Intelligent monitoring method, device, equipment and storage medium for computer room power

Abstract

The present invention relates to the field of intelligent monitoring technology, and discloses an intelligent monitoring method, device, equipment and storage medium for computer room power. The method collects the total current and total voltage values of each computer room device in multiple historical periods, as well as the current value of each electric energy unit, constructs training and test sets after normalization processing, and calculates actual electric energy data. Based on the change in the total current in adjacent historical periods, a timing weight is established for each device to form timing weight data. At the same time, the timing coefficient of the device is calculated using the unit current value and the total current value to obtain the timing coefficient data. Combining the timing coefficient and weight data, the impact value of each device and its probability distribution are calculated to obtain the total timing impact probability. Based on the total timing impact probability, data time series and actual electric energy data, a neural network model is trained to ultimately achieve accurate prediction of the electric energy consumption of computer room equipment. This method achieves accurate prediction of the electric energy consumed by computer room equipment.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to an intelligent monitoring method, device, equipment and storage medium for computer room power. Background Art

[0002] With the rapid development of information technology, the stability and security of computer rooms, as core facilities for data processing and storage, have become the lifeline of enterprise operations. Therefore, computer room power and environmental monitoring systems have become an indispensable component of modern computer room management. They not only comprehensively monitor the power system and computer room environment, but also provide management and control based on the monitoring data. By continuously monitoring key parameters such as voltage and current in the power system, the system ensures stable equipment operation. Furthermore, through intelligent regulation and control, it improves the efficiency and safety of the power system. Regarding the computer room environment, the system focuses on monitoring parameters such as temperature, humidity, and air quality to ensure that all equipment operates in an optimal environment, thereby improving equipment reliability and extending its service life. For routine computer room maintenance and management, the power and environmental monitoring system supports both remote and local monitoring modes. The former allows remote access to monitoring equipment in the computer room via the internet or a dedicated network, while the latter allows on-site monitoring and management via local equipment. Both monitoring methods rely on real-time monitoring data collected by various sensors and monitoring devices installed within the computer room. The type and number of sensors will vary depending on the specific size and complexity of the computer room, but together they form the foundation of the computer room power environment monitoring system, providing solid technical support for efficient and safe computer room management.

[0003] Currently, most data centers use energy consumption monitoring systems based on sensor networks to collect and analyze power data. These systems typically consist of smart meters and sensors installed in key locations, such as distribution cabinets and server racks. They collect real-time data on parameters such as voltage, current, and power factor, and transmit this data to a central management system via wired or wireless channels. Smart meters and sensors not only provide accurate power data but also monitor the operating status of equipment, promptly identifying potential faults and anomalies. Furthermore, data centers utilize cloud computing and big data technologies to conduct in-depth analysis of collected data to optimize energy usage strategies. Through cloud computing platforms, data centers can centrally manage and analyze data, enabling a better understanding and prediction of energy usage and the development of more effective energy-saving measures.

[0004] While existing power data monitoring technologies have achieved some success, they still face significant drawbacks in practical applications. For example, traditional monitoring systems lack accuracy and real-time performance in data prediction. This is particularly true when faced with the complex and ever-changing workloads of large-scale data centers. Existing solutions often struggle to accurately predict the power consumption of equipment in computer rooms. Summary of the Invention

[0005] The present invention provides an intelligent monitoring method, device, equipment and storage medium for computer room power, so as to achieve accurate prediction of the electric energy consumed by equipment in the computer room.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides an intelligent monitoring method for power in a computer room, comprising:

[0007] Collect the total current value and corresponding total voltage value of each equipment in the computer room within several historical periods, as well as the unit current value of each power unit;

[0008] Normalizing the total current value and the total voltage value to construct a training set and a test set;

[0009] Calculating according to the total current value and the total voltage value to obtain actual electric energy data;

[0010] Constructing a time series weight for each device in the computer room according to the increment of the total current value in adjacent historical periods to obtain time series weight data;

[0011] Calculate the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value to obtain timing coefficient data;

[0012] Constructing an impact value of each equipment in the computer room according to the timing coefficient data and the timing weight data, and calculating a probability distribution of the impact value to obtain a total timing impact probability;

[0013] Using the training set and the test set as training data, training a neural network model according to the total time series impact probability, the time sequence of the collected data, and the actual electric energy data, to obtain a trained neural network model;

[0014] The output of the neural network model is determined as electric energy prediction data, wherein the electric energy prediction data is used to predict the percentage of electric energy consumed by each equipment in the computer room at a future moment.

