Intelligent power distribution system and method for monitoring, regulating and controlling new energy box transformer substation

Abstract

The invention relates to the technical field of new energy, and discloses an intelligent power distribution system and method for monitoring, regulating and controlling a new energy box transformer substation, and aims to acquire real-time operation data of the box transformer substation through multiple sensors, construct a data feature vector through processing, obtain power distribution optimization data through an intelligent algorithm model, and realize intelligent power distribution. The method can comprehensively monitor data in real time, deeply mine data rules, accurately and flexibly regulate and control power distribution, and can also be integrated with other systems. Compared with a traditional mode, the operation reliability and the power distribution efficiency of the box transformer substation are effectively improved, equipment faults and energy waste are avoided, the operation cost is reduced, effective utilization of new energy is promoted, the effects in the aspects of improving the intelligent level of an electric power system, promoting energy structure transformation and the like are remarkable, and high application value is achieved.

Description

Technical Field

[0001] The present invention relates to the field of new energy technology, and in particular to an intelligent power distribution system and method for monitoring and controlling new energy box transformers. Background Art

[0002] At a time when the new energy industry is booming, the new energy box transformer, as a key device for power conversion and distribution, plays a decisive role in the performance of the entire new energy power generation system due to its operating stability and distribution efficiency. With the continuous increase in the installed capacity of new energy, such as the continuous expansion of wind farms and solar photovoltaic power stations, new energy box transformers are facing more complex and stringent operating environments and requirements, and traditional distribution monitoring and control methods have gradually exposed many drawbacks.

[0003] From the perspective of monitoring data acquisition, traditional means are difficult to obtain key data on the operation of new energy box transformers in a comprehensive and real-time manner. In actual operation, the operating status of new energy box transformers is affected by many factors, including ambient temperature, humidity, light intensity (for photovoltaic box transformers), wind speed (for wind power box transformers), and dynamic changes in power load. In the past, only limited monitoring equipment such as simple electric meters and thermometers were relied on to obtain only some basic electrical parameters, and important information such as partial discharge inside the box transformer and winding temperature distribution could not be effectively collected. Taking a large wind farm as an example, in extreme weather, the rapid change in wind speed will cause the output power of the wind turbine to fluctuate greatly. Since traditional monitoring methods cannot capture the comprehensive operating data of the box transformer in this case in a timely manner, it is impossible to accurately judge the working status of each component inside the box transformer, increasing the risk of equipment failure.

[0004] In terms of data processing and analysis, traditional methods lack depth and effectiveness. Faced with the limited data collected, they often only perform simple numerical statistics and threshold judgments, and are unable to explore the complex associations and potential laws hidden behind the data. For example, judging whether the box transformer is operating normally is based solely on the comparison between the real-time values ​​of voltage and current and the preset thresholds, ignoring the changing trends of these parameters in the time series and the mutual influence between different parameters. In fact, slight fluctuations in voltage may be associated with multiple factors such as changes in current, box transformer load characteristics, and grid harmonics. Traditional processing methods cannot conduct in-depth analysis of these complex relationships, making it difficult to detect potential fault hazards in advance.

[0005] For the power distribution control link, the flexibility and accuracy of the traditional mode are seriously insufficient. Usually, a pre-set fixed power distribution strategy is adopted, which cannot be adjusted intelligently according to the real-time operating status of the box transformer and the dynamic changes in power demand. In the scenario of new energy power generation, the uncertainty of power load is particularly prominent. For example, photovoltaic power generation is affected by light intensity, and the power generation power varies greatly at different times of the day; wind power has frequent power fluctuations due to the unpredictability of wind speed. When the box transformer load suddenly increases or the power generation power changes suddenly, the traditional control method cannot quickly make reasonable power distribution adjustments, which is easy to cause uneven power distribution, reduce energy utilization efficiency, and may also cause equipment overload, overheating and other problems, affecting equipment life and system stability. In addition, with the advancement of the intelligent development trend of new energy power generation systems, more efficient collaborative work and information interaction are needed between various devices. The traditional new energy box transformer monitoring and control system is highly independent, lacks effective communication and integration capabilities with other power equipment and management systems, and is difficult to integrate into the entire intelligent power network. This makes it impossible to fully consider the actual operation of the box transformer when conducting overall dispatch and optimization of the power system, which restricts the improvement of the intelligence level of the entire new energy power generation system. Summary of the invention

