5G communication equipment energy consumption management and control system

Abstract

The invention relates to the technical field of power consumption optimization, and discloses a 5G communication equipment energy consumption management and control system, which comprises a data acquisition module, a data preprocessing module, a load prediction module, a load classification module and a power consumption optimization module, the data acquisition module is used for acquiring equipment data, user behavior data, environment data and load level data of communication equipment and representing the data through a first feature sequence; the data preprocessing module is used for preprocessing the first feature sequence to obtain a second feature sequence; the load prediction module is used for constructing a load prediction model according to the second feature sequence and outputting load level data of the equipment; the load classification module is used for determining the range of load levels of the equipment in a second preset time period G through a clustering algorithm; and the power consumption optimization module is used for generating optimal parameter combinations corresponding to the three load levels according to the improved geyser algorithm. The energy efficiency of the communication equipment is improved, and energy waste is avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of power consumption optimization, and specifically relates to a 5G communication equipment energy consumption management and control system. Background Art

[0002] With the rapid development of 5G communication technology, the application of 5G communication equipment is becoming more and more extensive, covering multiple fields from smartphones, IoT devices to industrial automation. However, the high energy consumption of 5G communication equipment has become a key bottleneck restricting its sustainable development. Compared with previous generations of communication technology, 5G networks not only need to support higher data transmission rates and lower latency, but also need to handle more complex signal processing tasks, which leads to a significant increase in device power consumption.

[0003] Most existing energy management methods rely on static configuration and preset rules, lacking adaptability and flexibility. For example, some devices adjust power consumption through preset power levels, but this method cannot respond to load changes in real time, resulting in energy waste. Another method is to save energy by shutting down certain functional modules at a scheduled time, but this affects the availability of the device and the user experience.

[0004] Therefore, there is an urgent need for a 5G communication equipment energy consumption management and control system to solve the above problems. Summary of the invention

[0005] The present invention provides a 5G communication equipment energy consumption management and control system to solve the technical problem in the related technology that it is unable to respond to load changes in real time, resulting in low energy efficiency and energy waste.

[0006] The present invention provides a 5G communication equipment energy consumption management and control system, including a data acquisition module, a data preprocessing module, a load prediction module, a load classification module and a power consumption optimization module:

[0007] A data collection module, used to collect device data, user behavior data, environmental data and load level data of the communication device within a first preset time period T according to a preset time interval t, and represent them through a first feature sequence, where T and t are custom parameters;

[0008] The first feature sequence includes N sequence units, and the nth sequence unit represents device data, user behavior data, environment data, and load level data of the communication device at the nth time point, where N=T / t, 1≤n≤N;

[0009] The device data includes: transmission power, bandwidth utilization and modulation mode; the user behavior data includes: number of users, user active frequency and data traffic; the environmental data includes: ambient humidity and ambient temperature;

[0010] A data preprocessing module, used for preprocessing the first feature sequence to obtain a second feature sequence;

[0011] A load prediction module, used to construct a load prediction model according to the second feature sequence, the output of the load prediction model is the load level data of the equipment within the second preset time period G, where G is a custom parameter;

[0012] A load classification module, used for clustering the load level data within the second preset time period by a clustering algorithm to obtain a range of load levels, where the load levels include: low load, medium load and high load;

[0013] The power consumption optimization module is used to generate optimal parameter combinations corresponding to three load levels according to the improved geyser algorithm, and each optimal parameter combination includes the transmission power, bandwidth utilization rate and modulation mode that minimize the power consumption under the current load level.

[0014] Furthermore, the bandwidth utilization rate is represented by real number coding, indicating the ratio of the current bandwidth to the total bandwidth of the device; the modulation modes include: QPSK, 16-QAM and 64-QAM, and the modulation modes are represented by one-hot coding; the user active frequency is represented by real number coding, indicating the user's operation frequency.