[0015] In an optional implementation, constructing the time series weight of each equipment in the computer room according to the increment of the total current value in adjacent historical periods to obtain the time series weight data includes:

[0016] For the The following formula is used to calculate the equipment in each historical period. Total current increment within:

[0017]

[0018] in, Indicates the The equipment in the computer room is The total current value increment of the historical cycle, Indicates the The equipment in the computer room is The total current value of the historical cycle, Indicates the The equipment in the computer room is Total current value of historical cycles;

[0019] Collect equipment from each computer room In all historical periods The total current increments within form a total current increment set;

[0020] The total current increment set is used as an input variable and input into the following function to select the median as the timing weight:

[0021]

[0022] in, Indicates the The timing weight of each equipment in the computer room, Represents a preset computer program algorithm, Respectively represent 1st to 3rd of the equipment room The total current increment per cycle, Indicates the total number of historical cycles;

[0023] The timing weights of all equipment in the computer room are collected to form a timing weight set to obtain timing weight data.

[0024] In an optional embodiment, calculating the timing coefficient of the corresponding equipment in the computer room according to the total current value and the unit current value of the corresponding electric energy unit to obtain the timing coefficient data includes:

[0025] According to The total current value of the equipment in the computer room and the corresponding The unit current value of the electric energy unit is calculated by the following formula Among the equipment in the computer room The current ratio of each energy unit:

[0026]

[0027] in, Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The total current value of the equipment in the computer room, No. The first The unit current value of each electric energy unit, Represents historical cycles;

[0028] According to the current ratio, the following formula is used to calculate the current ratio: Timing coefficient of each equipment in the computer room:

[0029]

[0030]

[0031] in, Indicates the The timing coefficient of each equipment in the computer room, Indicates the The average current ratio of the equipment in each computer room is: Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The number of power units contained in each equipment room, Represents historical cycles;

[0032] The timing coefficients of each equipment in the computer room are collected to form a timing coefficient set, and timing coefficient data is obtained.

[0033] In an optional embodiment, constructing the influence value of each computer room device according to the timing coefficient data and the timing weight data and calculating the probability distribution of the influence value to obtain the total timing influence probability includes:

[0034] Accumulate and sum the timing coefficient data and the corresponding timing weight data of all equipment in the computer room to obtain the impact value data of each equipment in the computer room;

[0035] Taking the impact value data of each equipment in the computer room in several historical periods as discrete data, performing statistical classification on the discrete data, and obtaining probability distribution data of the impact values of several training levels;

[0036] Calculating the probability of the impact value of each training level according to the probability distribution data to obtain a plurality of level probability data;

[0037] Taking each training level as an index, several level probability data are accumulated and summed to obtain the total time series impact probability.

[0038] In an optional embodiment, the training process of the neural network model includes:

[0039] Inputting the training data into the input layer of the neural network model in chronological order for training, and obtaining predicted value data output by the output layer of the neural network model;

[0040] Substituting the predicted value data and the actual electric energy data into a loss function to calculate a loss value to obtain loss value data;

[0041] Calculating the gradient of the output layer output of the neural network model based on the loss value data, and passing the gradient forward layer by layer through the chain rule to calculate the gradient of the parameters of each layer to obtain gradient data;

[0042] Update the parameters of each layer of the neural network model based on the gradient data and the preset learning rate;

[0043] The parameters of each layer are updated repeatedly until the training is completed when the number of training times of the neural network model is greater than a preset number, or when the loss function value of the neural network model is less than a preset loss threshold.

[0044] In an optional embodiment, the calculating the loss value data using the predicted value and the actual electric energy data to obtain the loss value data includes:

[0045] The loss value data is calculated by the following loss function:

[0046]

[0047] in, Represents loss value data, Represents the power forecast data, Indicates the actual data of electric energy. Indicates the number of computer rooms.

[0048] In an optional embodiment, determining the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment, includes:

[0049] The power forecast data is determined by the following formula:

[0050]

[0051] in, Indicates the The equipment in the computer room is The percentage of electric energy in a historical period, is the output function of the neural network, Indicates the The equipment in the computer room is The normalized total current value of the historical cycle, Indicates the The equipment in the computer room is Normalized total voltage value of the historical period.

[0052] In a second aspect, the present invention provides an intelligent monitoring device for power in a computer room, comprising:

[0053] The data acquisition module is used to collect the total current value and the corresponding total voltage value of each equipment in the computer room within several historical periods, as well as the unit current value of each power unit;

[0054] A data preprocessing module, configured to normalize the total current value and the total voltage value to construct a training set and a test set;

[0055] an actual electric energy calculation module, configured to calculate according to the total current value and the total voltage value to obtain actual electric energy data;

[0056] A time series weight construction module is used to construct the time series weight of each equipment in the computer room based on the increment of the total current value in adjacent historical periods to obtain time series weight data;

[0057] A timing coefficient calculation module, configured to calculate the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value, and obtain timing coefficient data;

[0058] An influence value calculation module is used to construct an influence value of each equipment in the computer room using the time series coefficient data and the time series weight data, and calculate the probability distribution of the influence value to obtain the total time series influence probability;

[0059] A neural network training module is used to use the training set and the test set as training data, train the neural network model according to the total time series impact probability, the time sequence of the collected data and the real-time electric energy actual data, and obtain a trained neural network model;

[0060] An output module is configured to determine the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each computer room device at a future moment. In a third aspect, the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements any of the aforementioned methods for intelligently monitoring computer room power.