[0006] The purpose of the present invention is to provide an intelligent power distribution system and method for monitoring and controlling new energy box transformers, so as to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: an intelligent power distribution method for monitoring and controlling new energy box transformers, the method comprising: Obtain real-time data on the operation of new energy box transformers; Processing the real-time data of the new energy box transformer operation to obtain a data feature vector; According to the data feature vector, the box-type transformer monitoring and control power distribution optimization data is obtained based on the intelligent algorithm model; By using the box-type transformer monitoring and control power distribution optimization data, the intelligent power distribution results of the new energy box-type transformer monitoring and control are obtained.

[0008] Preferably, the real-time data of the new energy box transformer operation is processed to obtain the data feature vector, which includes: Determine the number of data types included in the real-time data of new energy box transformer operation; The calculation formula for constructing the data feature vector according to the number of data types is:

[0009] in, is the data feature vector, is the length of data collection time, is the number of data types, is a real number set, which is an array consisting of real numbers.

[0010] Preferably, the step of obtaining the box-type transformer monitoring and control power distribution optimization data based on the data feature vector and the intelligent algorithm model includes: The data feature vector is subjected to wavelet transformation to obtain an initial feature transformation result, the calculation formula of which is:

[0011] in is the initial feature transformation result, represents the wavelet transform operation; Utilizing the initial feature transformation results, a time series analysis module and a spatial correlation analysis module are respectively constructed based on a deep neural network.

[0012] Preferably, constructing a time series analysis module and a spatial association analysis module based on a deep neural network respectively includes: slicing and reorganizing the initial feature transformation results to construct a time series analysis module; and performing dimensionality reduction and fusion processing on the initial feature transformation results to construct a spatial association analysis module.

[0013] Preferably, the slicing and reorganizing the initial feature transformation results to construct a time series analysis module comprises: Slice the initial feature transformation result according to the time dimension corresponding to the data collection time length to obtain the initial feature slice; Using the initial feature slices, the reorganization process is performed based on the recurrent neural network, and the calculation formula for the time series analysis result is obtained as follows:

[0014] in, is the result of time series analysis. is a recurrent neural network, is the operation of splicing Q slices along a new dimension, where Q is the number of initial feature slices, is the length of each initial feature slice, which is obtained by dividing the data acquisition time length M by the number of initial feature slices Q and rounding up. is the initial feature transformation result In Dimension On slices, the time range is from arrive ; The time series analysis result is used as the output of the time series analysis module.

[0015] Preferably, the step of performing dimensionality reduction and fusion processing on the initial feature transformation results to construct a spatial correlation analysis module comprises: The principal component analysis is used to reduce the dimension of the initial feature transformation results to obtain the feature data after dimension reduction. The calculation formula is:

[0016] in, is the feature data after dimensionality reduction, represents the principal component analysis operation; The feature data after dimensionality reduction is fused with the preset spatial position information to obtain fused feature data. The calculation formula is:

[0017] in, is the preset spatial location information; The fused feature data is convolved to obtain the calculation formula for the spatial correlation analysis result:

[0018] in, is the result of spatial correlation analysis. It is a two-dimensional convolution operation; The spatial correlation analysis result is used as the output of the spatial correlation analysis module.

