An energy-saving processing method, apparatus, device, and computer-readable storage medium

Abstract

This application discloses an energy-saving processing method, the method comprising: acquiring first historical service data of a target cell and current service data of the target cell; wherein the first historical service data is service data that characterizes the historical network usage of the target cell; determining a service data prediction model; processing the first historical service data based on the service data prediction model to obtain target service data of the target cell; and determining whether to perform energy-saving processing on the target cell based on the target service data and the current service data. This application also discloses an energy-saving processing device, apparatus, and computer-readable storage medium.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to an energy-saving processing method, apparatus, device, and computer-readable storage medium. Background Technology

[0002] While ensuring network coverage and user experience, energy-saving strategies are employed to achieve the energy conservation and emission reduction goals of "green base stations." Specifically, service data from a cell over a historical period is acquired, and a prediction model is trained using this data. Based on this model, the service data for a specific time period within the cell is predicted, and if the prediction result exceeds a data threshold, the cell is determined to be in an energy-saving state for that period. However, this method of determining whether a cell enters an energy-saving state based on prediction model results and data thresholds in related technologies is highly dependent on the data threshold and cannot accurately perform energy-saving processing on cells. Summary of the Invention

[0003] To address the aforementioned technical problems, this application aims to provide an energy-saving processing method, apparatus, device, and computer-readable storage medium. This solves the problem in related technologies where there is a high dependence on data thresholds, making it impossible to accurately perform energy-saving processing on cells. It can improve the efficiency of energy-saving processing and also ensure the accuracy of energy-saving processing.

[0004] The technical solution of this application is implemented as follows:

[0005] An energy-saving treatment method, the method comprising:

[0006] Obtain first historical service data and current service data of the target cell; wherein, the first historical service data is service data that can characterize the historical network usage of the target cell;

[0007] Determine the business data prediction model;

[0008] The first historical service data is processed based on the service data prediction model to obtain the target service data of the target cell.

[0009] Based on the target service data and the current service data, determine whether to perform energy-saving processing on the target cell.

[0010] In the above scheme, determining the business data prediction model includes:

[0011] Obtain the second historical service data of the sample cell; wherein, the second historical service data is service data of the sample cell that can characterize the historical network usage of the sample cell;

[0012] An initial business data prediction model is constructed using multiple convolutional layers determined based on multiple different inflation factors;

[0013] The initial business data prediction model is trained based on the second historical business data to obtain the business data prediction model.

[0014] In the above scheme, the step of training the initial business data prediction model based on the second historical business data to obtain the business data prediction model includes:

[0015] The multiple convolutional layers of the initial business data prediction model are determined based on the multiple different inflation factors; wherein the inflation factors increase sequentially with the number of convolutional layers.

[0016] The second historical business data is processed by convolution based on the multiple convolutional layers to obtain the first business data.

[0017] The business data that meets the target conditions is determined from the second historical business data to obtain the second business data;

[0018] The business data prediction model is obtained by training the model based on the second historical business data, the first business data, and the second business data.

[0019] In the above scheme, the step of training the model based on the second historical business data, the first business data, and the second business data to obtain the business data prediction model includes:

[0020] The third business data is obtained by performing residual processing on the first business data and the second historical business data;

[0021] The second and third service data are processed using a full connection to obtain the fourth service data.

[0022] The business data prediction model is obtained by training the model based on the fourth business data.

[0023] In the above scheme, the step of training the model based on the fourth business data to obtain the business data prediction model includes:

[0024] Determine the target loss function;

[0025] The service data prediction model is obtained by training the model based on the fourth service data, the third historical service data of the sample cell, and the target loss function; wherein, the third historical service data is generated by the sample cell after the time corresponding to the second historical service data.

[0026] In the above scheme, determining the target loss function includes:

[0027] Determine the initial loss function;

[0028] A first weight, a second weight, and a third weight are determined; wherein, the first weight is the weight corresponding to when the output data of the initial business data prediction model is greater than the third historical business data and less than the first target threshold, the second weight is the weight corresponding to when the output data is greater than the second target threshold, and the third weight is the weight corresponding to when the output data is greater than or equal to the first target threshold and less than or equal to the second target threshold, wherein the second target threshold is greater than the first target threshold;

[0029] The target loss function is determined based on the initial loss function, the first weight, the second weight, and the third weight.

[0030] In the above scheme, determining whether to perform energy-saving processing on the target cell based on the target service data and the current service data includes:

[0031] If the target service data is less than the first service data threshold, the target cell is determined to enter a power-saving state.

[0032] When the target cell enters a power-saving state and the current service data is less than the second service data threshold, power-saving processing is performed on the target cell.