[0061] In a fourth aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned intelligent monitoring methods for computer room power.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The present invention discloses an intelligent monitoring method for computer room power, comprising collecting the total current value and the corresponding total voltage value of each computer room device within a plurality of historical periods, and collecting the unit current value of each electric energy unit; normalizing the total current value and the total voltage value to construct a training set and a test set; performing calculations based on the total current value and the total voltage value to obtain actual electric energy data; constructing a time series weight for each computer room device based on increments of the total current values of adjacent historical periods to obtain time series weight data; calculating a time series coefficient for the corresponding computer room device based on the unit current value and the total current value to obtain time series coefficient data; constructing an influence value for each computer room device based on the time series coefficient data and the time series weight data, and calculating a probability distribution of the influence value to obtain a time series total influence probability; using the training set and the test set as training data, training a neural network model based on the time series total influence probability, the time sequence of the collected data, and the actual electric energy data to obtain a trained neural network model; and determining the output of the neural network model as electric energy prediction data, wherein the electric energy prediction data is used to predict the percentage of electric energy consumed by each computer room device at a future moment.

[0064] The present invention collects and processes historical current and voltage data of equipment in a computer room, constructs training and test sets, calculates actual power usage and timing weights and coefficients, constructs a probability distribution of influence values, and trains a neural network model to predict the future power consumption percentage of each device, thereby achieving accurate prediction of the power consumption of equipment in a computer room. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 This is a schematic diagram of a process for providing intelligent monitoring of power in a computer room according to a first embodiment of the present invention;

[0066] Figure 2 It is a structural diagram of intelligent monitoring of computer room power provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] Reference Figure 1 The first embodiment of the present invention provides an intelligent monitoring method for a computer room power system, comprising the following steps:

[0069] S11, collecting the total current value and the corresponding total voltage value of each equipment in the computer room within several historical periods, and collecting the unit current value of each power unit;

[0070] S12, normalizing the total current value and the total voltage value to construct a training set and a test set;

[0071] S13, performing calculation based on the total current value and the total voltage value to obtain actual electric energy data;

[0072] S14, constructing a time series weight for each device in the computer room based on the increment of the total current value in adjacent historical periods to obtain time series weight data;

[0073] S15, calculating the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value to obtain timing coefficient data;

[0074] S16, constructing an influence value of each equipment in the computer room according to the time series coefficient data and the time series weight data, and calculating a probability distribution of the influence value to obtain a total time series influence probability;

[0075] S17, using the training set and the test set as training data, and training a neural network model according to the total time series impact probability, the time sequence of the collected data, and the actual electric energy data, to obtain a trained neural network model;

[0076] S18, determining the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment.

[0077] In step S11 , the total current value and the corresponding total voltage value of each equipment in the computer room in several historical periods are collected, and the unit current value of each electric energy unit is collected.

[0078] Smart meters and sensors deployed in the computer room collect the total current and corresponding total voltage values of each computer room device over several historical periods, as well as the unit current value of each power unit. These smart meters and sensors can monitor and record the power consumption of each computer room device and its internal power units in real time. The monitored data is sent to the cloud platform or server for storage via wireless transmission technology. The collected data not only reflects the overall power consumption pattern of the equipment, but also reveals the specific working status of each internal power unit.

[0079] In step S12, the total current value and the total voltage value are normalized to construct a training set and a test set.

[0080] In a specific embodiment, normalizing the total current value and the total voltage value to construct a training set and a test set includes:

[0081] Normalization is achieved by applying the Min-Max normalization method, that is, for each eigenvalue, its normalized value is calculated by the following formula:

[0082]

[0083] in, Indicates the normalized total current value or total voltage value, Indicates the total current value or total voltage value Indicates the minimum value of the total voltage or total current, Indicates the minimum value of the total voltage or total current.

[0084] Normalization ensures that all eigenvalues are scaled to the range [0, 1], thereby eliminating the impact of dimensions and magnitudes between different features and improving the efficiency and accuracy of model training.

[0085] After normalization, the entire dataset is randomly divided into a training set and a test set, with a ratio of 70% (training set) and 30% (test set), ensuring that the data distribution in the two sets is as consistent as possible so that the training set can fully train the model, while the test set is used to evaluate the model's generalization ability and prediction performance.

[0086] In step S13, calculation is performed based on the total current value and the total voltage value to obtain actual electric energy data.