[0019] Preferably, the use of the box-type transformer monitoring and control power distribution optimization data to obtain the intelligent power distribution results of the new energy box-type transformer monitoring and control includes: According to the distribution optimization data of the box-type transformer monitoring and control, the calculation formula for establishing the intelligent distribution prediction result is:

[0020] in, For intelligent power distribution prediction results, For the output network, is the feature conversion function, and The spatial correlation analysis module and the time series analysis module are respectively The output of the layer, is the number of network layers; According to the intelligent power distribution prediction results and the corresponding new energy box transformer operation real-time data, the calculation formula of the intelligent power distribution error evaluation function is established as follows:

[0021] in, Provide real-time data for the new power supply box. is the intelligent power distribution error evaluation function, is the number of data samples; The calculated value of the intelligent power distribution error evaluation function is used to adaptively adjust the intelligent algorithm model.

[0022] Preferably, the method of adaptively adjusting the intelligent algorithm model by using the calculated value of the intelligent power distribution error evaluation function includes: Setting the error threshold , the intelligent distribution error evaluation function The value of is compared with the error threshold; like , the simulated annealing algorithm is used to adjust the parameters of the intelligent algorithm model; the initial temperature of the simulated annealing algorithm is set to , the temperature drop coefficient is set to ,in 1; In each iteration, a new parameter combination is randomly generated , calculate the intelligent distribution error evaluation function under the new parameter combination , and calculate the error difference ; like , then accept new parameter combinations ;like , then with probability Accepts new parameter combinations, where is the current temperature; after each iteration, the temperature is updated , until the maximum number of iterations is reached or the temperature is lower than the preset termination temperature, .

[0023] Preferably, the method further comprises, in the process of adjusting the parameters of the intelligent algorithm model by using the simulated annealing algorithm, dynamically adjusting the acceptance probability, specifically: Introducing adaptive factors , as the number of iterations k increases, the adaptive factor Updated by the following formula:

[0024] in, is the preset iteration number threshold, Parameters that control the rate of change of the adaptive factor; in calculating the acceptance probability When , the acceptance probability is adjusted to ,in The basic acceptance probability.

[0025] Preferably, an intelligent power distribution system for monitoring and controlling new energy box transformers comprises: Data acquisition module, used to obtain real-time data of new energy box transformer operation; A data processing module, connected to the data acquisition module, for processing the real-time data of the new energy box transformer operation to obtain a data feature vector; An optimization calculation module, connected to the data processing module, for obtaining distribution optimization data for monitoring and regulating the box-type transformer based on the data feature vector and an intelligent algorithm model; The result output module is connected to the optimization calculation module and is used to obtain the intelligent distribution results of the new energy box transformer monitoring and regulation by using the box transformer monitoring and regulation power distribution optimization data.

[0026] Compared with the prior art, the present invention has the following beneficial effects: From the perspective of data monitoring, the present invention improves the comprehensiveness and real-time performance of data acquisition. By deploying various types of sensors on new energy transformers, such as current sensors, voltage sensors, temperature sensors, and partial discharge sensors, it is possible to collect all kinds of key data during the operation of the transformers, covering electrical parameters, temperature information, discharge conditions, etc. Taking a large photovoltaic power station as an example, during periods of drastic changes in sunlight intensity, the system can collect real-time data on the output current, voltage, temperature of key internal parts, and working status of photovoltaic components of the transformer. These rich data provide a solid foundation for accurately grasping the operating status of the transformer, help to promptly discover potential fault hazards, avoid delays in fault diagnosis due to missing data, and significantly improve the reliability of transformer operation.

[0027] In terms of data processing and analysis, the present invention uses intelligent algorithm models to conduct in-depth mining and analysis of large amounts of collected data, which can effectively reveal the complex associations and potential laws behind the data. By constructing a time series analysis module and a spatial correlation analysis module, it is possible not only to accurately capture the changing trend of data in the time dimension, but also to deeply explore the relationship between different parameters in the spatial dimension. For example, when analyzing the box transformer data of a certain wind farm, the time series analysis module can predict the fluctuation trend of the box transformer output power in the future as the wind speed changes; the spatial correlation analysis module can find the correlation between the temperature and electrical parameters at different locations inside the box transformer, and predict in advance the risk of failures that may be caused by local overheating. This in-depth analysis capability provides strong support for the formulation of scientific and reasonable distribution and control strategies, and effectively improves the ability to predict and prevent faults.