[0033] An energy-saving treatment device, the device comprising:

[0034] An acquisition unit is used to acquire first historical service data of a target cell and current service data of the target cell; wherein the first historical service data is service data that can characterize the historical network usage of the target cell;

[0035] The determination unit is used to determine the business data prediction model;

[0036] The acquisition unit is further configured to process the first historical service data based on the service data prediction model to obtain the target service data of the target cell;

[0037] The determining unit is further configured to determine, based on the target service data and the current service data, whether to perform energy-saving processing on the target cell.

[0038] An energy-saving processing device, the device comprising: a processor, a memory, and a communication bus;

[0039] The communication bus is used to realize the communication connection between the processor and the memory;

[0040] The processor is used to execute the energy-saving processing program in the memory to implement the steps of the energy-saving processing method described above.

[0041] A computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the energy-saving processing method described above.

[0042] The energy-saving processing method, apparatus, device, and computer-readable storage medium provided in the embodiments of this application can acquire first historical service data and current service data of a target cell, determine a service data prediction model, then process the first historical service data based on the service data prediction model to obtain target service data of the target cell, and then determine whether to perform energy-saving processing on the target cell based on the target service data and the current service data. This method, which determines whether to perform energy-saving processing on the target cell based on the target service data and current service data pre-acquired by the service data prediction model, does not rely on a single data threshold. Since it determines whether to perform energy-saving processing on the target cell based on both the target service data and the current service data, it can more accurately perform energy-saving processing on the target cell. This solves the problem in related technologies where there is a high dependence on data thresholds, making it impossible to accurately perform energy-saving processing on cells. It can improve the efficiency of energy-saving processing and also ensure the accuracy of energy-saving processing. Attached Figure Description

[0043] Figure 1 A schematic flowchart of an energy-saving processing method provided for an embodiment of this application;

[0044] Figure 2 A schematic flowchart of another energy-saving processing method provided for an embodiment of this application;

[0045] Figure 3 A schematic diagram of feature serialization provided for an embodiment of this application;

[0046] Figure 4 A schematic diagram of a dilated causal convolution provided for an embodiment of this application;

[0047] Figure 5 A schematic flowchart of another energy-saving treatment method provided for an embodiment of this application;

[0048] Figure 6 A schematic diagram illustrating the model training of a business data prediction model provided for an embodiment of this application;

[0049] Figure 7 A schematic diagram of a business data trend chart provided for an embodiment of this application;

[0050] Figure 8 A schematic diagram of an energy-saving processing device provided for an embodiment of this application;

[0051] Figure 9 A schematic diagram illustrating the interaction between an energy-saving processing device and a client, provided as an embodiment of this application;

[0052] Figure 10 A schematic diagram of the structure of an energy-saving processing device provided for an embodiment of this application;

[0053] Figure 11 This is a schematic diagram of the structure of an energy-saving processing device provided for an embodiment of this application. Detailed Implementation

[0054] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0055] It should be understood that the phrases "embodiments of this application" or "foreign embodiments" throughout the specification mean that a specific feature, structure, or characteristic related to an embodiment is included in at least one embodiment of this application. Therefore, "embodiments of this application" or "in the foreign embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0056] Unless otherwise specified, any step in the embodiments of this application performed by the electronic device may be executed by the processor of the electronic device. It is also worth noting that the embodiments of this application do not limit the order in which the electronic device performs the following steps. Furthermore, the methods used to process data in different embodiments may be the same or different methods. It should also be noted that any step in the embodiments of this application can be executed independently by the electronic device; that is, when the electronic device performs any step in the following embodiments, it may not depend on the execution of other steps.

[0057] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0058] This application provides an energy-saving treatment method, which can be applied to energy-saving treatment equipment. (Refer to...) Figure 1 As shown, the method includes the following steps:

[0059] Step 101: Obtain the first historical service data and the current service data of the target cell.

[0060] Among them, the first historical service data can characterize the service data of the target cell’s historical network usage.

[0061] In this embodiment, the target cell is the cell whose service data needs to be predicted. The current service data is the service data acquired by the target cell at the current moment. Service data may include uplink traffic volume, downlink traffic volume, number of users, uplink resource utilization rate, and downlink resource utilization rate, etc. As one possible approach, the energy-saving processing device can acquire the service data of the target cell every fifteen minutes on a past day as the first historical service data, with a fifteen-minute time granularity.

[0062] Step 102: Determine the business data prediction model.

[0063] In this embodiment of the application, the business data prediction model is used to predict business data for the next time period, such as predicting business data every fifteen minutes on a future day.

[0064] Step 103: Process the first historical service data based on the service data prediction model to obtain the target service data of the target cell.

[0065] In this embodiment of the application, the target service data is the service data of the target cell in the next time period obtained based on the service data prediction model. For example, the target service data can be the service data of the target cell every fifteen minutes tomorrow, obtained based on the service data prediction model.

[0066] Step 104: Based on the target service data and the current service data, determine whether to implement energy-saving measures for the target community.

