cell.

[0004] However, the first method requires manual collection of network data for different network scenarios and analysis of cell status and services, and then setting of sleep time periods, which has high labor costs. The second method makes judgments based on the real-time services of the cell, which will cause frequent activation of power saving (or wake-up) operations, which is likely to have a greater impact on the network and cause poor user perception. Summary of the invention

[0005] The embodiments of the present invention provide a method, device, equipment and readable storage medium for determining a cell energy-saving strategy, so as to more accurately configure energy-saving periods, tap energy-saving potentials and reduce the impact on the network.

[0006] In a first aspect, an embodiment of the present invention provides a cell energy saving control method, including:

[0007] Get historical data;

[0008] Using the historical data to predict the cell service data in the prediction period, and obtaining the service data change of the cell service data in the prediction period;

[0009] Determining a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell service data within the time period to be predicted;

[0010] According to the stable period and the sensitive period, a first energy-saving strategy for the period to be predicted is obtained;

[0011] Predicting the stable period to obtain a second energy-saving strategy for the stable period, and predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period;

[0012] Based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable period and the sensitive period.

[0013] The method of using the historical data to predict the cell service data in the prediction period to obtain the service data change of the cell service data in the prediction period includes:

[0014] When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted;

[0015] When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted;

[0016] Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

[0017] The step of determining the stable period and the sensitive period of the time period to be predicted according to the change of the service data of the cell service data within the time period to be predicted includes:

[0018] According to the change of the business data, comparing the business data of each sampling point in the time period to be predicted with a preset threshold, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval;

[0019] If the service data of the target sampling point is equal to the preset threshold, the period between the previous sampling point and the next sampling point of the target sampling point is used as the sensitive period, and the target sampling point is any sampling point among the multiple sampling points;

[0020] The other time periods in the time period to be predicted except the sensitive time period are used as the stable time periods.

[0021] Wherein, obtaining the first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period includes:

[0022] During the stable period, determining the first energy saving strategy includes putting the cell into sleep or wake-up;

[0023] During the sensitive period, determining the first energy saving strategy includes waking up the cell.

[0024] The predicting of the stable period to obtain a second energy-saving strategy for the stable period includes:

[0025] Using the first sub-model in the second prediction model corresponding to the stable period, predicting the business change in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval;

[0026] Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups;

[0027] If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group;

[0028] If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group;

[0029] If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

[0030] The step of predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period includes:

[0031] Using a second sub-model in a second prediction model corresponding to the sensitive period, predicting a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval;

[0032] Dividing the sampling points in the second time period into a plurality of groups;

[0033] If, in the second target group, the ratio of the number of sampling points whose traffic volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

[0034] The energy-saving control of the stable period and the sensitive period is performed based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy, respectively, including:

[0035] In the stable period, based on the first energy-saving strategy and the second energy-saving strategy, dynamically adjusting the wake-up and sleep of the cell according to the prediction result of the stable period;

[0036] During the sensitive period, based on the first energy-saving strategy and the third energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction result of the energy-saving period.

[0037] Wherein, the method further comprises:

[0038] Training a first sub-model and a second sub-model of the first prediction model;

[0039] Wherein, training the first sub-model includes:

[0040] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0041] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0042] Wherein, training the second sub-model includes:

[0043] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0044] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0045] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0046] Wherein, the method further comprises:

[0047] Training the first sub-model and the second sub-model of the second prediction model;

[0048] Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively;

[0049] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0050] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0051] In a second aspect, an embodiment of the present invention further provides a cell energy-saving control device, including:

[0052] A first acquisition module is used to acquire historical data;

[0053] A second acquisition module is used to predict the cell service data of the to-be-predicted period using the historical data, and obtain the service data change of the cell service data in the to-be-predicted period;

[0054] A first determination module, configured to determine a stable period and a sensitive period of the time period to be predicted according to a change in service data of the cell service data within the time period to be predicted;

[0055] A second determination module is used to obtain a first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period;

[0056] A third determination module is used to predict the stable period to obtain a second energy-saving strategy for the stable period, and to predict the sensitive period to obtain a third energy-saving strategy for the sensitive period;

[0057] The first processing module is used to perform energy-saving control on the stable period and the sensitive period based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively.

[0058] Wherein, the second acquisition module is used for:

[0059] When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted;

[0060] When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted;

[0061] Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

[0062] Wherein, the first determining module includes:

[0063] A first comparison submodule, configured to compare the business data of each sampling point in the time period to be predicted with a preset threshold value according to the change of the business data, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval;

[0064] A first determination submodule is configured to, if the service data of a target sampling point is equal to the preset threshold, use the period between the previous sampling point and the next sampling point of the target sampling point as the sensitive period, wherein the target sampling point is any sampling point among the multiple sampling points;

[0065] The second determining submodule is used to use other time periods in the time period to be predicted except the sensitive time period as the stable time period.

[0066] The second determination module is used to, during the stable period, determine that the first energy-saving strategy includes putting the cell into sleep or wake-up; during the sensitive period, determine that the first energy-saving strategy includes waking up the cell.

