Electric energy meter load dynamic scheduling method, system and device and electric energy meter

Abstract

The invention relates to the field of data processing, in particular to an electric energy meter load dynamic scheduling method, system and device and an electric energy meter, and the method comprises the steps: collecting the data of refrigeration equipment, data processing equipment and photovoltaic power generation in a machine room based on the electric energy meter; based on the collected data, using an LSTM model to predict the demand load of the refrigeration equipment; calculating the adjustable load of the refrigeration equipment based on the predicted demand load and the real-time load of the refrigeration equipment; and taking the adjustable load as a boundary constraint, performing global optimization by using an improved PSO algorithm, solving to obtain an optimal scheduling instruction, and issuing and executing the optimal scheduling instruction. According to the method, on the premise of the thermal stability of the machine room, the refrigeration load and the photovoltaic output are dynamically coordinated, so that energy efficiency maximization and energy cost optimization are realized while safe and reliable operation of key equipment is ensured, and the accuracy of load scheduling and the overall energy efficiency of the system are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method, system, apparatus, and electricity meter for dynamic load scheduling of electricity meters. Background Technology

[0002] In the current power system, the widespread use of diverse electrical devices (such as refrigeration equipment and data processing equipment) and the large-scale integration of distributed energy sources (such as photovoltaics) have placed higher demands on refined load scheduling. Existing load scheduling methods are mostly based on simple thresholds or single-objective optimization, which are difficult to adapt to the needs of complex scenarios.

[0003] The Particle Swarm Optimization (PSO) algorithm, due to its applicability to multivariable and nonlinear optimization problems, has been attempted for application in load scheduling. This algorithm uses swarm intelligence to find optimal load allocation schemes in multi-device scenarios. However, when solving multi-constraint, multi-objective problems, the traditional PSO algorithm's search mechanism exhibits poor adaptability due to complex factors such as equipment differences, energy fluctuations, and grid constraints, resulting in limited load scheduling effectiveness. Summary of the Invention

[0004] To address the aforementioned technical problem of limited load dispatching effectiveness due to the limitations of existing technologies, the present invention provides solutions in the following aspects.

[0005] A first aspect of the present invention provides a method for dynamic load scheduling of electricity meters, comprising: Data is collected from the electricity meter, including data from various cooling devices, data processing equipment, and photovoltaic power generation within the computer room. This data includes the real-time load and voltage of the cooling devices, the real-time temperature of the computer room, the real-time load of the data processing equipment, and the real-time output of the photovoltaic power generation. The collected data is then normalized. Based on the collected data, an LSTM model is used to predict the demand load of the refrigeration equipment. The LSTM model is trained using an adaptive step-size sliding window to obtain training samples, and the loss function of the LSTM model incorporates temperature constraints. Based on the predicted demand load and the real-time load of the refrigeration equipment, calculate the adjustable load of the refrigeration equipment; Using adjustable load as boundary constraint, an improved PSO algorithm is used for global optimization to obtain the optimal scheduling instruction, which is then issued for execution. The improved PSO algorithm includes a multi-objective fitness function and an adaptive inertial weight. The multi-objective fitness function integrates the sensitivity level, temperature deviation, and photovoltaic output coefficient of the refrigeration equipment. The adaptive inertial weight is based on the temperature deviation of the refrigeration equipment, the photovoltaic output coefficient, and the real-time load dynamic adjustment of the data processing equipment.

[0006] Preferably, the process of obtaining the sliding window with an adaptive step size includes: Obtain the standard temperature range and temperature warning range in the computer room, record the time it takes for the temperature to rise from normal to reach or exceed the temperature warning range, and use the average of all recorded time lengths as the window length of the sliding window. Calculate the median value of the standard temperature range, then calculate the absolute value of the difference between the real-time maximum temperature in the window and the median value. Use this absolute value of the difference as the degree of temperature deviation within the window. Calculate the product of the exponent of this degree of deviation and the window length, and round the product up to obtain the step size of the sliding window.