[0067] In this embodiment, the predictive device obtains target service data in advance based on a service data prediction model, and then determines whether to perform energy-saving processing on the target cell based on the target service data and the current service data. This can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0068] The energy-saving processing method provided in this application embodiment acquires first historical service data and current service data of a target cell, determines a service data prediction model, processes the first historical service data based on the service data prediction model to obtain target service data of the target cell, and then determines whether to perform energy-saving processing on the target cell based on the target service data and the current service data. This method, which determines whether to perform energy-saving processing on the target cell based on the target service data and current service data pre-acquired by the service data prediction model, does not rely on a single data threshold. Since it determines whether to perform energy-saving processing on the target cell based on both the target service data and the current service data, it can more accurately perform energy-saving processing on the target cell. This solves the problem in related technologies where there is a high dependence on data thresholds, making it impossible to accurately perform energy-saving processing on the cell. It can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0069] Based on the foregoing embodiments, this application provides an energy-saving processing method, referring to... Figure 2 As shown, the method includes the following steps:

[0070] Step 201: The energy-saving processing equipment acquires the first historical service data and the current service data of the target cell.

[0071] Among them, the first historical service data can characterize the service data of the target cell’s historical network usage.

[0072] Step 202: The energy-saving processing equipment acquires the second historical service data of the sample cell.

[0073] Among them, the second historical service data is the service data of the sample cell that can characterize the historical network usage of the sample cell.

[0074] In this embodiment, the sample cells can be n cells, and the second historical service data can be the service data of the historical network usage of n cells over t days. The second historical service data for each day is granular with a 15-minute time interval, totaling 96 time points. Optionally, before training the model based on the second historical service data, missing value backfilling and periodicity analysis can be performed on the second historical service data. As one feasible approach, missing service data time points can be backfilled using the average of preceding and following time points. Periodicity analysis can be performed using the autocorrelation function (ACF). If the ACF of a cell is greater than a periodicity threshold, the cell is determined to be periodic and can proceed to subsequent long-term prediction, i.e., model prediction. If the ACF of a cell is less than or equal to the periodicity threshold, the cell is determined not to be periodic, and statistical methods are used to predict the target service data for future time points, such as using the maximum value of past service data at the same time as the target service data for future time points.

[0075] In this embodiment of the application, when using deep learning to solve the time series prediction problem, it is usually necessary to transform the time series data into supervised data, i.e., paired input-output data. Taking the second historical service data of the i-th cell as an example, the time series data can be: x 1，1 (i) ,x 1，2 (i) ,...,x 1，96 (i) ,...,x T，1 (i) ,...,x T，96 (i) The time series data is divided into multiple subsequences using a preset sliding window, and then transformed into supervised feature sequences before being input into a temporal convolutional neural network. The specific division method is as follows: Figure 3 As shown, t is the sliding window size, which predicts the data for day t+1 based on the data from the previous t days. Since there are 96 time points each day, the input X has dimensions (i,j), and the output Y has dimensions (i,j), where i is the number of days (range [1, t]) and j is the time point (range [1, 96]). The same sliding window processing is applied to the remaining n-1 cell data to obtain supervised feature sequences for all cells. Then, all feature sequences are shuffled and randomly divided into training and test sets in a 9:1 ratio. Next, it is possible to further refine the data based on... The training set is normalized, where, This represents the normalized data, where μ is the mean of the input sequence and σ is the variance of the input sequence.

[0076] Step 203: The energy-saving processing equipment uses multiple convolutional layers determined based on multiple different expansion factors to construct an initial business data prediction model.

[0077] In this embodiment, the inflation factor of each convolutional layer in the initial business data prediction model is different. The initial business data prediction model can be constructed based on a temporal convolutional network (TCN). However, when training with business data from a long time ago, the number of convolutional layers in a TCN increases significantly. By using multiple convolutional layers determined by different inflation factors to construct the initial business data prediction model, the neural network can have a larger field of view with the same number of layers, reducing computational load and network complexity.

[0078] Step 204: The energy-saving processing equipment trains the initial business data prediction model based on the second historical business data to obtain the business data prediction model.

[0079] Step 204 can be implemented in the following way:

[0080] Step 204a: The energy-saving processing equipment determines multiple convolutional layers of the initial business data prediction model based on multiple different expansion factors.

[0081] The dilation factor increases sequentially with the number of convolutional layers.

[0082] In the embodiments of this application, such as Figure 4 As shown, h is the input data for the initial business data prediction model. i The output data of the initial business data prediction model has a convolution kernel of 2. The inflation factor d of the initial business data prediction model is 1, 2, and 4 from bottom to top. That is, the inflation factor of the first convolutional layer of the initial business data prediction model is 1, the inflation factor of the second convolutional layer is 2, and the inflation factor of the third convolutional layer is 4. This allows the neural network to have a larger field of view with the same number of layers, reducing the amount of computation and the complexity of the network.

[0083] Step 204b: The energy-saving processing equipment performs convolution processing on the second historical business data based on multiple convolutional layers to obtain the first business data.