[0067] Wherein, the third determination module includes:

[0068] A first prediction submodule, configured to predict business changes in a first time period after a current moment using a first submodel in a second prediction model corresponding to the stable time period, wherein the first time period includes a plurality of sampling points set at a second preset interval;

[0069] A first processing submodule is used to form two adjacent sampling points among the multiple sampling points into a group starting from the first sampling point, to obtain multiple groups;

[0070] A first determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, then the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group;

[0071] A second determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, then the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group;

[0072] The third determination submodule is used to determine that if, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy-saving strategy includes waking up the cell.

[0073] Wherein, the third determination module includes:

[0074] A second prediction submodule, configured to use a second submodel in a second prediction model corresponding to the sensitive time period to predict a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval;

[0075] A second processing submodule, used for dividing the sampling points in the second time period into a plurality of groups;

[0076] The fourth determination submodule is used to determine that if the ratio of the number of sampling points in the second target group whose business volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, then the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

[0077] Wherein, the first processing module includes:

[0078] A first processing submodule is used to dynamically adjust the wake-up and sleep of the cell during the stable period based on the first energy-saving strategy and the second energy-saving strategy according to the prediction result of the stable period;

[0079] The second processing submodule is used to dynamically adjust the wake-up and sleep of the cell during the sensitive period based on the first energy-saving strategy and the third energy-saving strategy according to the prediction result of the energy-saving period.

[0080] The device further comprises: a first training module, configured to train a first sub-model and a second sub-model of the first prediction model;

[0081] Wherein, when training the first sub-model, the first training module is used to:

[0082] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0083] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0084] Wherein, when training the second sub-model, the first training module is used to:

[0085] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0086] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0087] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0088] The device further comprises: a second training module, configured to train the first sub-model and the second sub-model of the second prediction model;

[0089] Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively;

[0090] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0091] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0092] In a third aspect, an embodiment of the present invention further provides a cell energy saving control device, including: a processor and a transceiver;

[0093] The processor is configured to:

[0094] Get historical data;

[0095] Using the historical data to predict the cell service data in the prediction period, and obtaining the service data change of the cell service data in the prediction period;

[0096] Determining a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell service data within the time period to be predicted;

[0097] According to the stable period and the sensitive period, a first energy-saving strategy for the period to be predicted is obtained;

[0098] Predicting the stable period to obtain a second energy-saving strategy for the stable period, and predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period;

[0099] Based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable period and the sensitive period.

[0100] Wherein, the processor is used for:

[0101] When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted;

[0102] When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted;

[0103] Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

[0104] Wherein, the processor is used for:

[0105] According to the change of the business data, comparing the business data of each sampling point in the time period to be predicted with a preset threshold, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval;

[0106] If the service data of the target sampling point is equal to the preset threshold, the period between the previous sampling point and the next sampling point of the target sampling point is used as the sensitive period, and the target sampling point is any sampling point among the multiple sampling points;

[0107] The other time periods in the time period to be predicted except the sensitive time period are used as the stable time periods.

[0108] Wherein, the processor is used for:

[0109] During the stable period, determining the first energy saving strategy includes putting the cell into sleep or wake-up;

[0110] During the sensitive period, determining the first energy saving strategy includes waking up the cell.

[0111] Wherein, the processor is used for:

[0112] Using the first sub-model in the second prediction model corresponding to the stable period, predicting the business change in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval;

[0113] Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups;

[0114] If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group;

[0115] If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group;

[0116] If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

[0117] Wherein, the processor is used for:

[0118] Using a second sub-model in a second prediction model corresponding to the sensitive period, predicting a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval;

[0119] Dividing the sampling points in the second time period into a plurality of groups;

[0120] If, in the second target group, the ratio of the number of sampling points whose traffic volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

[0121] Wherein, the processor is used for:

[0122] In the stable period, based on the first energy-saving strategy and the second energy-saving strategy, dynamically adjusting the wake-up and sleep of the cell according to the prediction result of the stable period;

[0123] During the sensitive period, based on the first energy-saving strategy and the third energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction result of the energy-saving period.

[0124] Wherein, the processor is used for:

[0125] Training a first sub-model and a second sub-model of the first prediction model;

[0126] Wherein, when training the first sub-model, the processor is used to:

[0127] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0128] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0129] Wherein, when training the second sub-model, the processor is used to:

[0130] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0131] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0132] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0133] Wherein, the processor is used for:

[0134] Training the first sub-model and the second sub-model of the second prediction model;

[0135] Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively;

[0136] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0137] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0138] In a fourth aspect, an embodiment of the present invention further provides a communication device, comprising: a transceiver, a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the cell energy saving control method as described above are implemented.

[0139] In a fifth aspect, an embodiment of the present invention further provides a readable storage medium, on which a program is stored, and when the program is executed by a processor, the steps in the cell energy saving control method as described above are implemented.

