Method and device for controlling electromagnetic loss of power transformer in intelligent power grid environment

Abstract

The invention provides a power transformer electromagnetic loss control method and device in an intelligent power grid environment, and relates to the field of intelligent power grids, and the method comprises the steps: carrying out the cooling demand prediction of a transformer according to a predicted power load sequence and a space temperature sequence, and outputting a predicted cooling load sequence; and with meeting of the predicted cooling load sequence as a constraint and minimization of energy consumption and oil temperature fluctuation as optimization targets, cooling parameters of the forced oil cooling device are optimized, an optimal cooling parameter sequence is output, and the forced oil cooling device is controlled to dissipate heat of the transformer. The method aims to solve the technical problems that a traditional control method lacks cooling strategy dynamic adjustment capacity, temperature rise abnormity is prone to occurring in a high-load or high-temperature environment, transformer iron core loss and winding loss are remarkably increased, and electromagnetic loss is high, accurate and efficient cooling control can be achieved, the electromagnetic loss is effectively reduced, and the service life of a transformer is prolonged. And the operation efficiency and stability of the transformer are improved.

Description

Technical Field

[0001] The present invention relates to the field of smart grids, and in particular to a method and device for controlling electromagnetic losses of a power transformer in a smart grid environment. Background Art

[0002] During operation, transformers generate electromagnetic losses, mainly including core loss and winding loss. Temperature rise is a key factor affecting the change of electromagnetic loss. For this reason, cooling systems (such as forced oil cooling systems) are widely used in power transformers to control the equipment temperature and ensure its safe and stable operation.

[0003] However, most transformer cooling systems in existing technologies adopt static or rule-driven control strategies, which usually trigger the start and stop of fans or oil pumps based on the current temperature threshold. They lack the ability to effectively perceive and predict future load changes and ambient temperature trends, and cannot achieve dynamic optimization and adjustment of cooling strategies. This control method has problems such as response lag, excessive energy consumption or overcooling. Especially under high load or external high temperature conditions, the transformer is prone to abnormal temperature rise, resulting in a significant increase in core hysteresis loss and winding resistance loss, causing high electromagnetic loss, which in turn affects the operating efficiency and service life of the equipment. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and device for controlling electromagnetic losses in power transformers in a smart grid environment, to address the technical problems that conventional transformer cooling system control methods lack the ability to dynamically adjust cooling strategies, are prone to abnormal temperature rises under high load or high temperature environments, and significantly increase transformer core and winding losses, leading to high electromagnetic losses. The method and device include: In a first aspect, the present invention provides a method for controlling electromagnetic losses of power transformers in a smart grid environment, comprising: analyzing and obtaining a predicted power load sequence of the transformer in a preset future time zone, and a spatial temperature sequence of the area in which the transformer is located; predicting the cooling demand of the transformer based on the predicted power load sequence and the spatial temperature sequence, and outputting a predicted cooling load sequence; optimizing the cooling parameters of a forced oil cooling device with the satisfaction of the predicted cooling load sequence as a constraint and minimizing energy consumption and oil temperature fluctuations as the optimization goal, and outputting an optimal cooling parameter sequence; and controlling the forced oil cooling device in accordance with the optimal cooling parameter sequence to dissipate heat from the transformer in the preset future time zone.

[0005] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment further includes: monitoring and obtaining the power load of the transformer at P consecutive monitoring time points within a historical period to obtain a historical power load sequence; monitoring and obtaining load fluctuation correlation data of the transformer in a preset future time zone based on a preset load fluctuation correlation factor; and predicting and obtaining a predicted power load sequence of the transformer in a preset future time zone based on the historical power load sequence and the load fluctuation correlation data.

[0006] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment further includes: taking the historical time period as a constraint, collecting a sample power load sequence set and a sample load fluctuation associated data set of the transformer within the same time period, and obtaining the historical power load sequence of the transformer within a historical preset time zone as a sample predicted power load sequence set, wherein the historical time period and the historical preset time zone are continuous time periods, and the time interval of the historical preset time zone is the same as the preset future time zone; using the sample power load sequence set and the sample load fluctuation associated data set as input, using the sample predicted power load sequence set as supervision, training the long short-term memory network until convergence, and obtaining a power load forecasting model; inputting the historical power load sequence and load fluctuation associated data into the power load forecasting model, and outputting a predicted power load sequence.

[0007] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment also includes: collecting a sample power load sequence set and a sample space temperature sequence set based on the historical operation records of the transformer, and obtaining the historical cooling load sequence required under different sample power load sequence and sample space temperature sequence scenarios to obtain a sample cooling load sequence set; using the sample power load sequence set, sample space temperature sequence set and sample cooling load sequence set to train a long short-term memory network until convergence to obtain a cooling demand prediction model; using the cooling demand prediction model to predict cooling demand based on the predicted power load sequence and space temperature sequence, and output a predicted cooling load sequence.

[0008] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment also includes: obtaining the continuous working time of the forced oil cooling device, and analyzing to obtain the continuous working time sequence at the same time node of the predicted cooling load sequence; performing cooling performance attenuation analysis on the forced oil cooling device, and constructing a time-performance attenuation comparison table; based on the time-performance attenuation comparison table, obtaining a performance attenuation coefficient sequence according to the continuous working time sequence; and compensating and updating the predicted cooling load sequence according to the performance attenuation coefficient sequence.

[0009] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment also includes: using the attribute characteristics of the forced oil cooling device as a constraint, retrieving historical operation logs of similar forced oil cooling devices, collecting a set of sample continuous working time, and obtaining the performance attenuation ratio corresponding to different sample continuous working times as a sample performance attenuation coefficient to obtain a set of sample performance attenuation coefficients; determining multiple continuous working time intervals based on the set of sample continuous working time, and randomly selecting a first continuous working time interval to extract a first sample performance attenuation coefficient set covered by the first continuous working time interval; performing anomaly elimination and mean calculation on the first sample performance attenuation coefficient set to obtain the mean of the first sample performance attenuation coefficient; constructing a time-performance attenuation comparison table based on the mapping relationship between the first continuous working time interval and the mean of the first sample performance attenuation coefficient.