[0007] Preferably, if the computer room temperature is within normal range before calculating the adjustable load, the adjustable load is not calculated; if the computer room temperature exceeds the temperature warning range, the adjustable load is calculated, and the calculation process includes: The difference between the predicted demand load and the real-time load of the refrigeration equipment is used as the initial value of the adjustable load; the ratio of the real-time voltage of the refrigeration equipment to the rated voltage of the refrigeration equipment is used as the voltage correction factor; the calculated voltage correction factor is multiplied by the initial value of the adjustable load to obtain the final adjustable load.

[0008] Preferably, the process of obtaining the multi-objective fitness function includes: Calculate the ratio of the absolute value of the adjustable load of each refrigeration unit to the load range of the system's refrigeration units, and use this as the first item; The product of the first term, the sensitivity level of the refrigeration equipment, and the exponential term of the temperature deviation is calculated, and the mean of the product is calculated by iterating through all refrigeration equipment. This mean is then multiplied by the reciprocal of the photovoltaic output coefficient to obtain the multi-objective fitness function.

[0009] Preferably, the ratio of the historical load standard deviation of each cooling device to the mean of the historical load standard deviation of all cooling devices is obtained as the first factor, and the mean of the historical load standard deviation of all data processing devices in the computer room is obtained as the second factor. The product of the first factor and the second factor is used as the sensitivity level of the cooling device. The ratio of the current real-time output of photovoltaic power to the historical maximum output is used as the photovoltaic output coefficient.

[0010] Preferably, the process of obtaining the adaptive inertia weight includes: The negative of the exponential function of the product of the mean of the exponential term of the temperature deviation of all refrigeration equipment, the complement of the photovoltaic output coefficient, and the mean load of all data processing equipment is used as the adaptive inertia weight.

[0011] Preferably, in the improved PSO algorithm, for each refrigeration device, the corresponding adjustable load is used as a reference, and the reference is scaled according to the calculated adaptive inertia weight to determine the search range of the load scheduling value for each refrigeration device.

[0012] A second aspect of the present invention provides a dynamic load scheduling system for electricity meters, comprising: a data acquisition module for synchronously acquiring electricity meter parameters and real-time output of photovoltaic power generation; a demand load prediction module for predicting the demand load of cooling equipment using an LSTM model based on the acquired data; an adjustable load calculation module for calculating a voltage-corrected adjustable load based on the predicted demand load and the real-time load; and a load scheduling optimization module for calculating the optimal load scheduling using an improved PSO algorithm based on the adjustable load.

[0013] A third aspect of the present invention provides a dynamic load scheduling device for electricity meters, comprising: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-described dynamic load scheduling method for electricity meters is implemented.

[0014] In a fourth aspect, the present invention provides an electricity meter that integrates the above-mentioned electricity meter load dynamic scheduling system, and the system includes the above-mentioned electricity meter load dynamic scheduling device for realizing the load dynamic scheduling function.

[0015] The beneficial effects of this invention are: First, by employing an LSTM model that incorporates temperature constraints for load forecasting and innovatively introducing an adaptive step-size sliding window mechanism, the system can dynamically capture the nonlinear and temporal correlation between computer room temperature and cooling load, significantly improving the accuracy of demand load forecasting.

[0016] Then, based on the high-precision predicted load, by calculating the difference between the predicted load and the real-time load and introducing a voltage correction coefficient, an adjustable load that is more in line with the actual operating state of the equipment is obtained. This effectively prevents equipment overload or adjustment failure caused by voltage fluctuations or inaccurate assessments, and defines a safe and feasible search space for the optimization algorithm.

[0017] Finally, using adjustable load as the boundary, an improved PSO algorithm is employed for solution. The core of this approach lies in the improved multi-objective fitness function and adaptive inertia weights. The multi-objective fitness function cleverly unifies the sensitivity level of the cooling equipment, the degree of temperature deviation, and the photovoltaic output coefficient into a single objective, ensuring that while pursuing energy conservation, the scheduling instructions must prioritize the stable operation of critical equipment and the safety of the computer room temperature. The adaptive inertia weights dynamically adjust the algorithm's exploration and development capabilities based on the system's real-time status (temperature, photovoltaic load, data processing equipment load), ensuring rapid and stable convergence to the optimal solution even under complex operating conditions, generating optimal scheduling instructions that balance safety, economy, and efficiency. Attached Figure Description

[0018] Figure 1 This is a flowchart of steps S1-S4 in a dynamic load scheduling method for electricity meters according to an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of a dynamic load scheduling system for electricity meters according to an embodiment of the present invention.