[0028] The accuracy and flexibility of power distribution regulation is one of the core advantages of the present invention. Based on the box-type transformer monitoring and control power distribution optimization data obtained by the intelligent algorithm model, the system can quickly make accurate power distribution adjustments according to the real-time operating status of the box-type transformer and the dynamic changes in power demand. In the case of large fluctuations in the power of renewable energy generation, such as a sudden drop in the power of photovoltaic power generation when blocked by clouds, or a large change in the power of wind power when the wind speed suddenly changes, the intelligent power distribution system of the present invention can adjust the power distribution parameters in real time to ensure the balance and stability of power distribution. By precisely controlling the output power and voltage of the box-type transformer, problems such as equipment overload and overheating are effectively avoided, energy utilization efficiency is improved, equipment loss is reduced, and the service life of equipment is extended.

[0029] From the perspective of system integration and intelligent development, the intelligent distribution system of the present invention has good compatibility and scalability, and can achieve efficient communication and integration with other power equipment and management systems. It can seamlessly access the entire intelligent power network and provide accurate box-type transformer operation information for the overall dispatch and optimization of the power system. When the power system is load dispatching, the system can respond to dispatch instructions in a timely manner, adjust the distribution strategy according to its own operating status, and achieve optimal allocation of power resources. This high degree of intelligence and collaborative working ability has effectively promoted the development of new energy power generation systems towards intelligence and efficiency, and improved the operation and management level of the entire power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a working principle diagram of the intelligent power distribution method of the present invention; Figure 2 A diagram showing the steps for building a spatial correlation analysis module; Figure 3 A step chart for obtaining the intelligent power distribution results of new energy box-type transformer monitoring and control by utilizing the box-type transformer monitoring and control power distribution optimization data; Figure 4 A flowchart for adaptively adjusting the intelligent algorithm model using the calculated value of the intelligent power distribution error evaluation function. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] See also Figure 1-4 The present invention provides a technical solution: an intelligent power distribution method for monitoring and controlling new energy box transformers, the method comprising: Obtaining real-time data of new energy transformer operation: By installing various sensors on the new energy transformer, such as current sensors, voltage sensors, temperature sensors, etc., the current, voltage, power, temperature and other data during the transformer operation are collected in real time. For example, in a new energy transformer at a wind farm, the above data is collected every 10 seconds to obtain the real-time status information of the transformer operation.

[0033] Process the real-time data of the new energy box transformer operation to obtain the data feature vector: Determine the number of types of collected data. Assume that the four types of data, current, voltage, power, and temperature, are collected, that is, the number of data types At the same time, set the data collection time length ,like Seconds (i.e. data collected within 1 minute). According to the formula Constructing data feature vector ,in is a real number set, which is an array consisting of the collected real numbers. In this example, It is a 60×4 array, each row represents 4 types of data collected at different times, and each column corresponds to current, voltage, power, and temperature data.

[0034] According to the data feature vector, the distribution optimization data of the box-type transformer monitoring and control is obtained based on the intelligent algorithm model: Perform wavelet transform processing and use the formula Get the initial feature transformation result ,in Represents the wavelet transform operation. Taking the actual data as an example, the above 60×4 data feature vector Perform wavelet transform to obtain the initial feature transformation result , The dimension and form of is determined according to the characteristics of wavelet transform. Then, using The time series analysis module and spatial correlation analysis module are constructed based on deep neural networks.