[0015] Furthermore, the step of preprocessing the first feature sequence includes:

[0016] Step S201, using the 3σ principle to identify outliers in the first feature sequence, and replacing the outliers with the median;

[0017] Step S202, calculate the transmission power change rate between adjacent time points, and splice the transmission power change rate to the sequence unit of the first characteristic sequence. The calculation formula is:

[0018] Step S203, calculate the user active frequency change rate between adjacent time points, and splice the user active frequency change rate to the sequence unit of the first feature sequence. The calculation formula is:

[0019] Step S204: normalize the first feature sequence using a z-score normalization method to obtain a second feature sequence.

[0020] Further, the load prediction model includes: a first hidden layer, an extraction layer, and a classifier;

[0021] The first hidden layer includes N first units, the nth first unit inputs the nth sequence unit of the second feature sequence, and outputs a first feature vector;

[0022] The extraction layer is used to extract the Nth first eigenvector output by the Nth first unit, represented by the second eigenvector;

[0023] The second feature vector is input into the classifier, and the classification space of the classifier represents the load of the equipment within the second preset time period G.

[0024] Furthermore, the first hidden layer is constructed based on the improved gated recurrent unit, and the improvements include: adding an adjustment coefficient in the update gate, adding a scaling factor in the reset gate, embedding a residual connection, and introducing a trend gate.

[0025] Further, the load level data within the second preset time period is clustered using a k-means algorithm, and the steps of using the k-means algorithm include:

[0026] Step S301, randomly initialize three cluster centers;

[0027] Step S302, calculating the Euclidean distance between each load level data and each cluster center;

[0028] Step S303, assigning each load level data to the nearest cluster center to form three clusters;

[0029] Step S304, recalculating the mean of all load level data in each cluster, and using the mean as a new cluster center;

[0030] Step S305, repeating steps S302 to S304 until the change of the cluster center is less than a preset threshold;

[0031] Step S306: The three cluster centers obtained through the above steps represent three load levels, and the range of the load level is obtained according to the median value of the cluster center.

[0032] Furthermore, the step of generating the optimal parameter combinations corresponding to the three load levels according to the improved geyser algorithm in the power consumption optimization module includes:

[0033] Step S401, randomly generating an initialization population with Q individuals meeting the constraint conditions for each load level, where Q is a custom parameter;

[0034] The individual positions of the initialized population are represented by a vector code X = {P, B, M}, where P, B and M represent the transmission power, bandwidth utilization and modulation mode respectively;

[0035] Step S402, constructing an objective function;

[0036] Step S403, obtaining the fitness value of each individual in the initialized population through objective function calculation;

[0037] Step S404, using the roulette wheel selection method to select half of the individuals with higher fitness values ​​in the population as channels, and the selection probability is proportional to the fitness value of the individual;

[0038] Step S405, calculating the Euclidean distance between each individual and other individuals, and selecting J individuals with the closest Euclidean distance as neighbors of the current individual;

[0039] Step S406, based on the individual with the largest current fitness value and the individual's neighbors, the individual's position is updated. The calculation formula for updating the individual's position is: in represents the updated position of the individual, X n，i represents the neighbor position of an individual, rand represents a random number between 0 and 1, and X c，i Indicates the individual position with the largest current fitness value, X i Indicates the current individual's position;

[0040] Step S407, calculating the pressure on the individual and the probability of selecting a new channel, and updating the position of the individual when selecting a new channel; the calculation formula for the pressure on the individual is: Where P i Represents individual X i The pressure, k represents the current iteration number, f min and f max represents the minimum and maximum fitness values ​​in the current population, f(X i ) represents individual X i The fitness value of the individual is: in represents the probability of an individual selecting a new channel; the formula for updating the individual position when selecting a new channel is: in represents the updated position of the individual when a new channel is selected, represents the individual with the largest current fitness value, cauchy(0,1) represents a random number drawn from a Cauchy distribution with a location parameter of 0 and a scale parameter of 1;

[0041] Step S408, the current number of iterations is increased by 1, and if it is determined that the current number of iterations reaches the maximum number of iterations, the current individual position code is output as the optimal parameter combination corresponding to the load level, otherwise, the process returns to step S403, where the maximum number of iterations is a custom parameter.