[0140] In an embodiment of the present invention, the changes in business data within the time period to be predicted are predicted based on historical data, so that the stable time period and sensitive time period of the time period to be predicted are determined based on the business changes, and then the first energy-saving strategy for the time period to be predicted is determined; then, the stable time period and the sensitive time period are predicted respectively to obtain the corresponding second energy-saving strategy and third energy-saving strategy. Therefore, by using the scheme of the embodiment of the present invention, not only the long-term business change rules of the day to be measured can be obtained, but also the business changes in a short period of time can be predicted, so that by using the scheme of the embodiment of the present invention, not only the long-term energy-saving strategy can be obtained, but also different energy-saving strategies can be set for short time periods, and no manual intervention is required. Therefore, by using the scheme of the embodiment of the present invention, the energy-saving period can be configured more accurately according to the changes in the cell business, the energy-saving potential can be tapped, and the impact on the network can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0141] Figure 1 is a flow chart of a cell energy saving control method provided by an embodiment of the present invention;

[0142] Figure 2 It is a schematic diagram of cell service history data in an embodiment of the present invention;

[0143] Figure 3 is a schematic diagram of the relationship between the service change situation and the threshold value in the time period to be predicted according to an embodiment of the present invention;

[0144] Figure 4 This is a schematic diagram of the processing flow of the LSTM network model;

[0145] Figure 5 This is the structure diagram of LSTM;

[0146] Figure 6 It is one of the structural diagrams of the cell energy-saving control device provided in an embodiment of the present invention;

[0147] Figure 7This is the second structural diagram of the cell energy-saving control device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0148] In the embodiments of the present invention, the term "and / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0149] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0150] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0151] See also Figure 1 , Figure 1 is a flow chart of a cell energy saving control method provided by an embodiment of the present invention, such as Figure 1 As shown, the following steps are included:

[0152] Step 101: Obtain historical data.

[0153] The historical data refers to the service data of the cell before the prediction period. The historical data mainly includes data reflecting the changes in the historical service volume of the cell, such as: uplink resource occupancy rate, downlink resource occupancy rate, uplink service volume (KB), downlink service volume (KB), number of users, etc. Figure 2 The figure shows an example of the obtained cell service history data, wherein the period to be predicted may be, for example, one day, one week, one month, etc.

[0154] Step 102: Use the historical data to predict the cell service data in the prediction period to obtain the service data change of the cell service data in the prediction period.

[0155] The business data change may refer to a change in a certain type of business data. In this step, taking the prediction period as a certain prediction day as an example, different models may be used for prediction depending on whether the prediction day is a working day or a weekend (Saturday or Sunday).

[0156] Specifically, when the time period to be predicted is a working day, the historical data and the first sub-model in the first prediction model are used to predict the cell service data of the time period to be predicted, and the service data changes of the cell service data in the time period to be predicted are obtained, wherein the historical data are the data of the M working days before the time period to be predicted; when the time period to be predicted is a weekend, the historical data and the second sub-model in the first prediction model are used to predict the cell service data of the time period to be predicted, and the service data changes of the cell service data in the time period to be predicted are obtained, wherein the historical data are the data of the N weekends before the time period to be predicted; wherein the first sub-model is used to predict the changes of the cell service data on weekdays, and the second sub-model is used to predict the changes of the cell service data on weekends, and M and N are both integers greater than 0.

[0157] Optionally, M=5, N=3.

[0158] Step 103: Determine a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell during the time period to be predicted.

[0159] In an embodiment of the present invention, the business data of each sampling point of the time period to be predicted is compared with a preset threshold value according to the change of the business data, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval. For example, the time period to be predicted (taking the time period to be predicted as one day, 24 hours as an example) is divided into 24 sampling points, and the interval between two adjacent sampling points is 1 hour. That is, each hour is regarded as a sampling point. The preset threshold value can be set as needed.

[0160] If the service data of the target sampling point is equal to the preset threshold, the period between the previous sampling point and the next sampling point of the target sampling point is taken as the sensitive period, and the target sampling point is any sampling point among the multiple sampling points. Then, the other periods in the period to be predicted except the sensitive period can be taken as the stable period.

[0161] For example, assuming that the business data at 12 o'clock is equal to the threshold, then the time period from 11:00 to 13:00 can be used as the sensitive time period.

[0162] like Figure 3 As shown in FIG. , it is assumed that this is a schematic diagram of the relationship between the business changes and thresholds in the period to be predicted. Figure 3 As shown in FIG. 1 , by comparing the service data of each sampling point with the threshold, the stable period and sensitive period of the cell can be determined. Taking the downlink resource occupancy rate as an example, the threshold Q = 20. Figure 3In the figure, the downlink resource occupancy rate at 11:00, 12:00, 17:00, 18:00, etc. is equal to 20. Therefore, these moments can be used as critical moments to determine the sensitive time periods: [10:00-12:00], [16:00-18:00], [17:00-19:00], [11:00-13:00], [20:00-22:00], [11:00-13:00], etc. Then, except for the sensitive time periods, all other time periods can be used as service stable time, such as [0:00-1:00], etc.

[0163] Step 104: Obtain a first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period.

[0164] Specifically, in this step, in the stable period, determining the first energy saving strategy includes putting the cell into sleep or wake-up, and in the sensitive period, determining the first energy saving strategy includes waking up the cell. In this way, a long-term energy saving strategy is determined for the predicted period.