[0020] Figure 3 This is a structural block diagram of an energy meter load dynamic scheduling device according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0022] The application scenario of this invention is as follows: by using electricity meters to monitor and coordinate the data processing equipment and cooling equipment in real time, while ensuring that the computer room environment meets the process requirements, the dynamic and optimized allocation of power load is achieved, and clean energy such as on-site photovoltaic is given priority, so as to achieve safe, efficient and green operation of the computer room.

[0023] The data processing equipment includes servers, computer clusters, etc., and the cooling equipment includes temperature control units such as precision air conditioners.

[0024] In one embodiment, refer to Figure 1 A method for dynamic load scheduling of electricity meters is provided, comprising steps S1-S4, as detailed below: S1: Data collected from the power meter in the computer room includes data from various cooling devices, data processing devices, and photovoltaic power generation.

[0025] The following key data are collected once per minute using an electricity meter: The data collected includes the real-time load of each cooling device in the computer room, the voltage in the power system, the real-time temperature of the computer room, the real-time load of each data processing device, and the real-time photovoltaic output (PV). Furthermore, all the collected data undergoes routine normalization processing.

[0026] After the above data collection and preprocessing steps, the obtained data is more standardized and easier to process, which can provide good input for multi-constraint and multi-objective optimization algorithms, help the algorithms find the optimal scheduling scheme in complex scenarios, and achieve efficient energy utilization and safe operation of equipment.

[0027] S2: Based on the collected data, use an LSTM model to predict the demand load of the refrigeration equipment.

[0028] The computer room contains cooling and data processing equipment. The operation of the data processing equipment generates heat, causing the room temperature to rise. Excessive temperature can affect the performance and lifespan of the data processing equipment. By predicting the cooling equipment's load demand—that is, predicting the load required to maintain the computer room temperature within a standard range—we can ensure that the cooling equipment operates reasonably according to the prediction results, avoiding insufficient cooling that could lead to excessively high temperatures and guaranteeing the stable operation of the data processing equipment within the computer room.

[0029] Historical data (obtained from the S1 data acquisition operation described above) is used to train an LSTM (Long Short-Term Memory) model, which learns how to maintain a stable computer room temperature and outputs the load value required by the cooling equipment.

[0030] First, the model input comes from historical time series data, including three feature dimensions: normalized real-time load of cooling equipment, normalized real-time temperature of the computer room, and normalized real-time load of data processing equipment.

[0031] Then, considering the complex environment of the computer room, the demand for cooling equipment and temperature fluctuate over time. For example, the data processing equipment operates at different intensities during different business periods, resulting in different amounts of heat generated, which leads to corresponding changes in the computer room temperature and cooling load demand.

[0032] Therefore, by using an adaptive step-size sliding window mechanism when acquiring training samples, the window step size can be adjusted according to the real-time temperature. When the temperature changes drastically and deviates significantly from the standard range, the step size is reduced, and data is captured more frequently to capture such dynamic changes in a timely manner, enabling the model to learn more detailed load change patterns.

[0033] For example, review the specifications of all critical equipment in the computer room to identify their operating temperature requirements. Within the permissible range of all equipment, and considering industry recommendations and your own policies, determine a final, unified standard temperature range. Then, set a temperature warning range based on this standard range. This is achieved by narrowing the temperature difference by 10% from both the upper and lower limits of the standard range, creating a more stringent temperature interval that is more likely to trigger an alert. For instance, if the standard temperature range is [20℃, 30℃], then the temperature warning range would be [21℃, 29℃].

[0034] The aforementioned standard temperature range is an ideal temperature range set to ensure the normal operation and stable performance of the equipment in the computer room. The temperature warning range, as a sub-range of the standard temperature range, is relatively more stringent and is used to provide early warning of situations where the temperature may have an adverse effect on the equipment.

[0035] Observe the temperature data input to the above model and record the length of time it takes for the temperature to rise from a normal state (within the standard temperature range) to reach or exceed the temperature warning range. This is because in actual data center scenarios, it is more common and necessary to pay attention to situations where the temperature rises or exceeds the upper limit.