[0010] Preferably, the method for controlling electromagnetic losses of a power transformer in a smart grid environment further includes: obtaining a cooling parameter adjustment space of a forced oil cooling device, and randomly generating a number of initial cooling parameter sequences within the cooling parameter adjustment space, wherein the cooling parameters include oil pump flow and fan speed; performing cooling performance analysis based on the number of initial cooling parameter sequences, and outputting a number of cooling efficiency sequences; based on the number of cooling efficiency sequences, screening the number of initial cooling parameter sequences with the constraint of satisfying the predicted cooling load sequence, and obtaining a number of qualified cooling parameter sequences; with minimizing energy consumption and oil temperature fluctuations as the optimization goal, performing optimization based on the multiple qualified cooling parameter sequences, and outputting an optimal cooling parameter sequence.

[0011] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment further includes: performing energy consumption simulation and oil temperature simulation according to the multiple qualified cooling parameter sequences, and outputting multiple energy consumption values ​​and multiple oil temperature sequences; performing temperature volatility analysis according to the multiple oil temperature sequences, and obtaining multiple oil temperature fluctuations, wherein the oil temperature fluctuation is the ratio of the oil temperature standard deviation to the oil temperature mean; taking minimizing energy consumption and oil temperature fluctuation as the optimization goal, determining multiple fitnesses based on the multiple energy consumption values ​​and multiple oil temperature fluctuations, wherein the fitnesses are negatively correlated with the energy consumption values ​​and the oil temperature fluctuations; performing optimization based on the multiple fitnesses and multiple qualified cooling parameter sequences, and outputting the optimal cooling parameter sequence.

[0012] Preferably, the method for controlling electromagnetic losses of power transformers in a smart grid environment also includes: setting a qualified cooling parameter sequence as an initial solution, and arranging the multiple qualified cooling parameter sequences from large to small according to the fitness based on the multiple fitnesses to obtain an initial solution sequence; setting the first solution of the initial solution sequence as an optimal solution, and the other solutions as inferior solutions, and adjusting the multiple inferior solutions according to a preset optimization step size with the optimal solution as the direction to obtain multiple updated inferior solutions; sorting the optimal solution and the multiple updated inferior solutions from large to small according to the fitness to obtain an updated solution sequence, and eliminating a preset proportion of tail solutions in the updated solution sequence, and using the cooling parameter adjustment space for equal value supplementation, wherein the preset proportion is less than 5%; performing iterative optimization until a preset number of convergences is reached, and outputting the optimal solution in the current updated solution sequence as the optimal cooling parameter sequence.

[0013] In a second aspect, the present invention further provides an electromagnetic loss control device for an electric transformer in a smart grid environment, which is used to execute a method for controlling electromagnetic loss of an electric transformer in a smart grid environment as described in the first aspect, and includes: a power load prediction module, which is used to analyze and obtain a predicted power load sequence of the transformer in a preset future time zone, and a spatial temperature sequence of the area where the transformer is located; a cooling demand prediction module, which is used to predict the cooling demand of the transformer based on the predicted power load sequence and the spatial temperature sequence, and output a predicted cooling load sequence; a cooling parameter optimization module, which is used to optimize the cooling parameters of the forced oil cooling device with the satisfaction of the predicted cooling load sequence as a constraint and the minimization of energy consumption and oil temperature fluctuation as the optimization goal, and output an optimal cooling parameter sequence; a transformer heat dissipation control module, which is used to control the forced oil cooling device to dissipate heat for the transformer according to the optimal cooling parameter sequence in the preset future time zone.

[0014] The embodiments of the present invention include the following advantages: By analyzing and obtaining the predicted power load sequence of the transformer within a preset future time zone, as well as the spatial temperature sequence of the area in which the transformer is located, the cooling demand of the transformer is then predicted based on the predicted power load sequence and spatial temperature sequence, and a predicted cooling load sequence is output. Then, with the predicted cooling load sequence as a constraint and minimizing energy consumption and oil temperature fluctuations as the optimization goal, the cooling parameters of the forced oil cooling system are optimized, and an optimal cooling parameter sequence is output. Finally, within the preset future time zone, the forced oil cooling system is controlled according to the optimal cooling parameter sequence to dissipate heat from the transformer. In other words, by predicting the future cooling load demand and with the cooling load sequence as a constraint, an optimization model is constructed with the goal of minimizing energy consumption and oil temperature fluctuations, and the control parameters of the forced oil cooling system are optimized to obtain the optimal cooling parameter sequence to control the operation of the cooling system, thereby achieving precise and efficient cooling control, effectively reducing electromagnetic losses, and improving the operating efficiency and stability of the transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flowchart of the steps of a method for controlling electromagnetic loss of a power transformer in a smart grid environment according to the present invention; Figure 2 This is a structural schematic diagram of a power transformer electromagnetic loss control device in a smart grid environment according to the present invention.

[0016] Description of reference numerals: Power load prediction module 11, cooling demand prediction module 12, cooling parameter optimization module 13, transformer heat dissipation control module 14. DETAILED DESCRIPTION

[0017] The present invention provides a method and device for controlling electromagnetic losses in power transformers in a smart grid environment, addressing the technical problem that conventional transformer cooling system control methods lack the ability to dynamically adjust cooling strategies, resulting in abnormal temperature rises under high load or high temperature environments, significantly increasing transformer core and winding losses and leading to high electromagnetic losses. By predicting future cooling load requirements and constraining the requirements to meet this cooling load sequence, an optimization model is constructed with the goal of minimizing energy consumption and oil temperature fluctuations. The control parameters of the forced oil cooling system are optimized to obtain the optimal cooling parameter sequence to control the cooling system's operation, thereby achieving precise and efficient cooling control, effectively reducing electromagnetic losses, and improving the operating efficiency and stability of the transformer.

[0018] Below, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the accompanying drawings.

[0019] For example, see the attached Figure 1 The present invention provides a method for controlling electromagnetic loss of a power transformer in a smart grid environment, which is applied to a device for controlling electromagnetic loss of a power transformer in a smart grid environment, and specifically comprises the following steps: S10: Analyze and obtain the predicted power load sequence of the transformer in the preset future time zone, and the spatial temperature sequence of the area where the transformer is located.