[0087] In a specific embodiment, the calculating according to the total current value and the total voltage value to obtain the actual electric energy data includes:

[0088] The total power of the equipment in the computer room can be calculated using the following power formula:

[0089]

[0090] in, Indicates the total power of the equipment in the computer room. Indicates the total voltage value, Indicates the total current value;

[0091] The actual energy consumption percentage is calculated using the following formula to obtain the actual energy data:

[0092]

[0093] in, Indicates the percentage of actual energy consumed. Indicates the total power of the equipment in the computer room. Indicates the maximum rated power of the equipment in the equipment room.

[0094] In step S14, the time sequence weight of each equipment in the computer room is constructed according to the increment of the total current value in adjacent historical periods to obtain time sequence weight data.

[0095] In a specific embodiment, constructing the time series weight of each equipment in the computer room according to the increment of the total current value in adjacent historical cycles to obtain the time series weight data includes:

[0096] For the The following formula is used to calculate the equipment in each historical period. Total current increment within:

[0097]

[0098] in, Indicates the The equipment in the computer room is The total current value increment of the historical cycle, Indicates the The equipment in the computer room is The total current value of the historical cycle, Indicates the The equipment in the computer room is Total current value of historical cycles;

[0099] Collect equipment from each computer room In all historical periods The total current increments within form a total current increment set;

[0100] The total current increment set is used as an input variable and input into the following function to select the median as the timing weight:

[0101]

[0102] in, Indicates the The timing weight of each equipment in the computer room, Represents a preset computer program algorithm, Respectively represent 1st to 3rd of the equipment room The total current increment per cycle, Indicates the total number of historical cycles;

[0103] The timing weights of all equipment in the computer room are collected to form a timing weight set to obtain timing weight data.

[0104] Specifically, to construct the time series weight of each equipment in the computer room, we first need to measure the change in the equipment's power consumption at different time points based on the total current value increment of adjacent historical cycles. The equipment in the computer room can be Calculate its historical period The total current increment is achieved within Indicates the The equipment in the computer room is The total current value increment of the historical cycle, Indicates the The equipment in the computer room is The total current value of the historical cycle, Indicates the The equipment in the computer room is Total current value of historical cycles;

[0105] For example, if the total current value of a certain equipment in the first historical cycle is 10A and the total current value in the second historical cycle is 12A, then the total current increment of the equipment between the two cycles is Similarly, the total current increment of the device in all historical cycles can be calculated to form a total current increment set.

[0106] Next, these total current increment sets are used as input variables, and the timing weight of each equipment room is determined by selecting the median. Equipment in the computer room, its timing weight You can use the function To obtain, Refers to a preset computer program algorithm for finding the median from a given set of values. Respectively represent 1st to 3rd of the equipment room The total current increment per cycle, Indicates the total number of historical cycles

[0107] Continuing with the above example, if the total current increments of the equipment in the computer room in all historical cycles are 2A, 3A, 1A, and 4A respectively, then the timing weight of the equipment in the computer room is the median of these increment values, that is, .

[0108] Finally, after calculating the respective timing weights for all computer room equipment through the above process, these weights can be collected to form a timing weight set. This set constitutes the timing weight data, which can reflect the changing characteristics of the power consumption of each computer room equipment at different time points, and plays an important role in the subsequent power consumption prediction.

[0109] In step S15, the timing coefficient of the corresponding equipment in the computer room is calculated according to the unit current value and the total current value to obtain timing coefficient data.

[0110] In a specific embodiment, the calculating the timing coefficient of the corresponding equipment in the computer room according to the total current value and the unit current value of the corresponding electric energy unit to obtain the timing coefficient data includes:

[0111] According to The total current value of the equipment in the computer room and the corresponding The unit current value of the electric energy unit is calculated by the following formula Among the equipment in the computer room The current ratio of each energy unit:

[0112]

[0113] in, Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The total current value of the equipment in the computer room, No. The first The unit current value of each electric energy unit, Represents historical cycles;

[0114] According to the current ratio, the following formula is used to calculate the current ratio: Timing coefficient of each equipment in the computer room:

[0115]

[0116]

[0117] in, Indicates the The timing coefficient of each equipment in the computer room, Indicates the The average current ratio of the equipment in each computer room is: Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The number of power units contained in each equipment room, Represents historical cycles;

[0118] The timing coefficients of each equipment in the computer room are collected to form a timing coefficient set, and timing coefficient data is obtained.

[0119] Specifically, for the The first of the equipment in the computer room electrical energy units, can be obtained by the formula To calculate the energy unit in a specific historical period Current ratio within .in, Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The total current value of the equipment in the computer room, No. The first The unit current value of each electric energy unit, For example, if the total current value of the equipment in the first computer room is 10A and the current value of the first power unit inside it is 3A, then the current ratio of the power unit is .

[0120] Next, to calculate the The timing coefficient of the equipment in the computer room must first be calculated by the formula To calculate the average current ratio of all power units in the device .