[0165] Step 105: predict the stable period to obtain a second energy-saving strategy for the stable period, and predict the sensitive period to obtain a third energy-saving strategy for the sensitive period.

[0166] In the embodiment of the present invention, different prediction models are used to predict the stable period and the sensitive period to determine the energy saving strategy. When predicting the service stable time, a larger time granularity prediction can be performed, while when predicting the sensitive period, a shorter time granularity prediction can be performed.

[0167] 1. Predicting the stable period to obtain a second energy-saving strategy for the stable period, including:

[0168] (1) using the first sub-model in the second prediction model corresponding to the stable period to predict business changes in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval.

[0169] The first time period may be, for example, 2 hours, and the second preset interval may be arbitrarily set, for example, 15 minutes. The historical data used in the prediction may be historical data before the current moment.

[0170] Specifically, in the embodiment of the present invention, for the current moment in the stable period, the business changes in the next 2 hours can be predicted. Then, the 2 hours can be divided into 8 sampling points at intervals of 15 minutes. Through this step, the predicted business volume of the 8 sampling points can be obtained.

[0171] (2) Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups.

[0172] For the 8 sampling points obtained in (1), starting from the first sampling point, two adjacent sampling points are combined, so that 4 groups can be formed here.

[0173] (3) If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at the time corresponding to each sampling point in the first target group.

[0174] If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group.

[0175] If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

[0176] For the four groups obtained in (2), the predicted traffic volume of the sampling points in each group is compared with the preset threshold. If the traffic volume is greater than or equal to the preset threshold, the corresponding sampling point is marked as 1, otherwise it is marked as 0. Then, after comparison, multiple arrangements such as 00110011, 00000000, 11111111, and 01010101 can be formed.

[0177] If the traffic volume corresponding to each sampling point in a group is greater than or equal to a preset threshold, that is, the group is marked as 11, then the second energy-saving strategy includes waking up the cell at the time corresponding to each sampling point in the first target group.

[0178] If the traffic volume corresponding to each sampling point in a group is less than a preset threshold, that is, the group is marked as 00, the second energy-saving strategy includes putting the cell to sleep at the time corresponding to each sampling point in the first target group.

[0179] If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

[0180] That is, if the marks of the groups form an alternating arrangement such as 01010101 or 10101010, then the second energy saving strategy includes waking up the cell.

[0181] 2. Predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period, including:

[0182] (1) Using the second sub-model in the second prediction model corresponding to the sensitive time period, predict the business changes in the second time period after the current moment.

[0183] The second time period includes a plurality of sampling points arranged at a third preset interval, the length of the second time period is less than the length of the first time period, and the third preset interval is less than the second preset interval. The historical data used in the prediction may be historical data before the current moment.

[0184] For example, the second time period may be 30 minutes after the current moment, and the third preset interval may be set as required, for example, to 1 minute. Then, the second time period may be set to include 30 sampling points.

[0185] (2) Divide the sampling points in the second time period into multiple groups.

[0186] In the embodiment of the present invention, the sampling points can be divided into groups in various ways, for example, every two adjacent sampling points among the sampling points are regarded as a group, etc. In combination with the example in (1), the multiple sampling points can be divided into two groups, that is, starting from the first sampling point, every 15 sampling points form a group.

[0187] (3) If, in the second target group, the ratio of the number of sampling points whose traffic volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, the third energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at a time corresponding to each sampling point in the second target group.

[0188] For example, if the traffic volumes corresponding to more than 8 sampling points among 15 sampling points are greater than the preset threshold, then the third energy-saving strategy includes waking up the cell at the time corresponding to each sampling point, otherwise the third energy-saving strategy includes putting the cell to sleep at the time corresponding to each sampling point.

[0189] Optionally, if the business volume corresponding to 15 consecutive points in a group is greater than a preset threshold, the cell will be awakened at the time corresponding to each sampling point to ensure network quality; if the business volume corresponding to 15 consecutive points in a group is less than the preset threshold, the cell will continue to be dormant to tap the energy-saving potential.

[0190] In this way, more fine-grained predictions can be made for key moments within sensitive periods, which can be combined with short-term prediction results to achieve resistance to sudden events and fluctuations, and to timely adjust the cell's sleep or wake-up to ensure user perception.

[0191] Step 106: Based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable period and the sensitive period.

[0192] Specifically, during the stable period, based on the first energy-saving strategy and the second energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction results of the stable period; during the sensitive period, based on the first energy-saving strategy and the third energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction results of the energy-saving period, the time window is increased, and user perception is protected.

[0193] From the above description, it can be seen that by using the solution of the embodiment of the present invention, not only the long-term service change law of the day to be tested can be obtained, but also the service change in a short period of time can be predicted, so that by using the solution of the embodiment of the present invention, not only the long-term energy-saving strategy can be obtained, but also different energy-saving strategies can be set for short time periods, and no manual participation is required. Therefore, by using the solution of the embodiment of the present invention, the energy-saving period can be more accurately configured according to the change of cell service, the energy-saving potential can be tapped, and the impact on the network and the cost of energy-saving control can be reduced.