[0084] In this embodiment of the application, the first business data is the data obtained by convolution processing the second historical business data.

[0085] Step 204c: The energy-saving processing equipment determines the business data that meets the target conditions from the second historical business data, and obtains the second business data.

[0086] In this embodiment of the application, the target condition can be the maximum value of the service data at each time point, so that the second service data is the maximum value of the service data at each time point determined from the second historical service data.

[0087] Step 204d: The energy-saving processing equipment trains a model based on the second historical business data, the first business data, and the second business data to obtain a business data prediction model.

[0088] In the embodiments of this application, the business data prediction model trained based on the second historical business data, the first business data, and the second business data can make the prediction results as close as possible to the true values, thereby improving the accuracy of the prediction.

[0089] Step 205: The energy-saving processing equipment processes the first historical business data based on the business data prediction model to obtain the target business data of the target cell.

[0090] Step 206: When the target service data is less than the first service data threshold, the energy-saving processing equipment determines that the target cell has entered the energy-saving state.

[0091] In this embodiment, the first service data threshold can be pre-set; for example, the first service data threshold can be set to: 2 online users. If the target service data is less than the first service data threshold, it indicates that the current network utilization rate of the target cell is not high, and it can enter a power-saving state; if the target service data is greater than or equal to the first service data threshold, it indicates that the current network utilization rate of the target cell is high, and it should maintain a normal network state.

[0092] In this embodiment, the first service data can be the service data of the target cell every 15 minutes on a certain day in the past. Then, the target service data obtained based on the service data prediction model can be the service data every 15 minutes tomorrow. Based on the first service data threshold, it can be determined whether the target cell enters the energy-saving state at each time point tomorrow. Cells that enter the energy-saving state at each time point are added to the deactivated cell list, and cells that do not enter the energy-saving state at each time point are added to the wake-up cell list, so as to facilitate the processing of the network status of the cell at each time point according to these two lists.

[0093] Step 207: When the target cell enters the energy-saving state and the current business data is less than the second business data threshold, the energy-saving processing equipment performs energy-saving processing on the target cell.

[0094] In this embodiment, the second service data threshold may be the same as or different from the first service data threshold, and can be set according to actual service needs. This embodiment does not limit this. Energy-saving processing is performed on the cell when it enters a power-saving state and the current service data is less than the second service data threshold. This means focusing on cells in the deactivated cell list, and performing energy-saving processing on cells when the current service data of cells in the obtained deactivated cell list is less than the second service data threshold.

[0095] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0096] The energy-saving processing method provided in this application, which determines whether to perform energy-saving processing on a target cell based on the target service data and current service data obtained in advance by the service data prediction model, does not rely on a single data threshold. Instead, it determines whether to perform energy-saving processing on a target cell based on both the target service data and the current service data. Therefore, it can perform energy-saving processing on the target cell more accurately, solving the problem in related technologies that rely heavily on data thresholds and cannot accurately perform energy-saving processing on cells. This can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0097] Based on the foregoing embodiments, this application provides an energy-saving processing method, referring to... Figure 5 As shown, the method includes the following steps:

[0098] Step 301: The energy-saving processing equipment acquires the first historical service data and the current service data of the target cell.

[0099] Among them, the first historical service data can characterize the service data of the target cell’s historical network usage.

[0100] Step 302: The energy-saving processing equipment acquires the second historical business data of the sample cell.

[0101] Among them, the second historical service data is the service data of the sample cell that can characterize the historical network usage of the sample cell.

[0102] Step 303: The energy-saving processing equipment uses multiple convolutional layers determined based on multiple different expansion factors to construct an initial business data prediction model.

[0103] Step 304: The energy-saving processing equipment determines multiple convolutional layers of the initial business data prediction model based on multiple different expansion factors.

[0104] The dilation factor increases sequentially with the number of convolutional layers.

[0105] Step 305: The energy-saving processing equipment performs convolution processing on the second historical business data based on multiple convolutional layers to obtain the first business data.

[0106] In the embodiments of this application, such as Figure 6 As shown, after performing convolution processing on the second historical business data, the energy-saving processing equipment can also perform normalization processing, activation processing through the ReLU (Rectified LinearUnits) activation function, and dropout processing on the convolutional business data. Then, it continues to perform convolution processing, normalization processing, activation processing, and dropout processing on the next layer until the first business data is obtained.

[0107] Step 306: The energy-saving processing equipment determines the business data that meets the target conditions from the second historical business data to obtain the second business data.

[0108] Step 307: The energy-saving processing equipment performs residual processing on the first business data and the second historical business data to obtain the third business data.