[0035] Using the distribution optimization data of box-type transformer monitoring and control, the intelligent distribution results of new energy box-type transformer monitoring and control are obtained: Based on the obtained distribution optimization data of box-type transformer monitoring and control, the calculation formula of intelligent distribution prediction results is established:

[0036] in For intelligent power distribution prediction results, For the output network, is the feature conversion function, and The spatial correlation analysis module and the time series analysis module are respectively The output of the layer, is the number of network layers. Assume that the number of network layers ,The intelligent power distribution prediction results are calculated through this formula. At the same time, according to the intelligent power distribution prediction results and the corresponding new energy box transformer operation real-time data, the intelligent power distribution error evaluation function is established

[0037] in Provide real-time data for the new power supply box. is the intelligent power distribution error evaluation function, is the number of data samples. If , calculate the error evaluation function value. Use the calculated value to adaptively adjust the intelligent algorithm model to optimize the power distribution results.

[0038] Embodiment 1:

[0039] This embodiment describes in detail how to determine the number of data types contained in the real-time data of the new energy box transformer operation, and construct a data feature vector based on this, providing a basic data structure for subsequent intelligent algorithm analysis. The specific implementation steps include: In the new energy transformer monitoring system of a solar power station, the collected data is sorted out. By analyzing the key parameters of the transformer operation, it is determined that the collected data types include three-phase current , three-phase voltage , Active Power , reactive power And the internal temperature of the box transformer , a total of 8 data types, namely .

[0040] Set the data collection time length seconds, and collect data every 0.5 seconds. In these 30 seconds, a total of 60 sets of data are collected. According to the formula Constructing data feature vector The collected 60 groups of 8 types of data are arranged in sequence to form a The data feature vector For example, the three-phase currents collected at the first second are , the three-phase voltages are , , Active power is , the reactive power is , the internal temperature of the box transformer is , then the data feature vector The first row of elements is And so on, the construction of the entire data feature vector is completed.

[0041] Embodiment 2:

[0042] This embodiment focuses on the process of building a time series analysis module based on data feature vectors. Through wavelet transform and recurrent neural network, it deeply mines the characteristics and laws of data in the time dimension, and provides a time series analysis basis for intelligent power distribution optimization. The specific implementation steps include: Wavelet transform processing: Continuing with the data of the above solar power station as an example, the constructed data feature vector (60×8) is processed by wavelet transform. The commonly used Daubechies wavelet (such as db4 wavelet) is used for transformation. According to the formula Get the initial feature transformation result After wavelet transform, The dimension and data distribution of have changed, and it contains data features of different frequency components, which is more conducive to subsequent analysis. For example, It may become a dimension The specific dimension is determined by the wavelet transform parameters and data characteristics, in which different elements reflect the characteristics of the original data at different time scales and frequencies.

[0043] Slice reorganization processing to build a time series analysis module: according to the length of data collection time, The time dimension corresponding to seconds, the result of initial feature transformation Slice the Cut initial feature slices, each slice has a length The length of data collection time Divide by the number of initial feature slices And round up to get, that is .

[0044] right Perform slicing operation to obtain Slices of the form Represents the dimension, Indicates the slice number. For example, the first slice Included Middle Data from the 1st element to the 6th element in the dimension.

[0045] Using these initial feature slices, we reorganize them based on a recurrent neural network (RNN). We use a long short-term memory network (LSTM, a special type of RNN) for calculations, according to the formula

[0046] Get the time series analysis results In the LSTM network, the data of each slice is input sequentially, and the network learns the temporal dependencies between the data. For example, the input layer of the LSTM network receives the first slice After the hidden layer processes the data and the memory unit updates, the second slice is received. The data, and so on, finally output the time series analysis results . It reflects the changing trends and characteristics of the data in the time dimension, and provides important time series information for subsequent intelligent power distribution optimization.