[0194] On the basis of the above embodiment, the embodiment of the present invention can also train the first sub-model and the second sub-model of the first prediction model, and train the first sub-model and the second sub-model of the second prediction model.

[0195] 1. Training the first prediction model

[0196] Specifically, training the first sub-model includes:

[0197] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0198] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0199] Wherein, training the second sub-model includes:

[0200] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0201] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0202] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0203] In practical applications, a detailed statistical analysis is made on the distribution characteristics of base station traffic in both time and space dimensions based on actual base station traffic data. The business models of different cells are quite different. For example, residential areas and industrial areas will present different business change models within a day. However, the business models of the same cell are relatively similar in the time dimension. For example, the same cell has similar traffic change models on many days of the week, such as Monday and Tuesday. Further, a correlation analysis is performed on the business changes of the cell on different days of the week. The weekdays show a strong correlation, and Saturdays and Sundays show a strong correlation. Therefore, in an embodiment of the present invention, modeling is performed for weekdays and weekends respectively, and the first sub-model and the second sub-model of the first prediction model are generated to predict the business conditions on weekdays and weekends, respectively.

[0204] Specifically, when modeling for working days, for the same indicator, the data of m working days before the current forecast day is taken, and each working day has n data (determined by the time granularity). Then the service change sampling point data of the current cell for one day can be shown in the following matrix:

[0205]

[0206] Among them, t p,q It represents the qth data of the pth day, 1≤p≤m, 1≤q≤n.

[0207] The business volume on the forecast day is:

[0208]

[0209] That is, the average value of the data of the previous m working days at the same time point is the business volume at that time on the predicted day. Taking the prediction of downlink resource occupancy rate as an example, the data collection time granularity is 15 minutes, then there are 96 data in a day, n = 96. According to the actual data of the existing network, the MSE (Mean Square Error) between the actual value and the predicted value is used as the optimization target. Finally, when m = 5, this model achieves the best prediction effect.

[0210] 2. Training the Second Prediction Model

[0211] During the training process, the first sub-model and the second sub-model are trained respectively using the historical cell service data of the target day before the period to be predicted and a long short-term memory network (Long Short-Term Memory, LSTM);

[0212] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0213] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0214] LSTM has its own unique memory mode and forgetting mode, which enables the network to fully explore the temporal characteristics of time series data and learn the temporal dependency between input data. Figure 4 As shown in FIG. 1 , the processing flow of the LSTM network model includes: collecting historical data, preprocessing the collected historical data, extracting features of the data through a long short-term memory network, fitting the wireless network traffic data, and then predicting the wireless network traffic data at the next moment.

[0215] like Figure 5 As shown in Figure 1, the structure of the LSTM used for short-term prediction of network traffic is shown. After preprocessing the original data set, p training set samples are obtained, each of which has T traffic data. The corresponding label is {y (1) ,y (2) ,...,y (i) ,...,y (p)}, where x (i) represents the i-th sample, represents the tth data of the i-th sample, y (i) It represents the label corresponding to the i-th sample, that is, the traffic value at the next moment predicted by the i-th sample.

[0216] The i-th sample x (i) The wireless network traffic value y at the next moment predicted as the input of the long short-term memory network (i) It takes T time steps to train a network for a sample, and each time step corresponds to an LSTM cell. The long short-term memory network uses a self-loop method, and only one data can be input into the network at each moment.

[0217] In this step, historical data and LSTM models are mainly used to train the first sub-model (a model that predicts the next 2 hours every 15 minutes), and the first 96 data (i.e., data with a granularity of 15 minutes per day), i.e., T=96, are used to predict the business volume of the next 2 hours (8 points). On the other hand, in order to mine more fine-grained time series features, historical data and LSTM models are used to train the second sub-model (a model that predicts the next 0.5 hours (30 points) every 1 minute). During the training process, the first 360 (T=360) data (i.e., 6 hours of data) are used to predict the business volume of the next 0.5 hours. The LSTM model can be used for more accurate short-term predictions.

[0218] In the embodiment of the present invention, the long-term and short-term prediction methods are effectively combined to solve the problems of unreasonable configuration and low accuracy of existing cell energy-saving schemes. By using long-term prediction to predict the long-term change rules of cell services, the sleep and wake-up windows can be determined in a steady state, avoiding frequent activation of power-saving (or wake-up) operations, and providing prediction time for network security, which is more secure. At the same time, on the basis of long-term prediction and combined with short-term prediction results, it can resist sudden and fluctuations, adjust sleep or wake-up in time, and ensure user perception. The entire strategy first predicts the business model of the cell over a long period of time, determines the sleep and wake-up windows in a steady state, and then combines the short-term prediction results to perform adaptive adjustment of details, ensuring user perception while reducing the amount of calculation.