[0020] Furthermore, step S10 of the present invention further includes: S11: Monitor and obtain the power load of the transformer at P consecutive monitoring time points in a historical period to obtain a historical power load sequence; S12: Monitor and obtain the load fluctuation correlation data of the transformer in a preset future time zone based on a preset load fluctuation correlation factor.

[0021] Specifically, first, the number of continuous monitoring time points is set to P, where P is a positive integer not less than 5 to ensure that the sequence data has a certain time span and fluctuation representativeness; then, the power load of the transformer at P consecutive monitoring time points in the historical period is monitored and obtained. For example, if the system collects load data every 15 minutes, then P equals 12, which means that the power load changes in the past 3 hours are obtained. By collecting indicators such as power or current at P time points, a complete historical power load sequence is constructed. This sequence will be used in subsequent modeling as one of the inputs of the prediction model to reflect the load evolution trend and periodic characteristics of the transformer.

[0022] On the other hand, a preset load fluctuation correlation factor is obtained. The load fluctuation correlation factor refers to an external variable that is highly correlated with the transformer load change in statistical or machine learning modeling, such as the operation plan or electricity consumption characteristics of the connected load side (such as the start and stop schedule of industrial equipment, electricity price trends), weather information (such as temperature, humidity, sunshine intensity), etc.; then, according to the time range within the set future prediction window (for example, the next 1 to 4 hours), the time series data of the above correlation factors are collected to form a load fluctuation correlation data set to assist in modeling and analyzing the potential load trend changes of the transformer in the future time zone.

[0023] By integrating historical operating data and external fluctuation correlation data, an accurate input basis is provided for subsequent cooling load prediction and optimization control, thereby improving the foresight and scientific nature of the overall control strategy.

[0024] S13: According to the historical power load sequence and load fluctuation correlation data, a predicted power load sequence of the transformer in a preset future time zone is predicted and obtained.

[0025] Furthermore, step S13 of the present invention further includes: S131: Taking the historical period as a constraint, collect the sample power load sequence set and the sample load fluctuation associated data set of the transformer in the same period, and obtain the historical power load sequence of the transformer in the historical preset time zone as the sample predicted power load sequence set, wherein the historical period and the historical preset time zone are continuous time periods, and the time interval of the historical preset time zone is the same as the preset future time zone; S132: Using the sample power load sequence set and the sample load fluctuation associated data set as input, and using the sample predicted power load sequence set as supervision, train the long short-term memory network until convergence to obtain a power load forecasting model; S133: Input the historical power load sequence and load fluctuation associated data into the power load forecasting model, and output the predicted power load sequence.

[0026] Specifically, first, using the historical time period as a constraint, a set of transformer sample power load series and a sample load fluctuation correlation dataset are collected within the same time period. The historical time period refers to a continuous time period selected by the system for collecting multiple sample data. This time period can cover the past hours, days, or even weeks. The specific length is determined based on the training requirements of the prediction model and the actual data distribution. Within this historical time period, rolling sampling is performed with a fixed step size to extract multiple time-segment samples. Each sequence in the sample power load series represents the load change trajectory of the transformer within a time window. Each record in the sample load fluctuation correlation dataset contains correlation factors (such as temperature, industrial load status, etc.) within the same time period as the corresponding load series. Next, for each sample, a historical preset time zone with the same length as the future prediction target time zone is specified, which immediately follows the sampling time period. For example, if the model predicts a future load series of four points every 15 minutes within an hour, the historical preset time zone for each sample is also the hour immediately following the sample. The historical power load series of the transformer within the historical preset time zone is then obtained as the sample prediction power load series set.

[0027] Next, a long short-term memory (LSTM) network is trained using the sample power load sequence set and the sample load fluctuation association dataset as input, and the sample predicted power load sequence set as supervision. The LSTM network comprises an input layer, multiple hidden layers, and an output layer. The input layer receives the multidimensional time series described above; the hidden layers utilize multi-layer LSTM units to capture long-term load fluctuation trends; and the output layer is a regression layer (e.g., a Dense layer) that outputs predicted load values ​​for the next T time steps. During training, all sample data is first standardized to eliminate the effects of feature scale differences. Subsequently, forward and backward propagation are performed using the training set, and the model parameters are iteratively updated using an optimizer (e.g., Adam) to minimize the error between the predicted and true values. A validation set is introduced during training to monitor model performance in real time to determine overfitting or convergence. When the maximum number of training rounds (e.g., 100 rounds) is reached, or the validation error falls below a preset threshold, the model is considered converged and the training process is terminated, resulting in a trained power load forecasting model.

[0028] Then, the historical power load sequence and load fluctuation correlation data are used as input features and sent to the trained power load forecasting model. After forward propagation, the model outputs the predicted power load sequence for the future preset time period, reflecting the load trend of the transformer in the future time zone, providing an accurate basis for subsequent cooling demand prediction and control strategy optimization.

[0029] S20: Predicting the cooling demand of the transformer according to the predicted power load sequence and the space temperature sequence, and outputting a predicted cooling load sequence.

[0030] Furthermore, step S20 of the present invention further includes: S21: Based on the historical operation records of the transformer, a sample power load sequence set and a sample space temperature sequence set are collected, and the historical cooling load sequence required under different sample power load sequence and sample space temperature sequence scenarios is obtained to obtain a sample cooling load sequence set; S22: Using the sample power load sequence set, the sample space temperature sequence set and the sample cooling load sequence set, a long short-term memory network is trained until convergence to obtain a cooling demand prediction model; S23: Using the cooling demand prediction model, cooling demand is predicted according to the predicted power load sequence and space temperature sequence, and a predicted cooling load sequence is output.

[0031] Specifically, based on the historical operation records of the transformer, we first collect the power load sequences and corresponding spatial temperature sequences within multiple sample periods to form a sample power load sequence set and a sample spatial temperature sequence set; then, for different load and temperature combination scenarios, combined with the actual cooling needs of the transformer, we further obtain the corresponding cooling load sequences under these historical conditions and construct a sample cooling load sequence set. This sample set reflects the changing pattern of the cooling demand of the transformer under different environments and operating conditions, providing rich and accurate data support for the training and optimization of the cooling demand prediction model.