[0047] Embodiment 3:

[0048] This embodiment introduces in detail the process of building a spatial correlation analysis module based on data feature vectors. Through principal component analysis, information fusion and convolution processing, the correlation and characteristics of data in the spatial dimension are mined, which complements the time series analysis module to provide more comprehensive optimization data for intelligent power distribution. The specific implementation steps include: Principal component analysis dimensionality reduction: Taking the data of solar power stations as an example, the initial feature transformation results Perform principal component analysis (PCA) dimensionality reduction. The purpose of PCA is to reduce the dimension of the data and reduce the amount of calculation while retaining the main features of the data. Through PCA operation, according to the formula Get the feature data after dimensionality reduction . Assume that after PCA analysis, The dimension from Reduce to 8, that is, 8 principal components are retained. In this process, the covariance matrix of the data is calculated, the eigenvalue decomposition of the covariance matrix is ​​performed, and the first 8 eigenvectors with larger eigenvalues ​​are selected as principal components. The original data is projected onto these principal components to obtain the eigendata after dimensionality reduction. .

[0049] Fusion processing: the feature data after dimensionality reduction With preset spatial location information In a solar power station, spatial location information It may include the geographical coordinates of the box transformer in the power station (such as longitude and latitude), the relative position relationship between the box transformer and other equipment, etc. Assuming spatial location information Represented as a The first column represents the longitude information and the second column represents the latitude information. Fusion is performed to obtain fusion feature data , whose dimensions are During the fusion process, and By adding the data in corresponding dimensions, the fused data contains both the features of the original data and the spatial location information.

[0050] Convolution processing builds a spatial correlation analysis module: Convolution processing is performed on the fused feature data Fused, using a two-dimensional convolution operation (Conv2D). According to the formula Get the results of spatial correlation analysis . Choose an appropriate convolution kernel size (such as Step size (such as 1) for convolution calculation. During the convolution process, the convolution kernel fuses the feature data Slide up and perform convolution operation on the data in each sliding window to extract the spatial features of the data. Starting from the upper left corner of The weighted summation and other operations are performed on the data in the window to obtain an element of the convolution result. Slide on the top to finally get the spatial correlation analysis results . It reflects the correlation and characteristics of data in the spatial dimension, and provides a spatial analysis basis for intelligent power distribution optimization.

[0051] Embodiment 4:

[0052] This embodiment mainly describes how to use the box-type transformer monitoring and control distribution optimization data to obtain intelligent distribution prediction results, and establish an intelligent distribution error evaluation function, and how to use this function to adaptively adjust the intelligent algorithm model to improve the accuracy and reliability of intelligent distribution.

[0053] The method for calculating the intelligent power distribution prediction result is as follows: according to the output of the spatial correlation analysis module obtained above and the time series analysis module output , and the formula

[0054] Calculate intelligent power distribution forecast results Assuming the number of network layers , output network Using a multi-layer perceptron (MLP) structure, feature conversion function A simple linear transformation function At a certain moment, the outputs of the spatial correlation analysis module and the time series analysis module at the first layer are and , calculated by the feature conversion function

[0055] Add the calculation results of the 4 layers and input them into the output network Finally, the intelligent power distribution prediction results are obtained , such as predicting the optimal output power value of the transformer at the next moment.

[0056] Establishing intelligent power distribution error evaluation function: Based on the prediction results of intelligent power distribution Establish an intelligent power distribution error evaluation function based on the real-time operation data of the corresponding new energy box transformer

[0057] Assume that within a period of time, a total of Group data, The first one in the intelligent power distribution prediction results data, The first one is the real-time data of the new energy box transformer operation actually collected. For example, is the output power of the box-type transformer at a certain moment, The actual measured output power of the box transformer at that moment, the sum of square errors of these 50 sets of data is calculated to obtain the value of the intelligent power distribution error evaluation function