[0219] The embodiment of the present invention also provides a cell energy saving control device. Figure 6 , Figure 6 Since the principle of the cell energy-saving control device to solve the problem is similar to the cell energy-saving control method in the embodiment of the present invention, the implementation of the cell energy-saving control device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0220] like Figure 6 As shown, the cell energy-saving control device 600 includes:

[0221] The first acquisition module 601 is used to acquire historical data;

[0222] The second acquisition module 602 is used to predict the cell service data in the prediction period by using the historical data to obtain the service data change of the cell service data in the prediction period;

[0223] A first determination module 603 is used to determine a stable period and a sensitive period of the period to be predicted according to a change in service data of the cell service data within the period to be predicted;

[0224] A second determination module 604 is used to obtain a first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period;

[0225] The third determination module 605 is used to predict the stable period to obtain the second energy-saving strategy for the stable period, and to predict the sensitive period to obtain the third energy-saving strategy for the sensitive period;

[0226] The first processing module 606 is used to perform energy-saving control on the stable period and the sensitive period based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively.

[0227] Wherein, the second acquisition module is used for:

[0228] When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted;

[0229] When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted;

[0230] Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

[0231] Wherein, the first determining module includes:

[0232] A first comparison submodule, configured to compare the business data of each sampling point in the time period to be predicted with a preset threshold value according to the change of the business data, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval;

[0233] A first determination submodule is configured to, if the service data of a target sampling point is equal to the preset threshold, use the period between the previous sampling point and the next sampling point of the target sampling point as the sensitive period, wherein the target sampling point is any sampling point among the multiple sampling points;

[0234] The second determining submodule is used to use other time periods in the time period to be predicted except the sensitive time period as the stable time period.

[0235] The second determination module is used to, during the stable period, determine that the first energy-saving strategy includes putting the cell into sleep or wake-up; during the sensitive period, determine that the first energy-saving strategy includes waking up the cell.

[0236] Wherein, the third determination module includes:

[0237] A first prediction submodule, configured to predict business changes in a first time period after a current moment using a first submodel in a second prediction model corresponding to the stable time period, wherein the first time period includes a plurality of sampling points set at a second preset interval;

[0238] A first processing submodule is used to form two adjacent sampling points among the multiple sampling points into a group starting from the first sampling point, to obtain multiple groups;

[0239] A first determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, then the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group;

[0240] A second determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, then the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group;

[0241] The third determination submodule is used to determine that if, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy-saving strategy includes waking up the cell.

[0242] Wherein, the third determination module includes:

[0243] A second prediction submodule, configured to use a second submodel in a second prediction model corresponding to the sensitive time period to predict a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval;

[0244] A second processing submodule, used for dividing the sampling points in the second time period into a plurality of groups;

[0245] The fourth determination submodule is used to determine that if the ratio of the number of sampling points in the second target group whose business volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, then the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

[0246] Wherein, the first processing module includes:

[0247] A first processing submodule is used to dynamically adjust the wake-up and sleep of the cell during the stable period based on the first energy-saving strategy and the second energy-saving strategy according to the prediction result of the stable period;

[0248] The second processing submodule is used to dynamically adjust the wake-up and sleep of the cell during the sensitive period based on the first energy-saving strategy and the third energy-saving strategy according to the prediction result of the energy-saving period.

[0249] The device further comprises: a first training module, configured to train a first sub-model and a second sub-model of the first prediction model;

[0250] Wherein, when training the first sub-model, the first training module is used to:

[0251] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0252] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0253] Wherein, when training the second sub-model, the first training module is used to:

[0254] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0255] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0256] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0257] The device further comprises: a second training module, configured to train the first sub-model and the second sub-model of the second prediction model;

[0258] Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively;

[0259] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0260] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0261] The device provided in the embodiment of the present invention can execute the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0262] It should be noted that the division of units in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0263] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0264] The embodiment of the present invention also provides a cell energy saving control device. Figure 7 , Figure 7 Since the principle of the cell energy-saving control device to solve the problem is similar to the cell energy-saving control method in the embodiment of the present invention, the implementation of the cell energy-saving control device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0265] like Figure 7As shown, the cell energy saving control device 700 includes: a processor 701 and a transceiver 702;

[0266] The processor 701 is configured to:

[0267] Get historical data;

[0268] Using the historical data to predict the cell service data in the prediction period, and obtaining the service data change of the cell service data in the prediction period;

[0269] Determining a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell service data within the time period to be predicted;

[0270] According to the stable period and the sensitive period, a first energy-saving strategy for the period to be predicted is obtained;

[0271] The stable period is predicted to obtain a second energy-saving strategy for the stable period, and the sensitive period is predicted to obtain a third energy-saving strategy for the sensitive period.

[0272] The processor 701 is used for:

[0273] When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted;

[0274] When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted;

[0275] Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

[0276] The processor 701 is used for:

[0277] According to the change of the business data, comparing the business data of each sampling point in the time period to be predicted with a preset threshold, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval;

[0278] If the service data of the target sampling point is equal to the preset threshold, the period between the previous sampling point and the next sampling point of the target sampling point is used as the sensitive period, and the target sampling point is any sampling point among the multiple sampling points;

[0279] The other time periods in the time period to be predicted except the sensitive time period are used as the stable time periods.

[0280] The processor 701 is used for:

[0281] During the stable period, determining the first energy saving strategy includes putting the cell into sleep or wake-up;

[0282] During the sensitive period, determining the first energy saving strategy includes waking up the cell.