[0036] For example, if the temperature rises from 22℃ (normal) to 29.5℃ (exceeding the upper limit of the warning range) within a certain time period, and this process takes 30 minutes, then this 30-minute period is a recorded time interval. All such recorded time intervals are aggregated, and the average is calculated. For example, if five time intervals are recorded: 30 minutes, 40 minutes, 35 minutes, 25 minutes, and 45 minutes, then the average is 35 minutes, and this 35-minute period is the window length of the sliding window.

[0037] If the calculated mean is not an integer, it needs to be rounded to the nearest integer.

[0038] Furthermore, the window sliding step size is dynamically adjusted based on the degree of temperature deviation within the window length. The specific operation is as follows: Calculate the median value of the standard temperature range, then calculate the absolute value of the difference between the real-time maximum temperature in the window and the median value. Use this absolute value of the difference as the degree of temperature deviation within the window. Calculate the product of the exponent of this degree of deviation and the window length, and round the product up to obtain the step size of the sliding window.

[0039] The formula for calculating the step size is as follows: In the formula, The step size of the sliding window. The degree of temperature deviation within the window, This is the window length.

[0040] In summary, by obtaining the window length of the sliding window and the calculation method for the sliding step size at each slide, an adaptive step-size sliding window mechanism is derived. This mechanism is used to extract training samples and their corresponding labels from the input data of the aforementioned model. The labels are defined after the window ends. The average actual load of the refrigeration equipment within each time step.

[0041] Secondly, the loss function required for model training improves the model's performance in maintaining stable ambient temperature by introducing temperature constraints.

[0042] The loss function required for model training is obtained by fusing the mean squared error of all existing samples with the load scheduling sensitivity index of samples calculated based on temperature.

[0043] For example, the sum of squares of the deviations of all temperature values ​​within the calculated window (i.e., the absolute values ​​of the differences between each temperature value within the calculated window and the median value of the standard temperature range) is transformed using a hyperbolic tangent function to obtain the load scheduling sensitivity of the sample. This transformation converts the temperature deviation into a sensitivity index within a specific range, such that the index value increases as the actual temperature approaches or exceeds the standard temperature threshold (i.e., the deviation is greater), thus highlighting the loss weight of samples with high load scheduling sensitivity.

[0044] The loss function described above can be expressed as follows: In the formula, For the sample Forecasted demand load Compared with the actual load The mean square error, For the sample The sensitivity of load scheduling The total number of training samples. It is the hyperbolic tangent function.

[0045] Through the above operations, an LSTM model for predicting the demand load of cooling equipment is obtained. This model uses an adaptive sliding window to process real-time data and is trained using a comprehensive loss function, enabling it to accurately predict the load required to maintain normal temperatures. This provides a crucial input for intelligent load scheduling systems, ensuring that the computer room temperature is always controlled within the standard range.

[0046] Finally, based on the trained LSTM model described above, the past... Real-time data at each time step, outputting future... The predicted average load that the refrigeration equipment needs to provide within a given time step, i.e., the predicted demand load of the refrigeration equipment.

[0047] It should be noted that the above S2 operation is performed on a single refrigeration unit.

[0048] S3: Calculate adjustable load based on predicted demand load and real-time load of refrigeration equipment.

[0049] Although S2 above has obtained the predicted demand load for the cooling equipment, it's important to consider that changes in load directly affect the heat generation of the equipment in the computer room, thus impacting the room temperature. If load scheduling is based solely on the predicted load without considering the current temperature, it could lead to further temperature escalation. For example, the predicted load might indicate a need to reduce the load, but if the current temperature is already close to the warning level, blindly reducing the load could cause the cooling equipment to operate insufficiently, resulting in a continued temperature rise. Therefore, using a temperature exceeding the warning level as a trigger condition allows for a comprehensive consideration of the interaction between load and temperature.