[0032] Next, the collected sample power load sequence set, sample spatial temperature sequence set, and corresponding sample cooling load sequence set are used as input features and supervision labels to train a long short-term memory network. Multi-layer LSTM units are used to perform deep learning on the time dependency and nonlinear relationship in time series data. During the training process, power load and temperature sequences are input, and the model predicts the future cooling load sequence, which is compared with the actual sample cooling load sequence to calculate the error. The model parameters are continuously adjusted through the backpropagation algorithm and optimizer (such as Adam) to minimize the prediction error. The training iterations continue until the error of the model on the validation set tends to stabilize or meets the preset convergence conditions, thereby obtaining a cooling demand prediction model that can accurately predict future cooling demand.

[0033] Then, the trained cooling demand prediction model is used to take the predicted power load sequence and the corresponding spatial temperature sequence as input. Based on these time series characteristics and combined with the historically learned influence of load and temperature on cooling demand, the model performs in-depth time series analysis and mapping operations. After forward reasoning of the model, it outputs the predicted cooling load sequence in the future preset time zone, reflecting the changing trend of the transformer's cooling demand under different load and ambient temperature conditions, providing a scientific basis for the subsequent parameter optimization and dynamic regulation of the cooling system.

[0034] Furthermore, step S20 of the present invention further includes: S24: Obtain the continuous working time of the forced oil cooling system, and analyze to obtain a continuous working time sequence at the same time node of the predicted cooling load sequence.

[0035] Specifically, the current continuous operating time of the forced oil cooling system is first obtained—the cumulative operating time from the most recent startup to the current moment. Then, combining each future time node in the predicted cooling load sequence, the estimated continuous operating time corresponding to each time node is calculated. This time node is equal to the current continuous operating time plus the time interval from the current moment to that time node, resulting in a continuous operating time sequence. Analyzing this continuous operating time sequence accurately reflects the operating load state of the cooling system within the predicted timeframe, providing key time parameter support for cooling performance degradation assessment and control strategy optimization.

[0036] S25: performing cooling performance degradation analysis on the forced oil cooling system, and constructing a duration-performance degradation comparison table.

[0037] Furthermore, step S25 of the present invention further includes: S251: Taking the attribute characteristics of the forced oil cooling system as constraints, retrieve the historical operation logs of similar forced oil cooling systems, collect the sample continuous working time set, and obtain the performance attenuation ratio corresponding to different sample continuous working times as the sample performance attenuation coefficient to obtain the sample performance attenuation coefficient set; S252: Based on the sample continuous working time set, determine multiple continuous working time intervals, and randomly select the first continuous working time interval to extract the first sample performance attenuation coefficient set covered by the first continuous working time interval; S253: Perform anomaly elimination and mean calculation on the first sample performance attenuation coefficient set to obtain the mean of the first sample performance attenuation coefficient; S254: Based on the mapping relationship between the first continuous working time interval and the first sample performance attenuation coefficient mean, construct a time-performance attenuation comparison table.

[0038] Specifically, the method first retrieves historical operation logs of similar or similar forced oil cooling systems from a database based on their specific attributes (such as model, capacity, and operating environment). This data is then collected to determine their continuous operating duration under different operating conditions, forming a sample continuous operating duration set. Furthermore, the method combines actual maintenance and inspection records to determine the performance degradation ratio of these samples under corresponding continuous operating durations, defining this ratio as the sample performance degradation coefficient. By aggregating the performance degradation coefficients of multiple samples, a sample performance degradation coefficient set is constructed, providing data support for subsequent cooling system performance evaluation and degradation prediction based on continuous operating duration.

[0039] Next, the sample continuous operating duration set is divided into multiple consecutive operating duration intervals. For example, the continuous operating duration is divided into intervals of 0 to 100 hours, 101 to 200 hours, and 201 to 300 hours, for every 100 hours. One of these intervals is then randomly selected as the first continuous operating duration interval, such as the 101 to 200 hour interval. The performance decay coefficients of all samples corresponding to the continuous operating duration within this interval are then extracted from the sample performance decay coefficient set to form the first sample performance decay coefficient set. This first sample performance decay coefficient set is then first detected for outliers. Statistical methods (such as the standard deviation method) are used to identify and remove anomalous data points that significantly deviate from the overall data distribution to ensure data accuracy and representativeness. The average of the first sample performance decay coefficients after removing the anomalies is then calculated to obtain the mean value. This mean value reflects the typical level of performance decay of the forced oil cooling system within the selected continuous operating duration interval, providing a reliable statistical basis for subsequent performance evaluation and prediction.

[0040] Based on the aforementioned calculation method for the first continuous operating time interval and its corresponding sample performance decay coefficient mean, a similar analysis is performed on the remaining continuous operating time intervals, removing outliers and calculating the mean performance decay coefficient for each interval. Each continuous operating time interval is then mapped one-to-one with the corresponding mean performance decay value, constructing a complete duration-performance decay comparison table. This table intuitively reflects the performance decay patterns of the forced oil cooling system under different operating time periods, facilitating subsequent performance evaluation and prediction based on continuous operating time.

[0041] S26: Based on the duration-performance attenuation comparison table, a performance attenuation coefficient sequence is obtained by matching the continuous working duration sequence; S27: Compensating and updating the predicted cooling load sequence according to the performance attenuation coefficient sequence.

[0042] Specifically, first, based on the established duration-performance degradation comparison table, the operating time corresponding to each time point in the continuous operating time sequence is matched with the duration interval in the comparison table to obtain the corresponding performance degradation coefficient, forming a performance degradation coefficient sequence that corresponds one-to-one with the continuous operating time sequence. This sequence reflects the degree of degradation of the cooling system performance due to the increase in operating time at different time points. Next, the predicted cooling load value is adjusted according to the performance degradation coefficient to correct the cooling capacity reduction caused by the decline in cooling system performance. For example, the product of the performance degradation coefficient and the predicted cooling load is used as the updated predicted cooling load to obtain the predicted cooling load sequence. This allows the updated cooling load sequence to more accurately reflect the actual cooling demand, improving the accuracy and effectiveness of cooling control.

[0043] S30: Optimizing the cooling parameters of the forced oil cooling system with the predicted cooling load sequence as a constraint and minimizing energy consumption and oil temperature fluctuation as an optimization goal, and outputting an optimal cooling parameter sequence.