[0283] Wherein, the processor is used for:

[0284] Using the first sub-model in the second prediction model corresponding to the stable period, predicting the business change in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval;

[0285] Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups;

[0286] If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group;

[0287] If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group;

[0288] If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

[0289] The processor 701 is used for:

[0290] Using a second sub-model in a second prediction model corresponding to the sensitive period, predicting a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval;

[0291] Dividing the sampling points in the second time period into a plurality of groups;

[0292] If, in the second target group, the ratio of the number of sampling points whose traffic volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

[0293] Wherein, the processor is used for:

[0294] In the stable period, based on the first energy-saving strategy and the second energy-saving strategy, dynamically adjusting the wake-up and sleep of the cell according to the prediction result of the stable period;

[0295] During the sensitive period, based on the first energy-saving strategy and the third energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction result of the energy-saving period.

[0296] The processor 701 is used for:

[0297] Training a first sub-model and a second sub-model of the first prediction model;

[0298] Wherein, when training the first sub-model, the processor is used to:

[0299] Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments;

[0300] The average value of the data corresponding to the tth moment in m working days is taken as the predicted value;

[0301] Wherein, when training the second sub-model, the processor is used to:

[0302] Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments;

[0303] Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value;

[0304] Among them, m, n, p, q, i, and t are all integers greater than 0.

[0305] Wherein, the processor is used for:

[0306] Training the first sub-model and the second sub-model of the second prediction model;

[0307] Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively;

[0308] Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90;

[0309] A plurality of sampling points are set for the target day according to the third preset interval, the first preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

[0310] The device provided in the embodiment of the present invention can execute the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0311] An embodiment of the present invention further provides a communication device, comprising: a transceiver, a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor implements the steps in the cell energy saving control method as described above when executing the program.

[0312] The embodiment of the present invention also provides a readable storage medium, on which a program is stored. When the program is executed by the processor, each process of the above-mentioned cell energy-saving control method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. Among them, the readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to magnetic storage (such as floppy disk, hard disk, magnetic tape, magneto-optical disk (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid state drive (SSD)), etc.

[0313] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0314] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for a terminal (which can be a mobile phone, computer, server, air conditioner, or network equipment, etc.) to execute the methods described in each embodiment of the present invention.

[0315] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A cell energy saving control method, characterized in that: include: Get historical data; Using the historical data to predict the cell service data in the prediction period, and obtaining the service data change of the cell service data in the prediction period; Determining a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell service data within the time period to be predicted; According to the stable period and the sensitive period, a first energy-saving strategy for the period to be predicted is obtained; Predicting the stable period to obtain a second energy-saving strategy for the stable period, and predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period; Based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable period and the sensitive period; The predicting of the stable period to obtain a second energy-saving strategy for the stable period includes: Using the first sub-model in the second prediction model corresponding to the stable period, predicting the business change in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval; Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups; If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group; If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group; If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

2. The method according to claim 1, characterized in that The using the historical data to predict the cell service data in the prediction period to obtain the service data change of the cell service data in the prediction period includes: When the period to be predicted is a working day, the cell service data of the period to be predicted is predicted using the historical data and the first sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of M working days before the period to be predicted; When the period to be predicted is a weekend, the cell service data of the period to be predicted is predicted using the historical data and the second sub-model in the first prediction model to obtain the service data change of the cell service data in the period to be predicted, wherein the historical data is the data of N weekends before the period to be predicted; Among them, the first sub-model is used to predict the change of cell service data on weekdays, and the second sub-model is used to predict the change of cell service data on weekends. M and N are both integers greater than 0.

3. The method according to claim 1, characterized in that The determining, according to the change of the service data of the cell service data in the time period to be predicted, the stable time period and the sensitive time period of the time period to be predicted comprises: According to the change of the business data, comparing the business data of each sampling point in the time period to be predicted with a preset threshold, wherein the time period to be predicted includes a plurality of sampling points set at a first preset interval; If the service data of the target sampling point is equal to the preset threshold, the period between the previous sampling point and the next sampling point of the target sampling point is used as the sensitive period, and the target sampling point is any sampling point among the multiple sampling points; The other time periods in the time period to be predicted except the sensitive time period are used as the stable time periods.

4. The method according to claim 1, characterized in that: The step of obtaining a first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period includes: During the stable period, determining the first energy saving strategy includes putting the cell into sleep or wake-up; During the sensitive period, determining the first energy saving strategy includes waking up the cell.

5. The method according to claim 1, characterized in that The predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period includes: Using a second sub-model in a second prediction model corresponding to the sensitive period, predicting a service change in a second time period after the current moment, wherein the second time period includes a plurality of sampling points arranged at a third preset interval, a length of the second time period is less than a length of the first time period, and the third preset interval is less than the second preset interval; Dividing the sampling points in the second time period into a plurality of groups; If, in the second target group, the ratio of the number of sampling points whose traffic volume is greater than the preset threshold to the number of sampling points included in the second target group is greater than 1 / 2, the third energy-saving strategy includes waking up the cell at the moment corresponding to each sampling point in the second target group; otherwise, the third energy-saving strategy includes putting the cell to sleep at the moment corresponding to each sampling point in the second target group.