[0050] When the computer room temperature exceeds the temperature warning range, load scheduling is triggered based on the prediction results of S2 above, thereby executing the calculation of the adjustable load of the cooling equipment. The calculation process of the adjustable load is as follows: The difference between the predicted demand load and the real-time load is used as the initial value of the adjustable load. Combined with the electrical safety constraints of the refrigeration equipment, the initial value of the adjustable load is corrected to avoid electrical hazards caused by excessive adjustment of the equipment load when the voltage deviates from the rated voltage.

[0051] The ratio of the real-time voltage of the refrigeration equipment to its rated voltage is used as the voltage correction coefficient. The calculated voltage correction coefficient is multiplied by the initial value of the adjustable load to obtain the final adjustable load.

[0052] S4: Using adjustable load as boundary constraint, the improved PSO algorithm is used for global optimization to obtain the optimal scheduling instruction, which is then issued for execution.

[0053] Since the S3 calculation above yields the theoretical adjustable load for each refrigeration unit, in order to avoid all units adjusting significantly at the same time, it is necessary to reasonably allocate the adjustment tasks among multiple units and find the overall optimal scheduling scheme so that the load scheduling effect can be maximized.

[0054] Considering the principles of prioritizing data center safety and energy efficiency, this study integrates three constraints—data center temperature deviation, equipment sensitivity, and photovoltaic power output—into the fitness function and inertia weight of the PSO algorithm. This allows the scheduling algorithm to dynamically adjust its search strategy based on actual conditions, ensuring both safety and improved energy utilization.

[0055] The aforementioned improvements to the PSO algorithm include obtaining the multi-objective fitness function and adaptive inertia weights.

[0056] The steps for obtaining the multi-objective fitness function are as follows: First, the ratio of the historical load standard deviation of each cooling device to the mean of the historical load standard deviation of all cooling devices is obtained as the first factor. The mean of the historical load standard deviation of all data processing devices in the computer room is obtained as the second factor. The product of the first factor and the second factor is used as the sensitivity level of the cooling device.

[0057] Then, the absolute value of the difference between the real-time temperature and the center value of the standard temperature range is converted into an exponential term (in the form of...). This index item is used to determine the degree of temperature deviation of the refrigeration equipment.

[0058] Secondly, the ratio of the current real-time photovoltaic output to the historical maximum output is used as the photovoltaic output coefficient.

[0059] Finally, the ratio of the absolute value of the adjustable load of each refrigeration unit to the load range of the system's refrigeration units is calculated and used as the first term. The product of the first term, the sensitivity level of the refrigeration unit, and the exponential term of the temperature deviation is then multiplied, and the average of this product is calculated for all refrigeration units. This average is then multiplied by the reciprocal of the photovoltaic output coefficient to obtain the multi-objective fitness function.

[0060] For example, the above multi-objective fitness function can be expressed as a relational formula: In the formula, For multi-objective fitness functions, For the first The ratio of the absolute value of the adjustable load of the refrigeration equipment to the load range of the system's refrigeration equipment (i.e., the difference between the maximum and minimum load range). For the first Sensitivity level of Taiwan refrigeration equipment For the first An index of the degree of temperature deviation of the refrigeration equipment. This represents the total number of refrigeration equipment. This is the photovoltaic power output coefficient. Let it be a constant to prevent the denominator from being zero. The first part... The second part reflects the comprehensive cost or impact of refrigeration equipment load adjustment. Introducing the impact of photovoltaic output, when photovoltaic output is high, Reduce and encourage load adjustments when photovoltaic power generation is sufficient to reduce dependence on the grid.

[0061] By minimizing the multi-objective fitness function value, an optimal set of load scheduling strategies can be found, that is, the load settings for each refrigeration unit can be determined, thereby balancing multiple objectives: making full use of the unit's regulation capacity, maintaining temperature comfort, utilizing renewable energy and reducing operating costs.

[0062] The process of obtaining adaptive inertia weights is as follows: The negative of the exponential function of the product of the mean of the exponential term of the temperature deviation of all refrigeration equipment, the complement of the photovoltaic output coefficient, and the mean load of all data processing equipment is used as the adaptive inertia weight.