[0121] For example, the first equipment room has three power units, and their current ratios are 0.3, 0.4, and 0.3 respectively. Then the average current ratio of the equipment is that is .

[0122] Furthermore, in order to measure the difference between the current proportion of each power unit and its average value, the formula Calculate the timing coefficient of the i-th equipment in the computer room. Continuing with the above example, for the first equipment in the computer room, its timing coefficient is .

[0123] Ultimately, through the above process, the timing coefficients for each device in the computer room can be calculated and collected to form a set of timing coefficients, known as timing coefficient data. This set reflects the temporal trends in the current proportions of the power units within each device in the computer room, which is crucial for gaining a deeper understanding of the device's operating status and optimizing power management.

[0124] In step S16, the influence value of each equipment in the computer room is constructed according to the time series coefficient data and the time series weight data, and the probability distribution of the influence value is calculated to obtain the total time series influence probability.

[0125] In a specific embodiment, constructing the influence value of each equipment in the computer room according to the timing coefficient data and the timing weight data and calculating the probability distribution of the influence value to obtain the total timing influence probability includes:

[0126] Accumulate and sum the timing coefficient data and the corresponding timing weight data of all equipment in the computer room to obtain the impact value data of each equipment in the computer room;

[0127] Taking the impact value data of each equipment in the computer room in several historical periods as discrete data, performing statistical classification on the discrete data, and obtaining probability distribution data of the impact values of several training levels;

[0128] Calculating the probability of the impact value of each training level according to the probability distribution data to obtain a plurality of level probability data;

[0129] Taking each training level as an index, several level probability data are accumulated and summed to obtain the total time series impact probability.

[0130] In step S16, the process of constructing the impact value of each equipment in the computer room and calculating the probability distribution of the impact value to obtain the total time series impact probability can be specifically implemented in several stages:

[0131] Specifically, based on the previously obtained time series coefficient data and time series weight data, the impact value of each equipment room is constructed. The equipment in the computer room can be The impact value is calculated, where Representative The impact value of each equipment room; Represents the The timing coefficient of each equipment in the computer room reflects the degree of fluctuation of the current proportion of its internal power unit; It represents the timing weight of the device and reflects the changing characteristics of its power consumption at different time points.

[0132] Next, the impact value data for each device in the computer room over several historical periods is treated as discrete data and statistically classified to obtain probability distribution data for impact values at several training levels. This step primarily converts continuous impact values into discrete levels to facilitate subsequent probability calculations. For example, all data can be categorized into three levels based on the impact value: low, medium, and high, with each level corresponding to a specific impact value range.

[0133] Next, the probability of the impact value at each training level is calculated based on the probability distribution data, resulting in a number of level probability data. This means calculating the proportion of impact values that fall within each level range. Assuming that in all historical data, 30% of the impact values fall within the low level range, 50% fall within the medium level range, and 20% fall within the high level range, then the probabilities of the low, medium, and high levels are 0.3, 0.5, and 0.2, respectively.

[0134] Finally, using each training level as an index, the probability data of several levels are accumulated and summed to obtain the total time series impact probability. This step is to comprehensively consider the impact of all equipment in the computer room at different time points to evaluate the overall power consumption impact of the entire computer room system. For example, if the probabilities of the low, medium, and high levels of all equipment in the computer room calculated by the above method are added together, that is, , which indicates that within the time range considered, the impact values of all equipment in the computer room have completely covered all possibilities, and the total time series impact probability is 1, which means that the impact values of all equipment have been fully considered without omission.

[0135] Through the above process, not only can the impact value of each computer room equipment be constructed, but also the probability distribution of these impact values at different levels can be calculated, and then the total time series impact probability can be obtained, providing data support for the intelligent monitoring of the computer room power.

[0136] In step S17, the training set and the test set are used as training data, and the neural network model is trained according to the total time series impact probability, the time sequence of the collected data and the actual electric energy data to obtain a trained neural network model.

[0137] In a specific embodiment, the training process of the neural network model includes:

[0138] Inputting the training data into the input layer of the neural network model in chronological order for training, and obtaining predicted value data output by the output layer of the neural network model;

[0139] Substituting the predicted value data and the actual electric energy data into a loss function to calculate a loss value to obtain loss value data;

[0140] Calculating the gradient of the output layer output of the neural network model based on the loss value data, and passing the gradient forward layer by layer through the chain rule to calculate the gradient of the parameters of each layer to obtain gradient data;

[0141] Update the parameters of each layer of the neural network model based on the gradient data and the preset learning rate;

[0142] The parameters of each layer are updated repeatedly until the training is completed when the number of training times of the neural network model is greater than a preset number, or when the loss function value of the neural network model is less than a preset loss threshold.