[0044] Furthermore, step S30 of the present invention further includes: S31: Acquire a cooling parameter adjustment space of a forced oil cooling system, and randomly generate a number of initial cooling parameter sequences within the cooling parameter adjustment space, wherein the cooling parameters include oil pump flow and fan speed; S32: Perform cooling performance analysis based on the several initial cooling parameter sequences, and output a number of cooling efficiency sequences; S33: Based on the several cooling efficiency sequences, and subject to the constraint of satisfying the predicted cooling load sequence, screen the several initial cooling parameter sequences to obtain a plurality of qualified cooling parameter sequences.

[0045] Specifically, the range of adjustable cooling parameters in the forced oil cooling system, namely the cooling parameter adjustment space, is first determined. This mainly includes the upper and lower limits of the oil pump flow and fan speed and their variation range. Then, based on this adjustment space, several groups of initial cooling parameter sequences are randomly generated. Each sequence consists of oil pump flow and fan speed values ​​randomly selected within the allowable range, covering different cooling strategy combinations. These initial parameter sequences will serve as the starting point of the optimization algorithm for subsequent optimization of cooling control parameters and performance evaluation.

[0046] Next, based on the several initial cooling parameter sequences, a cooling performance analysis is performed on each set of parameters in a simulated or actual operating environment, evaluating their cooling effects under different load and temperature conditions. The corresponding cooling efficiency sequence is then calculated. This cooling efficiency sequence reflects the dynamic performance of the cooling capacity and energy efficiency achieved using that parameter sequence over time. Based on these cooling efficiency sequences, the initial cooling parameter sequences are then screened, with the constraint of meeting the cooling requirements of the predicted cooling load sequence as the constraint. Parameter sequences that ensure cooling effects meet or exceed the predicted load requirements and exhibit high cooling efficiency are then identified as qualified cooling parameter sequences, providing effective alternatives for subsequent optimization and actual control.

[0047] S34: Taking minimizing energy consumption and oil temperature fluctuation as the optimization goal, optimizing is performed based on the multiple qualified cooling parameter sequences, and an optimal cooling parameter sequence is output.

[0048] Furthermore, step S34 of the present invention further includes: S341: Perform energy consumption simulation and oil temperature simulation according to the multiple qualified cooling parameter sequences, and output multiple energy consumption values ​​and multiple oil temperature sequences; S342: Perform temperature fluctuation analysis according to the multiple oil temperature sequences, and obtain multiple oil temperature fluctuations, wherein the oil temperature fluctuation is the ratio of the oil temperature standard deviation to the oil temperature mean; S343: With minimizing energy consumption and oil temperature fluctuation as the optimization goal, determine multiple fitnesses based on the multiple energy consumption values ​​and multiple oil temperature fluctuation evaluations, wherein the fitness is negatively correlated with the energy consumption value and the oil temperature fluctuation.

[0049] Specifically, for each qualified cooling parameter sequence, energy consumption and oil temperature simulations are first performed. The energy consumption simulation evaluates the energy consumption of equipment such as the oil pump and fan under that parameter configuration, generating corresponding energy consumption values. The oil temperature simulation, based on the cooling effect and system thermal balance model, predicts the oil temperature changes at different time points and generates corresponding oil temperature sequences. These simulation results provide a comprehensive assessment of the energy efficiency and temperature control performance of each qualified parameter sequence, providing a scientific basis for further optimization. Next, a temperature volatility analysis is performed on each oil temperature sequence to assess the stability of the oil temperature throughout the operating cycle. This is done by calculating the standard deviation and mean of the temperature sequence and dividing them to obtain the oil temperature fluctuation index. This index reflects the relative degree of oil temperature variation. Smaller values ​​indicate more stable temperature and smaller fluctuations, while larger values ​​indicate greater temperature fluctuations and unstable cooling system control. Multiple oil temperature fluctuation indexes are generated. By comparing the oil temperature fluctuation indexes under multiple parameter configurations, cooling strategies with superior temperature control performance can be identified.

[0050] Then, with minimizing energy consumption and oil temperature fluctuation as the optimization goal, a fitness evaluation model is constructed. For each qualified cooling parameter sequence, the fitness value is calculated in combination with its corresponding energy consumption value and oil temperature fluctuation. The fitness is negatively correlated with the energy consumption value and temperature fluctuation, that is, the lower the energy consumption and fluctuation, the higher the fitness, and vice versa. This fitness evaluation mechanism helps to select the control strategy with the best comprehensive performance from multiple candidate cooling parameter schemes, which is used to guide the actual operation of the forced oil cooling system.

[0051] S344: Optimizing based on the multiple fitness levels and multiple qualified cooling parameter sequences, and outputting an optimal cooling parameter sequence.

[0052] Furthermore, step S344 of the present invention further includes: S3441: Set the qualified cooling parameter sequence as the initial solution, and based on the multiple fitnesses, arrange the multiple qualified cooling parameter sequences in descending order of fitness to obtain an initial solution sequence; S3442: Set the first solution of the initial solution sequence as the optimal solution, and the other solutions as inferior solutions, and adjust the multiple inferior solutions according to the preset optimization step size with the optimal solution as the direction to obtain multiple updated inferior solutions; S3443: Sort the optimal solution and the multiple updated inferior solutions in descending order of fitness to obtain an updated solution sequence, and eliminate a preset proportion of tail solutions in the updated solution sequence, and use the cooling parameter adjustment space for equal supplementation, wherein the preset proportion is less than 5%; S3444: Perform iterative optimization until the preset number of convergences is reached, and output the optimal solution in the current updated solution sequence as the optimal cooling parameter sequence.

[0053] Specifically, first, the multiple qualified cooling parameter sequences obtained by the previous screening are used as the initial solution set, and based on the fitness values ​​corresponding to each sequence, they are sorted from large to small according to fitness to construct an initial solution sequence. This sorting process helps to identify the cooling control strategy with the best performance in the current parameter configuration and provides an ordered reference for the subsequent optimization process. Subsequently, the cooling parameter sequence with the highest fitness in the sorting is set as the optimal solution, and the remaining sequences are set as inferior solutions. With the optimal solution as the target direction, referring to the difference between each inferior solution and the optimal solution in the parameter space, combined with the preset optimization step size, the parameters of each inferior solution are fine-tuned to make it closer to the optimal solution. This process simulates heuristic search or local improvement strategy, aiming to further improve the performance of the inferior solution, gradually approach the optimal control parameter configuration, and obtain multiple updated inferior solutions.