[0109] In this embodiment, to avoid the gradient vanishing and gradient exploding problems that occur in temporal convolutional neural networks, a residual module is introduced. For example... Figure 6 As shown, the residual module can be defined as: H(x) = F(x) + x, where H(x) is the output of the residual module, and F(x) is the residual mapping that the network needs to learn. x represents the second historical service data. Considering that a shutdown strategy will only be adopted when the model prediction result is less than the threshold of the first service data, to avoid incorrect cell location and shutdown due to excessively small prediction results, the expected prediction result should be as close as possible to the true value while being greater than the first service data threshold. Therefore, a full connection is added after the residual module network structure to represent the maximum value of the service data at each time step. This ensures a certain degree of boosting of the predicted value, which is more in line with the application scenario of energy-saving shutdown. The third service data is the service data obtained after residual processing of the first and second historical service data, which is the output H(x) of the residual module.

[0110] Step 308: The energy-saving processing equipment performs full connection processing on the second and third business data to obtain the fourth business data.

[0111] In this embodiment, the fourth service data is the service data obtained after performing full connection processing on the second and third service data.

[0112] Step 309: The energy-saving treatment equipment trains a model based on the fourth business data to obtain a business data prediction model.

[0113] Step 309 can be implemented in the following way:

[0114] Step 309a: Determine the target loss function for the energy-saving treatment equipment.

[0115] In this embodiment of the application, in order to better fit the application scenario of energy-saving shutdown, considering that the value of business data in different ranges will have different impacts on the shutdown result, different weights are applied to the loss of different second historical business data to improve the accuracy of shutdown.

[0116] Step 309a can be implemented in the following way:

[0117] Step A: Determine the initial loss function for the energy-saving treatment equipment.

[0118] In this embodiment of the application, the initial loss function can be: y i The output of the initial business data prediction model. This is the third historical business data, which is the actual business data, and N is the number of business data.

[0119] Step B: Determine the first, second, and third weights for the energy-saving treatment equipment.

[0120] The first weight is the weight corresponding to the output data of the initial business data prediction model being greater than the third historical business data and less than the first target threshold; the second weight is the weight corresponding to the output data being greater than the second target threshold; and the third weight is the weight corresponding to the output data being greater than or equal to the first target threshold and less than or equal to the second target threshold, wherein the second target threshold is greater than the first target threshold.

[0121] In this embodiment, the first target threshold may be a first business data threshold, and the second target threshold is larger than the first target threshold. Preferably, the second target threshold may be five times or more than the first target threshold. Figure 7 As shown, the first weight is used when the output data of the initial business data prediction model is greater than the third historical business data and less than the first target threshold. This output data is business data near the shutdown threshold. Figure 6 (point P in the equation), therefore the first weight can be increased, and the first weight can be greater than 1. The second weight is used when the output data is greater than the second target threshold; this output data is business data far from the shutdown threshold. Figure 6The second weight can be reduced to less than 1 because it is the Q point in the initial business data prediction model. The third weight is used when the output data of the initial business data prediction model is between the first and second target thresholds, so the third weight can be 1.

[0122] Step C: The energy-saving treatment equipment determines the target loss function based on the initial loss function, the first weight, the second weight, and the third weight.

[0123] In this embodiment, the target loss function can be loss = α·mse1 + β·mse2 + mse3, where loss is the target loss function, α is the first weight, and β is the second weight. And y i <a, a is the first target threshold, b is the second target threshold, and y is the third target threshold. i The output of the initial business data prediction model. This is the third historical business data, which is the actual business data, and N is the number of business data.

[0124] Step 309b: The energy-saving processing equipment trains a model based on the fourth business data, the third historical business data, and the target loss function to obtain a business data prediction model.

[0125] The third historical business data was acquired after the acquisition time of the second historical business data.

[0126] In this embodiment, the difference between the fourth business data and the third historical business data is determined based on the target loss function, and the model parameters are adjusted accordingly until a business data prediction model is trained. The business data prediction model trained based on the fourth business data, the third historical business data, and the target loss function can make the prediction results as close as possible to the true values, thus improving the prediction accuracy.

[0127] Step 310: The energy-saving processing equipment processes the first historical business data based on the business data prediction model to obtain the target business data of the target cell.

[0128] In this embodiment of the application, the target service data is the service data of the target cell obtained based on the service data prediction model.

[0129] Step 311: When the target service data is less than the first service data threshold, the energy-saving processing equipment determines that the target cell has entered the energy-saving state.

[0130] Step 312: When the target cell enters the energy-saving state and the current business data is less than the second business data threshold, the energy-saving processing equipment performs energy-saving processing on the target cell.