[0143] In a specific embodiment, the calculating the loss value data using the predicted value and the actual electric energy data to obtain the loss value data includes:

[0144] The loss value data is calculated by the following loss function:

[0145]

[0146] in, Represents loss value data, Represents the power forecast data, Indicates the actual data of electric energy. Indicates the number of computer rooms.

[0147] Specifically, the training and test sets are first fed into the neural network model's input layer in the chronological order of data collection for training. This means that the input data enters the model sequentially, ensuring that the model captures the temporal dependencies within the time series data. The input data includes the total time series impact probability, the chronological order of the collected data, and actual power data. This data is passed through the neural network's input layer, processed by the hidden layer, and ultimately generates predicted values at the output layer.

[0148] Next, the predicted value data output by the neural network model and the actual power data are substituted into the loss function to calculate the loss value and obtain the loss value data. The loss function is used to quantify the difference between the model's predicted value and the actual value. The loss value data can be calculated using the following formula:

[0149]

[0150] in, Represents loss value data, Represents the power forecast data, Indicates the actual data of electric energy. Represents the number of computer rooms. The smaller the loss value, the smaller the difference between the model's predicted value and the actual value, and the better the model performance.

[0151] Then, the gradient of the output layer of the neural network model is calculated based on the loss value data. The gradient is then passed forward layer by layer using the chain rule to calculate the gradient of the parameters of each layer and obtain the gradient data. Gradient calculation is the core of the backpropagation algorithm and is used to determine in which direction the model parameters should be adjusted to minimize the loss function. Specifically, for each neuron in the output layer, its gradient can be calculated using the following formula:

[0152]

[0153] in, Represents the output layer The input of a neuron Represents the output layer The output of a neuron, Represents loss value data.

[0154] Then, the gradient is passed forward layer by layer through the chain rule, the gradient of each layer parameter is calculated, and finally the gradient data is obtained.

[0155] Next, based on the gradient data and the preset learning rate, the parameters of each layer of the neural network model are updated. Parameter update is achieved through the gradient descent algorithm, and the specific formula is:

[0156]

[0157] in, Indicates the parameters, represents the learning rate, Represents the loss function for the The learning rate determines the step size of the parameter update. A too large learning rate may cause the parameters to update too quickly, resulting in model oscillation and non-convergence; a too small learning rate may cause the parameters to update too slowly, resulting in slow model convergence.

[0158] Finally, the parameters of each layer are updated repeatedly until the number of training times of the neural network model is greater than the preset number, or when the loss function value of the neural network model is less than the preset loss threshold, the training is determined to be completed:

[0159] For example, assuming the preset maximum number of training times is 1000 and the preset loss threshold is 0.01, during the training process, the loss function value of the model gradually decreases. When the number of training times reaches 1000, or when the loss function value is less than 0.01 for the first time, the training process ends and a trained neural network model is obtained.

[0160] Through the above process, the neural network model can be effectively trained to enable it to accurately predict future power consumption based on the input total time series impact probability, the time sequence of the collected data and the actual power data.

[0161] In step S18, the output of the neural network model is determined as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment.

[0162] In a specific embodiment, the output of the neural network model is determined as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment, including:

[0163] The power forecast data is determined by the following formula:

[0164]

[0165] in, Indicates the The equipment in the computer room is The percentage of electric energy in a historical period, is the output function of the neural network, Indicates the The equipment in the computer room is The normalized total current value of the historical cycle, Indicates the The equipment in the computer room is Normalized total voltage value of the historical period.

[0166] The significance of this step is that by predicting the percentage of electricity consumed by each computer room device in the future through the output of the neural network model, computer room managers can understand the power requirements of the equipment in advance, reasonably arrange power supply, and avoid power shortages or waste. At the same time, by predicting power consumption, they can optimize energy use, improve energy efficiency, prevent equipment failures, perform maintenance in a timely manner, reduce unexpected downtime, and ensure service continuity and stability.

[0167] To facilitate understanding of the present invention, some preferred embodiments of the present invention are further described below.

[0168] The following describes the working process of the present invention using a common scenario as an example. Figure 1 , an intelligent monitoring method for computer room power, comprising the following steps:

[0169] First, the intelligent monitoring system regularly collects power consumption data from each device in the computer room and records the timestamp of this data. This raw data is transmitted to the central processing unit, which processes and analyzes it. The system also collects information on environmental factors such as temperature and humidity, as these factors also affect equipment energy consumption.

[0170] The system then preprocesses the collected data to ensure data quality for subsequent analysis. This preprocessed data is used to calculate the timing coefficient and timing weight for each device. The timing coefficient reflects the fluctuation in the current share of the device's internal energy units, while the timing weight reflects the changing characteristics of the device's energy consumption at different time points. These two metrics together determine the impact value of each device. Furthermore, through statistical classification, the probability distribution data of these impact values can be obtained, forming the total timing impact probability.