[0054] Next, the optimized optimal solution is combined with multiple updated inferior solutions to form an updated solution set. Each solution is sorted from high to low according to its corresponding fitness value to generate an updated solution sequence. To continuously improve the overall solution set quality, the tail solutions in this sequence (i.e., those with the lowest fitness) are eliminated at a preset threshold (typically less than 5%) to remove the worst-performing cooling parameter configurations. Subsequently, new solutions are randomly generated within the cooling parameter adjustment space, equal to the number of eliminated solutions, to ensure a stable solution set size and provide diversity and exploration capabilities for the next round of optimization, thereby enhancing overall optimization efficiency. Then, based on the above-mentioned sorting, elimination and supplementation mechanism, iterative optimization is continuously performed. In each round of iteration, the current optimal solution is used as a guide to fine-tune the parameters of the inferior solution, and the solution sequence is updated according to the fitness. At the same time, low-quality solutions are eliminated and new solutions are supplemented. This process is repeated until the preset number of convergences is reached or other convergence conditions are met (such as the fitness change of the optimal solution tends to be stable). Finally, the solution with the highest fitness in the updated solution sequence at the end of the iteration is determined as the optimal cooling parameter sequence, which is used to guide the forced oil cooling system to achieve precise and efficient heat dissipation control, thereby effectively reducing electromagnetic losses and improving the operating efficiency and reliability of the transformer.

[0055] S40: In the preset future time zone, controlling the forced oil cooling system according to the optimal cooling parameter sequence to dissipate heat from the transformer.

[0056] Specifically, within the predetermined future time zone, the forced oil cooling system is dynamically controlled based on the optimal cooling parameter sequence obtained through prior optimization. This involves dynamically adjusting the oil pump flow and fan speed settings within this parameter sequence at each time point, precisely adjusting the cooling device's operating state to match the predicted cooling load. This control approach not only ensures the transformer maintains a reasonable operating temperature under future load and ambient temperature conditions, but also avoids energy waste and equipment aging caused by overcooling or undercooling, thereby maximizing heat dissipation efficiency and minimizing electromagnetic losses, improving the overall stability and economic efficiency of the transformer's operation.

[0057] In summary, the method for controlling electromagnetic loss of a power transformer in a smart grid environment provided by the present invention has the following technical effects: By analyzing and obtaining the predicted power load sequence of the transformer within a preset future time zone, as well as the spatial temperature sequence of the area in which the transformer is located, the cooling demand of the transformer is then predicted based on the predicted power load sequence and spatial temperature sequence, and a predicted cooling load sequence is output. Then, with the predicted cooling load sequence as a constraint and minimizing energy consumption and oil temperature fluctuations as the optimization goal, the cooling parameters of the forced oil cooling system are optimized, and an optimal cooling parameter sequence is output. Finally, within the preset future time zone, the forced oil cooling system is controlled according to the optimal cooling parameter sequence to dissipate heat from the transformer. In other words, by predicting the future cooling load demand and with the cooling load sequence as a constraint, an optimization model is constructed with the goal of minimizing energy consumption and oil temperature fluctuations, and the control parameters of the forced oil cooling system are optimized to obtain the optimal cooling parameter sequence to control the operation of the cooling system, thereby achieving precise and efficient cooling control, effectively reducing electromagnetic losses, and improving the operating efficiency and stability of the transformer.

[0058] In the second embodiment, based on the same inventive concept as the method for controlling electromagnetic loss of a power transformer in a smart grid environment in the above embodiment, the present invention also provides an electromagnetic loss control device for a power transformer in a smart grid environment, as shown in the attached FIG. Figure 2 ,include: The power load prediction module 11 is used to analyze and obtain the predicted power load sequence of the transformer in a preset future time zone, as well as the spatial temperature sequence of the area where the transformer is located; the cooling demand prediction module 12 is used to predict the cooling demand of the transformer based on the predicted power load sequence and the spatial temperature sequence, and output a predicted cooling load sequence; the cooling parameter optimization module 13 is used to optimize the cooling parameters of the forced oil cooling system with the satisfaction of the predicted cooling load sequence as a constraint and the minimization of energy consumption and oil temperature fluctuation as the optimization goal, and output an optimal cooling parameter sequence; the transformer heat dissipation control module 14 is used to control the forced oil cooling system to dissipate heat for the transformer according to the optimal cooling parameter sequence in the preset future time zone.

[0059] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: monitor and obtain the power load of the transformer at P consecutive monitoring time points in a historical period to obtain a historical power load sequence; monitor and obtain load fluctuation correlation data of the transformer in a preset future time zone based on a preset load fluctuation correlation factor; and predict and obtain a predicted power load sequence of the transformer in a preset future time zone based on the historical power load sequence and the load fluctuation correlation data.

[0060] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: with the historical time period as a constraint, collect the sample power load sequence set and the sample load fluctuation associated data set of the transformer in the same time period, and obtain the historical power load sequence of the transformer in the historical preset time zone as the sample predicted power load sequence set, wherein the historical time period and the historical preset time zone are continuous time periods, and the time interval of the historical preset time zone is the same as the preset future time zone; use the sample power load sequence set and the sample load fluctuation associated data set as input, use the sample predicted power load sequence set as supervision, train the long short-term memory network until convergence, and obtain a power load forecasting model; input the historical power load sequence and load fluctuation associated data into the power load forecasting model, and output the predicted power load sequence.

[0061] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: collect sample power load sequence sets and sample space temperature sequence sets based on the historical operation records of the transformer, and obtain the historical cooling load sequences required under different sample power load sequence and sample space temperature sequence scenarios to obtain a sample cooling load sequence set; use the sample power load sequence set, sample space temperature sequence set and sample cooling load sequence set to train a long short-term memory network until convergence to obtain a cooling demand prediction model; use the cooling demand prediction model to predict cooling demand based on the predicted power load sequence and space temperature sequence, and output a predicted cooling load sequence.