6. The method according to claim 1, characterized in that The performing energy-saving control on the stable period and the sensitive period respectively based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy includes: In the stable period, based on the first energy-saving strategy and the second energy-saving strategy, dynamically adjusting the wake-up and sleep of the cell according to the prediction result of the stable period; During the sensitive period, based on the first energy-saving strategy and the third energy-saving strategy, the wake-up and sleep of the cell are dynamically adjusted according to the prediction result of the sensitive period.

7. The method according to claim 2, characterized in that The method further comprises: Training a first sub-model and a second sub-model of the first prediction model; Wherein, training the first sub-model includes: Obtain historical cell service data for m working days before the period to be predicted, where each working day includes n data corresponding to n moments; The average value of the data corresponding to the tth moment in m working days is taken as the predicted value; Wherein, training the second sub-model comprises: Obtain historical cell service data for p weekends before the period to be predicted, where each weekend includes q data corresponding to q moments; Take the average value of the data corresponding to the i-th moment in p weekends as the predicted value; Among them, m, n, p, q, i, and t are all integers greater than 0.

8. The method according to claim 5, characterized in that The method further comprises: Training the first sub-model and the second sub-model of the second prediction model; Using historical cell service data of a target day before the time period to be predicted and a long short-term memory network to train the first sub-model and the second sub-model respectively; Wherein, a plurality of sampling points are set for the target day according to the second preset interval, the second preset interval is 15 minutes, and when training the first sub-model, the training step length is 90; A plurality of sampling points are set for the target day according to the third preset interval, the third preset interval is 1 minute, and when training the second sub-model, the training step length is 360.

9. A cell energy-saving control device, characterized in that: include: A first acquisition module is used to acquire historical data; A second acquisition module is used to predict the cell service data of the to-be-predicted period using the historical data, and obtain the service data change of the cell service data in the to-be-predicted period; A first determination module, configured to determine a stable period and a sensitive period of the time period to be predicted according to a change in service data of the cell service data within the time period to be predicted; A second determination module is used to obtain a first energy-saving strategy for the time period to be predicted according to the stable time period and the sensitive time period; A third determination module is used to predict the stable period to obtain a second energy-saving strategy for the stable period, and to predict the sensitive period to obtain a third energy-saving strategy for the sensitive period; A first processing module, configured to perform energy-saving control on the stable period and the sensitive period respectively based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy; Wherein, the third determination module includes: A first prediction submodule, configured to predict business changes in a first time period after a current moment using a first submodel in a second prediction model corresponding to the stable time period, wherein the first time period includes a plurality of sampling points set at a second preset interval; A first processing submodule is used to form two adjacent sampling points among the multiple sampling points into a group starting from the first sampling point, to obtain multiple groups; A first determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, then the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group; A second determining submodule, configured to, if the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, then the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group; The third determination submodule is used to determine that if, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy-saving strategy includes waking up the cell.

10. A cell energy-saving control device, characterized in that: include: processors and transceivers; The processor is configured to: Get historical data; Using the historical data to predict the cell service data in the prediction period, and obtaining the service data change of the cell service data in the prediction period; Determining a stable period and a sensitive period of the time period to be predicted according to changes in the service data of the cell service data within the time period to be predicted; According to the stable period and the sensitive period, a first energy-saving strategy for the period to be predicted is obtained; Predicting the stable period to obtain a second energy-saving strategy for the stable period, and predicting the sensitive period to obtain a third energy-saving strategy for the sensitive period; Based on the first energy-saving strategy, the second energy-saving strategy and the third energy-saving strategy respectively perform energy-saving control on the stable period and the sensitive period; Wherein, the processor is further used for: Using the first sub-model in the second prediction model corresponding to the stable period, predicting the business change in a first time period after the current moment, wherein the first time period includes a plurality of sampling points set at a second preset interval; Starting from the first sampling point, two adjacent sampling points among the plurality of sampling points are formed into a group to obtain a plurality of groups; If the traffic volume corresponding to each sampling point in the first target group is greater than or equal to a preset threshold, the second energy-saving strategy includes waking up the cell at a time corresponding to each sampling point in the first target group; If the traffic volume corresponding to each sampling point in the first target group is less than the preset threshold, the second energy-saving strategy includes putting the cell into sleep mode at the time corresponding to each sampling point in the first target group; If, among the multiple groups, the business volume of the first sampling point in each group is greater than or equal to the preset threshold, and the business volume of the second sampling point is less than the preset threshold, or the business volume of the first sampling point in each group is less than the preset threshold, and the business volume of the second sampling point is greater than or equal to the preset threshold, then the second energy saving strategy includes waking up the cell.

11. A communication device, comprising: A transceiver, a memory, a processor, and a program stored in the memory and executable on the processor; characterized in that the processor is used to read the program in the memory to implement the steps in the cell energy saving control method as described in any one of claims 1 to 8.

12. A readable storage medium for storing a program, characterized in that: When the program is executed by a processor, the steps of the cell energy saving control method according to any one of claims 1 to 8 are implemented.

Images