[0171] Next, the system trains a neural network model based on the constructed time-series total impact probability and the collected energy data. The training dataset includes historical energy consumption data, the time-series total impact probability, and the chronological order of data collection. The neural network model aims to learn the relationship between this input data and actual energy consumption, thereby accurately predicting energy demand at a specific point in the future.

[0172] During training, the system continuously adjusts the neural network's parameters to minimize the gap between predicted and actual values. This is achieved by calculating a loss function and then adjusting the model parameters based on this loss through backpropagation. As training progresses, the model's prediction accuracy gradually improves.

[0173] Once trained, the intelligent monitoring system will be able to predict future energy demand based on current and past energy consumption data. This predictive capability helps managers plan power supply in advance and avoid power shortages or surpluses.

[0174] In summary, the present invention collects and processes historical current and voltage data of computer room equipment, constructs training and test sets, calculates actual power usage and timing weights and coefficients, constructs a probability distribution of influence values, and trains a neural network model to predict the future power consumption percentage of each device, thereby achieving accurate prediction of the power consumption of computer room equipment.

[0175] Reference Figure 2 The second embodiment of the present invention provides an intelligent monitoring device for a computer room power system, comprising:

[0176] The data acquisition module is used to collect the total current value and the corresponding total voltage value of each equipment in the computer room within several historical periods, as well as the unit current value of each power unit;

[0177] A data preprocessing module, configured to normalize the total current value and the total voltage value to construct a training set and a test set;

[0178] an actual electric energy calculation module, configured to calculate according to the total current value and the total voltage value to obtain actual electric energy data;

[0179] A time series weight construction module is used to construct the time series weight of each equipment in the computer room based on the increment of the total current value in adjacent historical periods to obtain time series weight data;

[0180] A timing coefficient calculation module, configured to calculate the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value, and obtain timing coefficient data;

[0181] An influence value calculation module is used to construct an influence value of each equipment in the computer room using the time series coefficient data and the time series weight data, and calculate the probability distribution of the influence value to obtain the total time series influence probability;

[0182] A neural network training module is used to use the training set and the test set as training data, train the neural network model according to the total time series impact probability, the time sequence of the collected data and the real-time electric energy actual data, and obtain a trained neural network model;

[0183] The output module is used to determine the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment.

[0184] It should be noted that the intelligent monitoring device for computer room power provided in an embodiment of the present invention is used to execute all the process steps of the intelligent monitoring method for computer room power in the above embodiment. The working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0185] An embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an intelligent monitoring program for power in a computer room. When the processor executes the computer program, the steps of the above-mentioned intelligent monitoring method for power in a computer room are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are realized, such as the intelligent monitoring module for the power of the computer room.

[0186] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0187] The electronic device may be a computing device such as a desktop computer, notebook, PDA, or smart tablet. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that the aforementioned components are merely examples of electronic devices and do not constitute a limitation of the electronic device. The electronic device may include more or fewer components than those described above, or a combination of certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, and the like.

[0188] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the electronic device and connects various parts of the entire electronic device using various interfaces and lines.

[0189] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0190] If the module / unit integrated into the electronic device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium can be appropriately increased or decreased based on the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, based on legislation and patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0191] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0192] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An intelligent monitoring method for computer room power, characterized in that: The following steps are involved: Collect the total current value and corresponding total voltage value of each equipment in the computer room within several historical periods, as well as the unit current value of each power unit; Normalizing the total current value and the total voltage value to construct a training set and a test set; Calculating according to the total current value and the total voltage value to obtain actual electric energy data; Constructing a time series weight for each device in the computer room according to the increment of the total current value in adjacent historical periods to obtain time series weight data; Calculate the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value to obtain timing coefficient data; Constructing an impact value of each equipment in the computer room according to the timing coefficient data and the timing weight data, and calculating a probability distribution of the impact value to obtain a total timing impact probability; Using the training set and the test set as training data, training a neural network model according to the total time series impact probability, the time sequence of the collected data, and the actual electric energy data, to obtain a trained neural network model; The output of the neural network model is determined as electric energy prediction data, wherein the electric energy prediction data is used to predict the percentage of electric energy consumed by each equipment in the computer room at a future moment.

2. The intelligent monitoring method for computer room power according to claim 1, characterized in that: The step of constructing a time series weight for each device in the computer room according to the increment of the total current value in adjacent historical periods to obtain time series weight data includes: For the The following formula is used to calculate the equipment in each historical period. Total current increment within: in, Indicates the The equipment in the computer room is The total current value increment of the historical cycle, Indicates the The equipment in the computer room is The total current value of the historical cycle, Indicates the The equipment in the computer room is The total current value of the historical cycle, where , Indicates the total number of historical cycles; Collect equipment from each computer room The total current increments in all historical cycles form a total current increment set, which is: ; The total current increment set is used as an input variable and input into the following function to select the median as the timing weight: in, Indicates the The timing weight of each equipment in the computer room, represents a preset computer program algorithm; The timing weights of all equipment in the computer room are collected to form a timing weight set to obtain timing weight data.