[0062] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: obtain the continuous working time of the forced oil cooling system, and analyze to obtain the continuous working time sequence at the same time node of the predicted cooling load sequence; perform cooling performance attenuation analysis on the forced oil cooling system, and construct a time-performance attenuation comparison table; based on the time-performance attenuation comparison table, obtain a performance attenuation coefficient sequence according to the continuous working time sequence; and compensate and update the predicted cooling load sequence according to the performance attenuation coefficient sequence.

[0063] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: use the attribute characteristics of the forced oil cooling system as constraints, retrieve historical operation logs of similar forced oil cooling systems, collect sample continuous working time sets, and obtain the performance attenuation ratio corresponding to different sample continuous working times as the sample performance attenuation coefficient to obtain a sample performance attenuation coefficient set; determine multiple continuous working time intervals based on the sample continuous working time set, and randomly select a first continuous working time interval to extract a first sample performance attenuation coefficient set covered by the first continuous working time interval; perform anomaly elimination and mean calculation on the first sample performance attenuation coefficient set to obtain the mean of the first sample performance attenuation coefficient; and construct a time-performance attenuation comparison table based on the mapping relationship between the first continuous working time interval and the mean of the first sample performance attenuation coefficient.

[0064] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: obtain a cooling parameter adjustment space of a forced oil cooling system, and randomly generate a number of initial cooling parameter sequences within the cooling parameter adjustment space, wherein the cooling parameters include oil pump flow and fan speed; perform cooling performance analysis based on the number of initial cooling parameter sequences, and output a number of cooling efficiency sequences; based on the number of cooling efficiency sequences, screen the number of initial cooling parameter sequences with the constraint of satisfying the predicted cooling load sequence to obtain a number of qualified cooling parameter sequences; with minimizing energy consumption and oil temperature fluctuations as the optimization goal, perform optimization based on the multiple qualified cooling parameter sequences, and output an optimal cooling parameter sequence.

[0065] Furthermore, the electromagnetic loss control device for power transformers in the smart grid environment is also used to: perform energy consumption simulation and oil temperature simulation according to the multiple qualified cooling parameter sequences, and output multiple energy consumption values ​​and multiple oil temperature sequences; perform temperature volatility analysis according to the multiple oil temperature sequences, and obtain multiple oil temperature fluctuations, wherein the oil temperature fluctuation is the ratio of the oil temperature standard deviation to the oil temperature mean; with minimizing energy consumption and oil temperature fluctuation as the optimization goal, determine multiple fitnesses based on the multiple energy consumption values ​​and multiple oil temperature fluctuations, wherein the fitnesses are negatively correlated with the energy consumption values ​​and the oil temperature fluctuations; perform optimization based on the multiple fitnesses and multiple qualified cooling parameter sequences, and output the optimal cooling parameter sequence.

[0066] Furthermore, the power transformer electromagnetic loss control device in the smart grid environment is also used to: set a qualified cooling parameter sequence as an initial solution, and based on the multiple fitnesses, arrange the multiple qualified cooling parameter sequences in descending order of fitness to obtain an initial solution sequence; set the first solution of the initial solution sequence as an optimal solution, and the other solutions as inferior solutions, and adjust the multiple inferior solutions according to a preset optimization step size with the optimal solution as the direction to obtain multiple updated inferior solutions; sort the optimal solution and the multiple updated inferior solutions in descending order of fitness to obtain an updated solution sequence, and eliminate a preset proportion of tail solutions in the updated solution sequence, and use the cooling parameter adjustment space for equal value supplementation, wherein the preset proportion is less than 5%; perform iterative optimization until a preset number of convergences is reached, and output the optimal solution in the current updated solution sequence as the optimal cooling parameter sequence.

[0067] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The method for controlling electromagnetic loss of a power transformer in a smart grid environment and the specific examples in the aforementioned embodiment 1 are also applicable to the electromagnetic loss control device for a power transformer in a smart grid environment in this embodiment. Through the aforementioned detailed description of the electromagnetic loss control method for a power transformer in a smart grid environment, those skilled in the art can clearly understand the electromagnetic loss control device for a power transformer in a smart grid environment in this embodiment. Therefore, for the sake of brevity of the specification, it will not be described in detail here. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method description.

[0068] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0069] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is intended to include these modifications and variations.

Claims

1. A method for controlling electromagnetic losses of power transformers in a smart grid environment, characterized in that: Methods include: Analyze and obtain the predicted power load sequence of the transformer in the preset future time zone, as well as the spatial temperature sequence of the area where the transformer is located; Predicting transformer cooling demand based on the predicted power load sequence and the spatial temperature sequence, and outputting a predicted cooling load sequence; With the predicted cooling load sequence as a constraint and the minimization of energy consumption and oil temperature fluctuation as the optimization goal, the cooling parameters of the forced oil cooling device are optimized and the optimal cooling parameter sequence is output; In the preset future time zone, the forced oil cooling device is controlled according to the optimal cooling parameter sequence to dissipate heat from the transformer.

2. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 1, wherein: Analyze and obtain the predicted power load sequence of transformers in a preset future time zone, including: Monitor and obtain the power load of the transformer at P consecutive monitoring time points in a historical period to obtain a historical power load sequence; According to the preset load fluctuation correlation factor, monitor and obtain the load fluctuation correlation data of the transformer in the preset future time zone; According to the historical power load sequence and load fluctuation correlation data, a predicted power load sequence of the transformer in a preset future time zone is predicted and obtained.

3. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 2, wherein: According to the historical power load sequence and load fluctuation correlation data, a predicted power load sequence of the transformer in a preset future time zone is predicted and obtained, including: Taking the historical period as a constraint, collecting a sample power load sequence set and a sample load fluctuation associated data set of the transformer within the same period, and obtaining the historical power load sequence of the transformer within a historical preset time zone as a sample predicted power load sequence set, wherein the historical period and the historical preset time zone are continuous time periods, and the time interval of the historical preset time zone is the same as the preset future time zone; Using the sample power load sequence set and the sample load fluctuation associated data set as input, using the sample predicted power load sequence set as supervision, training the long short-term memory network until convergence, and obtaining a power load forecasting model; The historical power load sequence and load fluctuation associated data are input into the power load forecasting model, and a forecasted power load sequence is output.

4. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 1, wherein: The cooling demand of the transformer is predicted according to the predicted power load sequence and the space temperature sequence, and the predicted cooling load sequence is output, including: According to the historical operation records of the transformer, a sample power load sequence set and a sample space temperature sequence set are collected, and the historical cooling load sequences required under different sample power load sequence and sample space temperature sequence scenarios are obtained to obtain a sample cooling load sequence set; Using the sample power load sequence set, the sample space temperature sequence set, and the sample cooling load sequence set, a long short-term memory network is trained until convergence to obtain a cooling demand prediction model; The cooling demand prediction model is used to perform cooling demand prediction according to the predicted power load sequence and the space temperature sequence, and a predicted cooling load sequence is output.

5. The method for controlling electromagnetic loss of power transformers in a smart grid environment according to claim 1, wherein: The output forecast cooling load sequence further includes: Obtaining the continuous working time of the forced oil cooling device, and analyzing to obtain a continuous working time sequence at the same time node of the predicted cooling load sequence; Perform cooling performance degradation analysis on the forced oil cooling device and construct a duration-performance degradation comparison table; Based on the duration-performance attenuation comparison table, a performance attenuation coefficient sequence is obtained by matching the continuous working duration sequence; The predicted cooling load sequence is compensated and updated according to the performance degradation coefficient sequence.

6. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 5, characterized in that: Perform cooling performance degradation analysis on the forced oil cooling device and construct a duration-performance degradation comparison table, including: Using the attribute characteristics of the forced oil cooling device as a constraint, historical operation logs of similar forced oil cooling devices are retrieved to collect a set of sample continuous working durations. The performance attenuation ratios corresponding to different sample continuous working durations are set as sample performance attenuation coefficients to obtain a set of sample performance attenuation coefficients. Determine a plurality of continuous working duration intervals based on the sample continuous working duration set, randomly select a first continuous working duration interval, and extract a first sample performance attenuation coefficient set covered by the first continuous working duration interval; Performing outlier elimination and mean calculation on the first sample performance decay coefficient set to obtain a first sample performance decay coefficient mean; Based on the mapping relationship between the first continuous working time interval and the first sample performance attenuation coefficient average, a time-performance attenuation comparison table is constructed.

7. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 1, wherein: With the predicted cooling load sequence as a constraint and minimizing energy consumption and oil temperature fluctuation as the optimization goal, the cooling parameters of the forced oil cooling device are optimized, and the optimal cooling parameter sequence is output, including: Acquiring a cooling parameter adjustment space of a forced oil cooling device, and randomly generating a plurality of initial cooling parameter sequences within the cooling parameter adjustment space, wherein the cooling parameters include an oil pump flow rate and a fan speed; Performing cooling performance analysis based on the multiple initial cooling parameter sequences and outputting multiple cooling efficiency sequences; Based on the plurality of cooling efficiency sequences, the plurality of initial cooling parameter sequences are screened to obtain a plurality of qualified cooling parameter sequences, with the predicted cooling load sequence being a constraint; Taking minimizing energy consumption and oil temperature fluctuation as the optimization goal, optimization is performed based on the multiple qualified cooling parameter sequences, and the optimal cooling parameter sequence is output.

8. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 7, wherein: With minimizing energy consumption and oil temperature fluctuation as the optimization goal, optimization is performed based on the multiple qualified cooling parameter sequences, including: Performing energy consumption simulation and oil temperature simulation respectively according to the plurality of qualified cooling parameter sequences, and outputting a plurality of energy consumption values ​​and a plurality of oil temperature sequences; Performing temperature fluctuation analysis on the multiple oil temperature sequences to obtain multiple oil temperature fluctuations, wherein the oil temperature fluctuation is a ratio of an oil temperature standard deviation to an oil temperature mean; Minimizing energy consumption and oil temperature fluctuation is an optimization goal, and determining a plurality of fitnesses based on the plurality of energy consumption values ​​and the plurality of oil temperature fluctuations, wherein the fitnesses are negatively correlated with the energy consumption values ​​and the oil temperature fluctuations; Optimization is performed based on the multiple fitness levels and multiple qualified cooling parameter sequences, and an optimal cooling parameter sequence is output.

9. The method for controlling electromagnetic loss of a power transformer in a smart grid environment according to claim 8, characterized in that: Optimizing based on the multiple fitnesses and multiple qualified cooling parameter sequences, and outputting an optimal cooling parameter sequence, including: The qualified cooling parameter sequence is set as an initial solution, and based on the multiple fitnesses, the multiple qualified cooling parameter sequences are arranged in descending order of fitness to obtain an initial solution sequence; The first solution of the initial solution sequence is set as the optimal solution, and the other solutions are set as inferior solutions. With the optimal solution as the direction, multiple inferior solutions are adjusted according to the preset optimal step length to obtain multiple updated inferior solutions; Sort the best solution and multiple updated inferior solutions according to fitness from large to small to obtain an updated solution sequence, eliminate a preset proportion of tail solutions in the updated solution sequence, and use the cooling parameter adjustment space to supplement them with equal values, wherein the preset proportion is less than 5%; Iterative optimization is performed until a preset number of convergences is reached, and the best solution in the current updated solution sequence is output as the optimal cooling parameter sequence.

10. A power transformer electromagnetic loss control device in a smart grid environment, characterized in that: The steps for implementing the method for controlling electromagnetic loss of a power transformer in a smart grid environment according to any one of claims 1 to 9 include: The power load forecasting module is used to analyze and obtain the predicted power load sequence of the transformer in a preset future time zone, as well as the spatial temperature sequence of the area where the transformer is located; a cooling demand prediction module, configured to predict the cooling demand of the transformer based on the predicted power load sequence and the space temperature sequence, and output a predicted cooling load sequence; A cooling parameter optimization module is used to optimize the cooling parameters of the forced oil cooling device with the predicted cooling load sequence as a constraint and the minimization of energy consumption and oil temperature fluctuation as the optimization goal, and output an optimal cooling parameter sequence; The transformer heat dissipation control module is used to control the forced oil cooling device to dissipate heat for the transformer according to the optimal cooling parameter sequence within the preset future time zone.

Images