[0042] Furthermore, the objective function in step S402 is:

[0043]

[0044] Where f represents the fitness value of the individual, b f Indicates a custom parameter.

[0045] The beneficial effects of the present invention are as follows: the present invention uses the K-means clustering algorithm to automatically divide the load levels, ensuring that the system can be flexibly adjusted according to the actual load conditions, improving the adaptability and robustness of the system, avoiding deviations caused by personal experience and preferences, and making the results more objective and consistent;

[0046] The present invention generates the optimal parameter combination through the improved geyser algorithm, realizes intelligent energy consumption management, optimizes the working state of the equipment under different load levels, and improves the overall performance;

[0047] The present invention can respond to load changes in real time by dynamically adjusting parameters such as the transmission power, bandwidth utilization rate and modulation mode of 5G communication equipment, thereby significantly improving energy efficiency and avoiding energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a module schematic diagram of a 5G communication equipment energy consumption control system of the present invention;

[0049] Figure 2 is a flow chart of preprocessing the first feature sequence of the present invention;

[0050] Figure 3 is a flow chart of the use of the k-means algorithm of the present invention;

[0051] Figure 4 It is a flow chart of the use of the improved geyser algorithm of the present invention.

[0052] In the figure: a data acquisition module 101, a data preprocessing module 102, a load prediction module 103, a load classification module 104 and a power consumption optimization module 105. DETAILED DESCRIPTION

[0053] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0054] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0055] like Figure 1-Figure 4 As shown, a 5G communication equipment energy consumption management and control system includes a data acquisition module, a data preprocessing module, a load prediction module, a load classification module and a power consumption optimization module:

[0056] The data collection module 101 is used to collect device data, user behavior data, environmental data and load level data of the communication device within a first preset time period T according to a preset time interval t, and represent them through a first feature sequence, where T and t are custom parameters;

[0057] The first feature sequence includes N sequence units, and the nth sequence unit represents device data, user behavior data, environment data, and load level data of the communication device at the nth time point, where N=T / t, 1≤n≤N;

[0058] The device data includes: transmission power, bandwidth utilization and modulation mode; the user behavior data includes: number of users, user active frequency and data traffic; the environmental data includes: ambient humidity and ambient temperature;

[0059] A data preprocessing module 102 is used to preprocess the first feature sequence to obtain a second feature sequence;

[0060] The load prediction module 103 is used to construct a load prediction model according to the second feature sequence, and the output of the load prediction model is the load level data of the equipment within the second preset time period G, where G is a custom parameter;

[0061] The load classification module 104 is used to cluster the load level data within the second preset time period by a clustering algorithm to obtain a range of load levels, where the load levels include: low load, medium load and high load;

[0062] The power consumption optimization module 105 is used to generate optimal parameter combinations corresponding to three load levels according to the improved geyser algorithm, each optimal parameter combination including the transmission power, bandwidth utilization rate and modulation mode that minimize the power consumption under the current load level.

[0063] In one embodiment of the present invention, the bandwidth utilization rate is represented by real number coding, indicating the ratio of the current bandwidth to the total bandwidth of the device; the modulation modes include: QPSK, 16-QAM and 64-QAM, and the modulation mode is represented by one-hot coding, for example, when the modulation mode is QPSK, it is represented by [1, 0, 0]; the user activity frequency is represented by real number coding, indicating the user's operation frequency, such as the number of visits per minute.