[0131] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0132] The energy-saving processing method provided in this application, which determines whether to perform energy-saving processing on a target cell based on the target service data and current service data obtained in advance by the service data prediction model, does not rely on a single data threshold. Instead, it determines whether to perform energy-saving processing on a target cell based on both the target service data and the current service data. Therefore, it can perform energy-saving processing on the target cell more accurately, solving the problem in related technologies that rely heavily on data thresholds and cannot accurately perform energy-saving processing on cells. This can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0133] Based on the foregoing embodiments, referring to Figure 8 As shown in the figure, this application embodiment provides an energy-saving processing system, including a prediction module, a monitoring module, and a processing module. The prediction module uses a service data prediction model built on a temporal convolutional network for long-term prediction, capable of predicting service data for 96 time points in the future day. This allows for the identification of energy-saving periods for target cells, pre-filtering deactivated cell lists and wake-up cell lists for each time period. The prediction results of the service data prediction model are then sent to a remote dictionary server (Redis). The monitoring module performs corresponding operations based on the deactivated cell lists and wake-up cell lists in Redis. For the wake-up cell list, the monitoring module dispatches the request to the processing module to perform a wake-up operation, ensuring that the cells in the wake-up cell list maintain normal network status. For the deactivated cell list, the Operation and Maintenance Center (OMC) executes real-time performance query commands (such as online user count queries) via the Telnet interface to obtain the current service data of the cells in the deactivated cell list. If the current service data is greater than or equal to the second service data threshold, the target cell prediction is determined to have failed, and the target cell is removed from the deactivated cell list. If the target service data is less than the first service data threshold and the current service data is less than the second service data threshold, the processing module performs energy-saving processing on the target cell. This improves the efficiency of energy-saving processing and ensures its accuracy. The monitoring module can also set the monitoring frequency interval, such as t minutes / time, preferably t is 1, and the monitoring frequency can be k times, preferably k is 4.

[0134] like Figure 9As shown, the energy-saving processing device is configured to send the wake-up cell list and the deactivated cell list at each moment in a message to the client, so that the client can perform corresponding operations based on the wake-up cell list and the deactivated cell list in the message. The wake-up operation is performed on the cells in the wake-up cell list to keep the cells in the wake-up cell list in normal network status, and the energy-saving processing is performed on the cells in the deactivated cell list to put the cells in the wake-up cell list in an energy-saving state.

[0135] Based on the foregoing embodiments, embodiments of this application provide an energy-saving processing device, which can be applied to... Figures 1-2 In the energy-saving treatment method provided in the embodiment corresponding to 5, refer to Figure 10 As shown, the energy-saving processing device 4 includes:

[0136] The acquisition unit 41 is used to acquire the first historical service data of the target cell and the current service data of the target cell; wherein, the first historical service data can represent the historical network usage of the target cell.

[0137] Unit 42 is used to determine the business data prediction model;

[0138] The acquisition unit 41 is also used to process the first historical service data based on the service data prediction model to obtain the target service data of the target cell.

[0139] The determining unit 42 is also used to determine whether to perform energy-saving processing on the target cell based on the target service data and the current service data.

[0140] In this embodiment of the application, the determining unit 42 is further configured to obtain the second historical service data of the sample cell; wherein, the second historical service data is service data of the sample cell that can characterize the historical network usage of the sample cell;

[0141] The determination unit 42 is also used to construct an initial business data prediction model using multiple convolutional layers determined based on multiple different inflation factors;

[0142] The determining unit 42 is also used to train the initial business data prediction model based on the second historical business data to obtain the business data prediction model.

[0143] In this embodiment of the application, the determining unit 42 is further configured to determine multiple convolutional layers of the initial business data prediction model based on multiple different inflation factors; wherein, the inflation factor increases sequentially with the number of convolutional layers;

[0144] The determining unit 42 is also used to perform convolution processing on the second historical business data based on multiple convolutional layers to obtain the first business data;

[0145] The determining unit 42 is also used to determine the business data that meets the target conditions from the second historical business data, and obtain the second business data;

[0146] The determining unit 42 is also used to train the model based on the second historical business data, the first business data and the second business data to obtain a business data prediction model.

[0147] In this embodiment of the application, the determining unit 42 is further configured to perform residual processing on the first service data and the second historical service data to obtain the third service data;

[0148] The determining unit 42 is also used to perform full connection processing on the second business data and the third business data to obtain the fourth business data;

[0149] Unit 42 is also used to train a model based on the fourth business data to obtain a business data prediction model.

[0150] In this embodiment of the application, the determining unit 42 is further configured to determine the target loss function;

[0151] The determining unit 42 is also used to train the model based on the fourth business data, the third historical business data and the target loss function to obtain the business data prediction model; wherein the third historical business data is obtained after the acquisition time of the second historical business data.

[0152] In this embodiment of the application, the determining unit 42 is further configured to determine the initial loss function;

[0153] The determining unit 42 is also used to determine a first weight, a second weight, and a third weight; wherein, the first weight is the weight corresponding to when the output data of the initial business data prediction model is greater than the third historical business data and less than the first target threshold, the second weight is the weight corresponding to when the output data is greater than the second target threshold, and the third weight is the weight corresponding to when the output data is greater than or equal to the first target threshold and less than or equal to the second target threshold.

[0154] The determination unit 42 is also used to determine the target loss function based on the initial loss function, the first weight, the second weight, and the third weight.