[0063] For example, the adaptive inertia weights described above satisfy the following relationship: In the formula, For adaptive inertia weights, This is the mean of the index term representing the degree of temperature deviation across all refrigeration equipment. This is the photovoltaic power output coefficient. This represents the average load of all data processing equipment in the computer room. It is an exponential function with base e. When big, Low, When large, it leads to A smaller value indicates that the refrigeration equipment is under stress and requires more aggressive adjustments; conversely, a larger value indicates that the refrigeration equipment is in good condition and allows for more conservative exploration.

[0064] Based on the above operations, the core components of the improved PSO algorithm are obtained, and then the final operation is performed. The specific process is as follows: In the improved PSO algorithm of this invention, the particle is one. The dimension vector represents a possible scheduling scheme, and its search interval is dynamically determined based on the adaptive inertia weights calculated above and the schedulable load of the refrigeration equipment. Each dimension represents the first... Adjustable load of the refrigeration equipment.

[0065] For refrigeration equipment requiring scheduling, the corresponding adjustable load is used as a benchmark. This benchmark is then scaled according to the calculated adaptive inertia weights to determine the search range for the load scheduling values ​​of each refrigeration unit. The search interval of dimension satisfies: When the When the adjustable load of a dimension is greater than 0, the value range is: .

[0066] When the When the adjustable load of a dimension is less than 0, the value range is: .

[0067] In summary, the multi-objective fitness function, serving as an evaluation criterion for the quality of particles (i.e., scheduling schemes), directly guides the particle swarm to search in the direction with the lowest overall cost by minimizing the function value, ensuring that the final solution can simultaneously consider the fairness of equipment regulation, the safety of computer room temperature, and the efficient utilization of photovoltaic energy. The adaptive inertia weight dynamically regulates the search behavior and search range of the particle swarm. The weight is adaptively adjusted according to the real-time status of the system (average temperature deviation, photovoltaic output, and data processing equipment load), thereby achieving a balance between "global exploration" and "local fine search". This not only improves the convergence speed and accuracy of the algorithm, but also enables the generated scheduling instructions to respond to environmental changes in real time, making them more robust.

[0068] Ultimately, the improved PSO algorithm, based on the aforementioned mechanism, efficiently searches for the globally optimal load scheduling command for the cooling equipment within the solution space constrained by the dynamically determined search interval. This optimal load scheduling command is then converted into a series of specific load commands for the cooling equipment. These commands are directly distributed to the smart meters connected to each cooling device through the energy management system. The smart meters, according to the received commands, precisely adjust the cooling equipment to the target range.

[0069] In one embodiment, refer to Figure 2 A dynamic load dispatching system for electricity meters is provided, comprising: The data acquisition module is used to synchronously collect the parameters of the electricity meter and the real-time output of photovoltaic power generation; The demand load forecasting module is used to predict the demand load of refrigeration equipment based on the collected data and using an LSTM model. The adjustable load calculation module is used to calculate the voltage-corrected adjustable load based on the predicted demand load and the real-time load. The load scheduling optimization module is used to calculate the optimal load scheduling based on the adjustable load using an improved PSO algorithm.

[0070] In one embodiment, refer to Figure 3 A dynamic load dispatching device for electricity meters is provided, comprising: The system includes a processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the aforementioned dynamic load scheduling method for electricity meters.

[0071] In one embodiment, an electricity meter is provided that integrates the aforementioned electricity meter load dynamic scheduling system, and the system includes the aforementioned electricity meter load dynamic scheduling device for realizing the load dynamic scheduling function.

[0072] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for dynamic load scheduling of electricity meters, characterized in that, include: Data is collected from the electricity meter, including data from various cooling devices, data processing equipment, and photovoltaic power generation within the computer room. This data includes the real-time load and voltage of the cooling devices, the real-time temperature of the computer room, the real-time load of the data processing equipment, and the real-time output of the photovoltaic power generation. The collected data is then normalized. Based on the collected data, an LSTM model is used to predict the demand load of the refrigeration equipment. The LSTM model is trained using an adaptive step-size sliding window to obtain training samples, and the loss function of the LSTM model incorporates temperature constraints. Based on the predicted demand load and the real-time load of the refrigeration equipment, calculate the adjustable load of the refrigeration equipment; Using adjustable load as the boundary constraint, an improved PSO algorithm is used for global optimization to obtain the optimal scheduling instruction, which is then issued and executed. The improved PSO algorithm includes a multi-objective fitness function and an adaptive inertial weight. The multi-objective fitness function integrates the sensitivity level, temperature deviation, and photovoltaic output coefficient of the refrigeration equipment. The adaptive inertial weight is dynamically adjusted based on the temperature deviation, photovoltaic output coefficient, and real-time load of the data processing equipment.