[0064] In one embodiment of the present invention, the step of preprocessing the first feature sequence includes:

[0065] Step S201, using the 3σ principle to identify outliers in the first feature sequence, and replacing the outliers with the median;

[0066] Step S202, calculate the transmission power change rate between adjacent time points, and splice the transmission power change rate to the sequence unit of the first characteristic sequence. The calculation formula is:

[0067] Step S203, calculate the user active frequency change rate between adjacent time points, and splice the user active frequency change rate to the sequence unit of the first feature sequence. The calculation formula is:

[0068] Step S204: normalize the first feature sequence using a z-score normalization method to obtain a second feature sequence.

[0069] In one embodiment of the present invention, the load prediction model includes: a first hidden layer, an extraction layer, and a classifier;

[0070] The first hidden layer includes N first units, the nth first unit inputs the nth sequence unit of the second feature sequence, and outputs a first feature vector;

[0071] The extraction layer is used to extract the Nth first eigenvector output by the Nth first unit, represented by the second eigenvector;

[0072] The second feature vector is input into the classifier, and the classification space of the classifier represents the load of the equipment within the second preset time period G.

[0073] In one embodiment of the present invention, the first hidden layer is constructed based on the improved gated recurrent unit, and the specific improvements include: adding an adjustment coefficient in the update gate, adding a scaling factor in the reset gate, embedding a residual connection, and introducing a trend gate. In these improvements, the adjustment coefficient is used to flexibly adjust the influence of the hidden state at the previous moment, so that the model can more easily balance the proportion of historical information and current input, thereby improving the accuracy and adaptability of the prediction; adding a scaling factor in the reset gate to reduce the interference of short-term fluctuations on the model, helping the model to more accurately retain key long-term trend information, avoiding overreaction caused by short-term noise, and making the model more stable in dynamic changes; through residual connections, the original input information can be directly transmitted to subsequent levels to prevent the loss of key information when processing short-term fluctuations, and improve the model's expressiveness and anti-interference ability in complex time series; the trend gate specifically captures and retains the long-term trend information of the input data, so that the model has the ability to learn the data change pattern over a long period of time. Through this gated design, the model can effectively focus on long-term trends while processing short-term fluctuations, significantly improving the overall prediction effect.

[0074] Specifically, the calculation formula of the nth first unit of the first hidden layer includes:

[0075] The calculation formula of the update gate of the first unit of the first hidden layer is:

[0076]

[0077] Among them, z n represents the update gate of the first unit of the first hidden layer, x n represents the nth sequence unit of the second feature sequence currently input, h n-1 represents the first eigenvector output by the n-1th first unit, and They represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the update gate of the nth sequence unit of the second feature sequence, respectively. θ represents the adjustment coefficient, which is used to adjust the hidden state h at the previous moment. n-1 The weight in the update gate increases the sensitivity of the update gate to short-term changes. sigmoid represents the sigmoid activation function.

[0078] The calculation formula of the reset gate of the first unit of the first hidden layer is:

[0079]

[0080] Among them, r n represents the reset gate of the nth first unit in the first hidden layer, and denote the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the nth sequence unit of the second feature sequence respectively; δ denotes a scaling factor, which is used to adjust the response speed of the reset gate to the current input and hidden state, prevent overreaction to short-term noise, and retain long-term information;

[0081] The calculation formula of the trend gate of the nth first unit in the first hidden layer is:

[0082]

[0083] Among them, τ n represents the trend gate of the nth first unit in the first hidden layer, and Respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the trend gate of the nth sequence unit of the second feature sequence;

[0084] The calculation formula for the candidate hidden state of the nth first unit in the first hidden layer is:

[0085]

[0086] in, represents the candidate hidden state of the nth first unit of the first hidden layer, and Respectively represent a first weight parameter, a second weight parameter and a bias parameter corresponding to the candidate hidden state of the nth sequence unit of the second feature sequence;

[0087] The calculation formula for the hidden state update of the nth first unit in the first hidden layer is:

[0088]

[0089] Among them, h n represents the first eigenvector of the output of the nth first unit of the first hidden layer, ⊙ represents point-by-point multiplication, and γ represents the weight coefficient of the residual connection, which controls the direct impact of the current input on the hidden state and prevents excessive smoothing of information.