[0155] In this embodiment of the application, the determining unit 42 is further configured to determine that the target cell enters a power-saving state when the target service data is less than the first service data threshold.

[0156] The determining unit 42 is also used to perform energy-saving processing on the target cell when the target cell enters the energy-saving state and the current service data is less than the second service data threshold.

[0157] The energy-saving processing device provided in this application embodiment determines whether to perform energy-saving processing on a target cell based on the target service data and current service data obtained in advance by the service data prediction model. This method does not rely on a single data threshold, and it determines whether to perform energy-saving processing on a target cell based on both the target service data and the current service data. Therefore, it can perform energy-saving processing on the target cell more accurately, solving the problem in related technologies that have a high dependence on data thresholds and cannot accurately perform energy-saving processing on cells. This can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0158] Based on the foregoing embodiments, embodiments of this application provide an energy-saving processing device that can be applied to... Figures 1-2 In the energy-saving treatment method provided in the embodiment corresponding to 5, refer to Figure 11 As shown, the energy-saving processing device 5 may include: a processor 51, a memory 52, and a communication bus 53, wherein:

[0159] Communication bus 53 is used to realize the communication connection between processor 51 and memory 52;

[0160] The processor 51 is used to execute the energy-saving processing program in the memory 52 to perform the following steps:

[0161] Obtain the first historical service data and the current service data of the target cell; wherein, the first historical service data is service data that can characterize the historical network usage of the target cell;

[0162] Determine the business data prediction model;

[0163] The first historical service data is processed based on the service data prediction model to obtain the target service data of the target cell.

[0164] Based on the target business data and current business data, determine whether to implement energy-saving measures for the target community.

[0165] In other embodiments of this application, processor 51 is used to execute a power-saving processing program in memory 52 to determine a business data prediction model in order to perform the following steps:

[0166] Obtain the second historical service data of the sample cell; wherein, the second historical service data is the service data of the sample cell that can characterize the historical network usage of the sample cell;

[0167] An initial business data prediction model is constructed using multiple convolutional layers determined based on multiple different inflation factors;

[0168] The initial business data prediction model is trained based on the second set of historical business data to obtain the business data prediction model.

[0169] In other embodiments of this application, the processor 51 is used to execute an energy-saving processing program in the memory 52 to train an initial service data prediction model based on second historical service data, thereby obtaining a service data prediction model to implement the following steps:

[0170] Multiple convolutional layers of the initial business data prediction model are determined based on multiple different inflation factors; wherein, the inflation factors increase sequentially with the number of convolutional layers.

[0171] The first business data is obtained by performing convolution processing on the second historical business data using multiple convolutional layers.

[0172] The second business data is obtained by identifying business data that meets the target conditions from the second historical business data.

[0173] A business data prediction model is obtained by training the model based on the second historical business data, the first business data, and the second business data.

[0174] In other embodiments of this application, the first processor 51 is used to execute an energy-saving processing program in the first memory 52 to train a model based on second historical service data, first service data, and second service data to obtain a service data prediction model, in order to implement the following steps:

[0175] The third business data is obtained by performing residual processing on the first business data and the second historical business data;

[0176] The second and third business data are processed using a full connection to obtain the fourth business data.

[0177] The business data prediction model is obtained by training the model based on the fourth business data.

[0178] In other embodiments of this application, the first processor 51 is used to execute an energy-saving processing program in the first memory 52 to train a model based on fourth service data to obtain a service data prediction model, in order to implement the following steps:

[0179] Determine the target loss function;

[0180] The model is trained based on the fourth business data, the third historical business data, and the target loss function to obtain the business data prediction model; the third historical business data is obtained after the acquisition time of the second historical business data.

[0181] In other embodiments of this application, the first processor 51 is used to execute an energy-saving processing program in the first memory 52 to train a model based on second historical service data, first service data, and second service data to obtain a service data prediction model, in order to implement the following steps:

[0182] Determine the initial loss function;

[0183] Determine the first weight, the second weight, and the third weight; wherein, the first weight is the weight corresponding to when the output data of the initial business data prediction model is greater than the third historical business data and less than the first target threshold, the second weight is the weight corresponding to when the output data is greater than the second target threshold, and the third weight is the weight corresponding to when it is greater than or equal to the first target threshold and less than or equal to the second target threshold, wherein the second target threshold is greater than the first target threshold;

[0184] The target loss function is determined based on the initial loss function, the first weight, the second weight, and the third weight.

[0185] In other embodiments of this application, the first processor 51 is used to execute an energy-saving processing program in the first memory 52 to determine whether to perform energy-saving processing on the target cell based on target service data and current service data, in order to implement the following steps:

[0186] If the target service data is less than the first service data threshold, the target cell is determined to enter the energy-saving state.

[0187] When the target cell enters a state requiring energy saving and the current business data is less than the second business data threshold, energy saving processing is performed on the target cell.