3. The intelligent monitoring method for computer room power according to claim 1, characterized in that: The calculating the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value to obtain timing coefficient data includes: According to The total current value of the equipment in the computer room and the corresponding The unit current value of the electric energy unit is calculated by the following formula Among the equipment in the computer room The current ratio of each energy unit: in, Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The total current value of the equipment in the computer room, No. The first The unit current value of each electric energy unit, Represents historical cycles; According to the current ratio, the following formula is used to calculate the current ratio: Timing coefficient of each equipment in the computer room: in, Indicates the The timing coefficient of each equipment in the computer room, Indicates the The average current ratio of the equipment in each computer room is: Indicates the Among the equipment in the computer room The proportion factor of each electric energy unit, Indicates the The number of power units contained in each equipment room; The timing coefficients of each equipment in the computer room are collected to form a timing coefficient set, and timing coefficient data is obtained.

4. The intelligent monitoring method for computer room power according to claim 1, characterized in that: The step of constructing the influence value of each equipment in the computer room according to the time series coefficient data and the time series weight data and calculating the probability distribution of the influence value to obtain the total time series influence probability includes: Accumulate and sum the timing coefficient data and the corresponding timing weight data of all equipment in the computer room to obtain the impact value data of each equipment in the computer room; Taking the impact value data of each equipment in the computer room in several historical periods as discrete data, performing statistical classification on the discrete data, and obtaining probability distribution data of the impact values of several training levels; Calculating the probability of the impact value of each training level according to the probability distribution data to obtain a plurality of level probability data; Taking each training level as an index, several level probability data are accumulated and summed to obtain the total time series impact probability.

5. The intelligent monitoring method for computer room power according to claim 1, characterized in that: The training process of the neural network model includes: Inputting the training data into the input layer of the neural network model in chronological order for training, and obtaining predicted value data output by the output layer of the neural network model; Substituting the predicted value data and the actual electric energy data into a loss function to calculate a loss value to obtain loss value data; Calculating the gradient of the output layer output of the neural network model based on the loss value data, and passing the gradient forward layer by layer through the chain rule to calculate the gradient of the parameters of each layer to obtain gradient data; Update the parameters of each layer of the neural network model based on the gradient data and the preset learning rate; The parameters of each layer are updated repeatedly until the training is completed when the number of training times of the neural network model is greater than a preset number, or when the loss function value of the neural network model is less than a preset loss threshold.

6. The intelligent monitoring method for computer room power according to claim 5, characterized in that: Substituting the predicted value data and the actual electric energy data into a loss function to calculate the loss value to obtain the loss value data includes: The loss value data is calculated by the following loss function: in, Represents loss value data, Represents the power forecast data, Indicates the actual data of electric energy. Indicates the number of computer rooms. Represents a historical cycle, Indicates the period offset.

7. The intelligent monitoring method for computer room power according to claim 1, characterized in that: The step of determining the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each device in the computer room at a future moment, includes: The power forecast data is determined by the following formula: in, Indicates the The equipment in the computer room is The percentage of electric energy in a historical period, is the output function of the neural network, Indicates the The equipment in the computer room is The normalized total current value of the historical cycle, Indicates the The equipment in the computer room is Normalized total voltage value of the historical period.

8. An intelligent monitoring device for power in a machine room, characterized in that: include: The data acquisition module is used to collect the total current value and the corresponding total voltage value of each equipment in the computer room within several historical periods, as well as the unit current value of each power unit; A data preprocessing module, configured to normalize the total current value and the total voltage value to construct a training set and a test set; an actual electric energy calculation module, configured to calculate according to the total current value and the total voltage value to obtain actual electric energy data; A time series weight construction module is used to construct the time series weight of each equipment in the computer room based on the increment of the total current value in adjacent historical periods to obtain time series weight data; A timing coefficient calculation module, configured to calculate the timing coefficient of the corresponding equipment in the computer room according to the unit current value and the total current value, and obtain timing coefficient data; An influence value calculation module is used to construct an influence value of each equipment in the computer room using the time series coefficient data and the time series weight data, and calculate the probability distribution of the influence value to obtain the total time series influence probability; A neural network training module is used to use the training set and the test set as training data, train the neural network model according to the total time series impact probability, the time sequence of the collected data and the actual electric energy data, and obtain a trained neural network model; The output module is used to determine the output of the neural network model as power prediction data, wherein the power prediction data is used to predict the percentage of power consumed by each equipment in the computer room at a future moment.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for intelligent monitoring of computer room power according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the intelligent monitoring method for computer room power according to any one of claims 1 to 7.

Images