[0090] In one embodiment of the present invention, a k-means algorithm is used to cluster the load level data within the second preset time period, and the steps of using the k-means algorithm include:

[0091] Step S301, randomly initialize three cluster centers;

[0092] Step S302, calculating the Euclidean distance between each load level data and each cluster center, the formula of the Euclidean distance is: Where D(P i , Pj ) represents the data point P i (x i ,y i ) and the cluster center point P j (x j ,y j )’s Euclidean distance;

[0093] Step S303, assigning each load level data to the nearest cluster center to form three clusters;

[0094] Step S304, recalculating the mean of all load level data in each cluster, and using the mean as a new cluster center;

[0095] Step S305, repeating steps S302 to S304 until the change of the cluster center is less than a preset threshold;

[0096] Step S306: The three cluster centers obtained through the above steps represent three load levels, and the range of load levels is obtained according to the middle value of the cluster center. For example, the low-load cluster center C low The cluster center C is 0.2, which is a medium load. med is 0.5, the cluster center C with high load high is 0.8, the low load range is 0 to That is, 0 to 0.35, and the medium load range is arrive That is 0.35 to 0.65, and the high load range is to 1, that is, 0.65 to 1.

[0097] Automatic clustering of load levels using the k-means algorithm can better reflect the actual load situation, reduce the impact of human factors, avoid deviations caused by personal experience and preferences, and make the results more objective and consistent.

[0098] In one embodiment of the present invention, the step of generating the optimal parameter combinations corresponding to the three load levels according to the improved geyser algorithm in the power consumption optimization module includes:

[0099] Step S401, randomly generating an initialization population with Q individuals meeting the constraint conditions for each load level, where Q is a custom parameter;

[0100] The constraints of the low load level population include: the transmission power ranges from 0 to 100; the bandwidth utilization rate ranges from 0 to 0.2; the modulation mode ranges from 0 to 2 and is an integer;

[0101] The constraints of the medium load level population include: the transmission power ranges from 0 to 100; the bandwidth utilization rate ranges from 0.2 to 0.4; the modulation mode ranges from 0 to 2 and is an integer;

[0102] The constraints of the high load level population include: the transmission power ranges from 0 to 100; the bandwidth utilization rate ranges from 0.4 to 1; the modulation mode ranges from 0 to 2 and is an integer;

[0103] The individual positions of the initialized population are represented by a vector code X = {P, B, M}, where P, B and M represent the transmission power, bandwidth utilization and modulation mode respectively;

[0104] Step S402, constructing an objective function;

[0105] Step S403, obtaining the fitness value of each individual in the initialized population through objective function calculation;

[0106] Step S404, using the roulette wheel selection method to select half of the individuals with higher fitness values ​​in the population as channels, the selection probability is proportional to the fitness value of the individual, and the formula for the selection probability is: where p i represents the probability of selecting the current channel, f(X i ) and f(X j ) represents individual X i and individual X j The fitness value of

[0107] Step S405, calculate the Euclidean distance between each individual and other individuals, and select the J individuals with the closest Euclidean distance as the neighbors of the current individual. The neighbors of these individuals are usually individuals that are close to the current individual in the search space. The purpose of selecting neighbors is to use local information to guide the search direction of the individual. The distance formula is: Where d(X i , X j ) represents individual X i (P i , B i , M i ) and other individuals X j (P j , B j , M j ), preferably, J is set to 5;

[0108] Step S406, based on the individual with the largest current fitness value and the individual's neighbors, the individual's position is updated. The purpose of the position update is to guide the individual to move closer to a better solution. The calculation formula for the individual position update is: in represents the updated position of the individual, X n，i represents the neighbor position of an individual, rand represents a random number between 0 and 1, and X c，i Indicates the individual position with the largest current fitness value, X i Indicates the current individual's position;