[0188] It should be noted that a detailed description of the steps performed by the processor can be found in [reference needed]. Figures 1-2 The energy-saving treatment method provided in the embodiment corresponding to 5 will not be described again here.

[0189] The energy-saving processing device provided in this application embodiment determines whether to perform energy-saving processing on a target cell based on the target service data and current service data obtained in advance by the service data prediction model. This method does not rely on a single data threshold, and it determines whether to perform energy-saving processing on a target cell based on both the target service data and the current service data. Therefore, it can perform energy-saving processing on the target cell more accurately, solving the problem in related technologies that have a high dependence on data thresholds and cannot accurately perform energy-saving processing on cells. This can improve the efficiency of energy-saving processing, ensure the accuracy of energy-saving processing, and reduce resource waste.

[0190] Based on the foregoing embodiments, embodiments of this application provide a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement... Figures 1-2 The steps of the energy-saving treatment method provided in the embodiment corresponding to 5.

[0191] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0192] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0193] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0194] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0195] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. An energy-saving treatment method, characterized in that, The method includes: Obtain first historical service data and current service data of the target cell; wherein, the first historical service data is service data that can characterize the historical network usage of the target cell; Obtain the second historical service data of the sample cell; wherein, the second historical service data is service data of the sample cell that can characterize the historical network usage of the sample cell; An initial business data prediction model is constructed using multiple convolutional layers determined based on multiple different inflation factors; The initial business data prediction model is trained based on the second historical business data to obtain the business data prediction model. The first historical service data is processed based on the service data prediction model to obtain the target service data of the target cell. Based on the target service data and the current service data, determine whether to perform energy-saving processing on the target cell.

2. The method according to claim 1, characterized in that, The step of training the initial business data prediction model based on the second historical business data to obtain the business data prediction model includes: The multiple convolutional layers of the initial business data prediction model are determined based on the multiple different inflation factors; wherein the inflation factors increase sequentially with the number of convolutional layers. The second historical business data is processed by convolution based on the multiple convolutional layers to obtain the first business data. The business data that meets the target conditions is determined from the second historical business data to obtain the second business data; The business data prediction model is obtained by training the model based on the second historical business data, the first business data, and the second business data.

3. The method according to claim 2, characterized in that, The step of training the model based on the second historical business data, the first business data, and the second business data to obtain the business data prediction model includes: The third business data is obtained by performing residual processing on the first business data and the second historical business data; The second and third service data are processed using a full connection to obtain the fourth service data. The business data prediction model is obtained by training the model based on the fourth business data.

4. The method according to claim 3, characterized in that, The step of training the model based on the fourth business data to obtain the business data prediction model includes: Determine the target loss function; The service data prediction model is obtained by training the model based on the fourth service data, the third historical service data of the sample cell, and the target loss function; wherein, the third historical service data is generated by the sample cell after the time corresponding to the second historical service data.

5. The method according to claim 4, characterized in that, The determination of the target loss function includes: Determine the initial loss function; A first weight, a second weight, and a third weight are determined; wherein, the first weight is the weight corresponding to when the output data of the initial business data prediction model is greater than the third historical business data and less than the first target threshold, the second weight is the weight corresponding to when the output data is greater than the second target threshold, and the third weight is the weight corresponding to when the output data is greater than or equal to the first target threshold and less than or equal to the second target threshold, wherein the second target threshold is greater than the first target threshold; The target loss function is determined based on the initial loss function, the first weight, the second weight, and the third weight.

6. The method according to claim 1, characterized in that, The step of determining whether to perform energy-saving processing on the target cell based on the target service data and the current service data includes: If the target service data is less than the first service data threshold, the target cell is determined to enter a power-saving state. When the target cell enters a power-saving state and the current service data is less than the second service data threshold, power-saving processing is performed on the target cell.

7. An energy-saving treatment device, characterized in that, The device includes: An acquisition unit is used to acquire first historical service data of a target cell and current service data of the target cell; wherein the first historical service data is service data that can characterize the historical network usage of the target cell; The determining unit is used to acquire the second historical service data of the sample cell; wherein, the second historical service data is service data of the sample cell that can characterize the historical network usage of the sample cell; The determining unit is also used to construct an initial business data prediction model using multiple convolutional layers determined based on multiple different inflation factors; The determining unit is further configured to train the initial business data prediction model based on the second historical business data to obtain a business data prediction model. The acquisition unit is further configured to process the first historical service data based on the service data prediction model to obtain the target service data of the target cell; The determining unit is further configured to determine, based on the target service data and the current service data, whether to perform energy-saving processing on the target cell.

8. An energy-saving treatment device, characterized in that, The device includes: a processor, a memory, and a communication bus; The communication bus is used to realize the communication connection between the processor and the memory; The processor is used to execute an energy-saving processing program in the memory to implement the steps of the energy-saving processing method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the energy-saving processing method as described in any one of claims 1 to 6.

Images