2. The method for dynamic load scheduling of an electricity meter according to claim 1, characterized in that, The process of obtaining an adaptive step-size sliding window includes: Obtain the standard temperature range and temperature warning range in the computer room, record the time it takes for the temperature to rise from normal to reach or exceed the temperature warning range, and use the average of all recorded time lengths as the window length of the sliding window. Calculate the median value of the standard temperature range, then calculate the absolute value of the difference between the real-time maximum temperature in the window and the median value. Use this absolute value of the difference as the degree of temperature deviation within the window. Calculate the product of the exponent of this degree of deviation and the window length, and round the product up to obtain the step size of the sliding window.

3. The method for dynamic load scheduling of an electricity meter according to claim 2, characterized in that, If the computer room temperature is within normal limits before calculating the adjustable load, the adjustable load calculation will not be performed; if the computer room temperature exceeds the temperature warning range, the adjustable load calculation will be performed, and the calculation process includes: The difference between the predicted demand load and the real-time load of the refrigeration equipment is used as the initial value of the adjustable load; the ratio of the real-time voltage of the refrigeration equipment to the rated voltage of the refrigeration equipment is used as the voltage correction factor; the calculated voltage correction factor is multiplied by the initial value of the adjustable load to obtain the final adjustable load.

4. The method for dynamic load scheduling of an electricity meter according to claim 3, characterized in that, The process of obtaining the multi-objective fitness function includes: Calculate the ratio of the absolute value of the adjustable load of each refrigeration unit to the load range of the system's refrigeration units, and use this as the first item; The product of the first term, the sensitivity level of the refrigeration equipment, and the exponential term of the temperature deviation is calculated, and the mean of the product is calculated by iterating through all refrigeration equipment. This mean is then multiplied by the reciprocal of the photovoltaic output coefficient to obtain the multi-objective fitness function.

5. The method for dynamic load scheduling of an electricity meter according to claim 4, characterized in that, The ratio of the historical load standard deviation of each cooling device to the mean of the historical load standard deviation of all cooling devices is used as the first factor. The mean of the historical load standard deviation of all data processing devices in the computer room is used as the second factor. The product of the first factor and the second factor is used as the sensitivity level of the cooling device. The ratio of the current real-time output of photovoltaic power to the historical maximum output is used as the photovoltaic output coefficient.

6. The method for dynamic load scheduling of an electricity meter according to claim 5, characterized in that, The process of obtaining the adaptive inertia weight includes: The negative of the exponential function of the product of the mean of the exponential term of the temperature deviation of all refrigeration equipment, the complement of the photovoltaic output coefficient, and the mean load of all data processing equipment is used as the adaptive inertia weight.

7. The method for dynamic load scheduling of an electricity meter according to claim 6, characterized in that, In the improved PSO algorithm, for each refrigeration unit, the corresponding adjustable load is used as a reference, and the reference is scaled according to the calculated adaptive inertia weight to determine the search range of the load scheduling value for each refrigeration unit.

8. A dynamic load dispatching system for electricity meters, characterized in that, include: The data acquisition module is used to synchronously collect the parameters of the electricity meter and the real-time output of photovoltaic power generation; The demand load forecasting module is used to predict the demand load of refrigeration equipment based on the collected data and using an LSTM model. The adjustable load calculation module is used to calculate the voltage-corrected adjustable load based on the predicted demand load and the real-time load. The load scheduling optimization module is used to calculate the optimal load scheduling based on the adjustable load using an improved PSO algorithm.

9. A dynamic load dispatching device for electricity meters, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the dynamic load scheduling method for electricity meters according to any one of claims 1-7.

10. An electricity meter, characterized in that, The system integrates the dynamic load scheduling system for electricity meters as described in claim 8, and the system includes the dynamic load scheduling device for electricity meters as described in claim 9, for realizing the dynamic load scheduling function.

Images