[0109] Step S407, calculating the pressure on the individual and the probability of selecting a new channel, and updating the position of the individual when selecting a new channel; the calculation formula for the pressure on the individual is: Where P i Represents individual X i The pressure, k represents the current iteration number, f min and f max Represents the minimum and maximum fitness values ​​in the current population; the formula for the probability of an individual selecting a new channel is: in represents the probability of an individual selecting a new channel; the formula for updating the individual position when selecting a new channel is: in represents the updated position of the individual when a new channel is selected, represents the individual with the largest current fitness value, cauchy(0,1) represents a random number drawn from a Cauchy distribution with a location parameter of 0 and a scale parameter of 1;

[0110] The pressure of individuals guides them to move to specific channels. The purpose of selecting new channels is to increase the diversity of search and prevent the algorithm from converging to the local optimal solution too early. Integrating the Cauchy distribution into the individual position update can effectively utilize the long-tail characteristics of the Cauchy distribution, increase the global search ability of the algorithm, and jump out of the local optimal solution.

[0111] Step S408, the current number of iterations is increased by 1, and if it is determined that the current number of iterations reaches the maximum number of iterations, the current individual position code is output as the optimal parameter combination corresponding to the load level, otherwise, it returns to step S403, where the maximum number of iterations is a custom parameter, preferably, the maximum number of iterations is set to 200.

[0112] In one embodiment of the present invention, the objective function in step S402 is:

[0113]

[0114] Where f represents the fitness value of the individual, b f Indicates a custom parameter.

[0115] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A 5G communication equipment energy consumption management and control system, characterized in that: Including data acquisition module, data preprocessing module, load prediction module, load classification module and power consumption optimization module: A data collection module, used to collect device data, user behavior data, environmental data and load level data of the communication device within a first preset time period T according to a preset time interval t, and represent them through a first feature sequence, where T and t are custom parameters; The first feature sequence includes N sequence units, and the nth sequence unit represents device data, user behavior data, environment data, and load level data of the communication device at the nth time point, where N=T / t, 1≤n≤N; The device data includes: transmission power, bandwidth utilization and modulation mode; the user behavior data includes: number of users, user active frequency and data traffic; the environmental data includes: ambient humidity and ambient temperature; A data preprocessing module, used for preprocessing the first feature sequence to obtain a second feature sequence; A load prediction module, used to construct a load prediction model according to the second feature sequence, the output of the load prediction model is the load level data of the equipment within the second preset time period G, where G is a custom parameter; A load classification module, used for clustering the load level data within the second preset time period by a clustering algorithm to obtain a range of load levels, where the load levels include: low load, medium load and high load; The power consumption optimization module is used to generate optimal parameter combinations corresponding to three load levels according to the improved geyser algorithm, and each optimal parameter combination includes the transmission power, bandwidth utilization rate and modulation mode that minimize the power consumption under the current load level.

2. A 5G communication equipment energy consumption management and control system according to claim 1, characterized in that: The bandwidth utilization rate is represented by real number coding, which indicates the ratio of the current bandwidth to the total bandwidth of the device. The modulation modes include QPSK, 16-QAM and 64-QAM, which are represented by one-hot coding. The user active frequency is represented by real number coding, which indicates the user's operation frequency.

3. A 5G communication equipment energy consumption management and control system according to claim 1, characterized in that: The step of preprocessing the first feature sequence includes: Step S201, using the 3σ principle to identify outliers in the first feature sequence, and replacing the outliers with the median; Step S202, calculate the transmission power change rate between adjacent time points, and splice the transmission power change rate to the sequence unit of the first characteristic sequence. The calculation formula is: Step S203, calculate the user active frequency change rate between adjacent time points, and splice the user active frequency change rate to the sequence unit of the first feature sequence. The calculation formula is: Step S204: normalize the first feature sequence using a z-score normalization method to obtain a second feature sequence.

4. A 5G communication equipment energy consumption management and control system according to claim 1, characterized in that: The load prediction model includes: a first hidden layer, an extraction layer and a classifier; The first hidden layer includes N first units, the nth first unit inputs the nth sequence unit of the second feature sequence, and outputs a first feature vector; The extraction layer is used to extract the Nth first eigenvector output by the Nth first unit, represented by the second eigenvector; The second feature vector is input into the classifier, and the classification space of the classifier represents the load of the equipment within the second preset time period G.

5. A 5G communication equipment energy consumption management and control system according to claim 4, characterized in that: The first hidden layer is built based on the improved gated recurrent unit. The improvements include: adding an adjustment coefficient in the update gate, adding a scaling factor in the reset gate, embedding a residual connection, and introducing a trend gate.

6. A 5G communication equipment energy consumption management and control system according to claim 1, characterized in that: The load level data within the second preset time period is clustered using a k-means algorithm, and the steps of using the k-means algorithm include: Step S301, randomly initialize three cluster centers; Step S302, calculating the Euclidean distance between each load level data and each cluster center; Step S303, assigning each load level data to the nearest cluster center to form three clusters; Step S304, recalculating the mean of all load level data in each cluster, and using the mean as a new cluster center; Step S305, repeating steps S302 to S304 until the change of the cluster center is less than a preset threshold; Step S306: The three cluster centers obtained through the above steps represent three load levels, and the range of the load level is obtained according to the median value of the cluster center.

7. A 5G communication equipment energy consumption management and control system according to claim 1, characterized in that: The steps of generating the optimal parameter combinations corresponding to the three load levels according to the improved geyser algorithm in the power consumption optimization module include: Step S401, randomly generating an initialization population with Q individuals meeting the constraint conditions for each load level, where Q is a custom parameter; The individual positions of the initialized population are represented by a vector code X = {P, B, M}, where P, B and M represent the transmission power, bandwidth utilization and modulation mode respectively; Step S402, constructing an objective function; Step S403, obtaining the fitness value of each individual in the initialized population through objective function calculation; Step S404, using the roulette wheel selection method to select half of the individuals with higher fitness values ​​in the population as channels, and the selection probability is proportional to the fitness value of the individual; Step S405, calculating the Euclidean distance between each individual and other individuals, and selecting J individuals with the closest Euclidean distance as neighbors of the current individual; Step S406, based on the individual with the largest current fitness value and the individual's neighbors, the individual's position is updated. The calculation formula for updating the individual's position is: in represents the updated position of the individual, X n，i represents the neighbor position of an individual, rand represents a random number between 0 and 1, and X c，i Indicates the individual position with the largest current fitness value, X i Indicates the current individual's position; Step S407, calculating the pressure on the individual and the probability of selecting a new channel, and updating the position of the individual when selecting a new channel; the calculation formula for the pressure on the individual is: Where P i Represents individual X i The pressure, k represents the current iteration number, f min and f max represents the minimum and maximum fitness values ​​in the current population, f(X i ) represents individual X i The fitness value of the individual is: in represents the probability of an individual selecting a new channel; the formula for updating the individual position when selecting a new channel is: in represents the updated position of the individual when a new channel is selected, represents the individual with the largest current fitness value, cauchy(0,1) represents a random number drawn from a Cauchy distribution with a location parameter of 0 and a scale parameter of 1; Step S408, the current number of iterations is increased by 1, and if it is determined that the current number of iterations reaches the maximum number of iterations, the current individual position code is output as the optimal parameter combination corresponding to the load level, otherwise, the process returns to step S403, where the maximum number of iterations is a custom parameter.

8. A 5G communication equipment energy consumption management and control system according to claim 7, characterized in that: The objective function in step S402 is: Where f represents the fitness value of the individual, b f Indicates a custom parameter.

Images