[0058] Adaptive adjustment of intelligent algorithm model: using intelligent distribution error evaluation function The value of is used to adaptively adjust the intelligent algorithm model. Set the error threshold ,assuming . The error evaluation function The value of is compared with the error threshold. , the simulated annealing algorithm is used to adjust the parameters of the intelligent algorithm model. In the simulated annealing algorithm, the initial temperature is set to , the temperature reduction coefficient is set to . In each iteration, a new parameter combination is randomly generated , calculate the intelligent distribution error evaluation function under the new parameter combination , and calculate the error difference .like , then accept new parameter combinations ;like , then with probability Accepts new parameter combinations, where is the current temperature. After each iteration, the temperature is updated , until the maximum number of iterations (such as 100) is reached or the temperature is lower than the preset termination temperature By continuously adjusting the parameters of the intelligent algorithm model, the prediction results of intelligent power distribution can be closer to the actual situation, thus improving the accuracy of intelligent power distribution.

[0059] Embodiment 5:

[0060] This embodiment further optimizes the simulated annealing algorithm's adjustment process for the intelligent algorithm model parameters, and introduces an adaptive factor to dynamically adjust the acceptance probability, thereby improving the algorithm's search efficiency and convergence speed, so that the intelligent algorithm model can adapt to data changes more quickly and accurately, and optimize the intelligent power distribution results. The specific implementation steps include: Introducing adaptive factors to update acceptance probability: In the process of adjusting the parameters of the intelligent algorithm model using the simulated annealing algorithm, the adaptive factor γ is introduced. The adaptive factor γ is calculated according to the formula

[0061] Update, where is the preset iteration number threshold, assuming , To control the parameter of the adaptive factor change rate, set . In calculating the acceptance probability When , the acceptance probability is adjusted to ,in Assume that the basic acceptance probability .

[0062] Example of adjusting parameters of simulated annealing algorithm: Taking the simulated annealing algorithm parameters in the above embodiment as an example, in a certain iteration, the number of iterations is First, according to the adaptive factor formula, The value of . Randomly generate new parameter combinations After that, the intelligent distribution error evaluation function under the new parameter combination is calculated. , and calculate the error difference Assumptions , current temperature , then the original acceptance probability Adjusted probability of acceptance According to the adjusted acceptance probability Decide whether to accept new parameter combinations In the subsequent iterations, as The change of adaptive factor Keep updating and accept the probability The corresponding changes also enable the simulated annealing algorithm to better balance the global search and local search capabilities when searching for the optimal parameter combination, improve the convergence speed of the algorithm and the optimization effect of the intelligent algorithm model, and thus improve the accuracy and stability of intelligent power distribution. Through the above detailed description of the embodiments, the specific implementation process of the intelligent power distribution system and method of the present invention in the monitoring and control of new energy box transformers is clearly presented. From data collection and processing to model construction and optimization, each link is closely coordinated to provide a reliable technical solution for realizing accurate and efficient new energy box transformer power distribution.

[0063] The application of the present invention can also be combined with an integrated intelligent power distribution terminal. Take a large wind farm as an example, in which new energy box transformers are widely distributed. In the past, only traditional current sensors, voltage sensors, temperature sensors, etc. were used to collect data. Although basic operating parameters can be obtained, there are deficiencies in the precise control of box transformer power distribution and rapid response to faults. An integrated intelligent power distribution terminal (intelligent circuit breaker) with multiple functions is introduced, which not only integrates the opening and closing control functions of traditional circuit breakers, but also has built-in high-precision power monitoring modules, communication modules and intelligent control chips.

[0064] When the intelligent power distribution system calculates that the power distribution parameters need to be adjusted, the intelligent power distribution terminal (intelligent circuit breaker) performs further control. For example, if it is calculated that the current of a certain phase is too large, which may cause equipment overload, the intelligent power distribution terminal can quickly adjust the conduction time and current limit of the phase, and accurately control the circuit through the internal intelligent control chip to reduce the current of the phase, ensure balanced power distribution, and avoid equipment overload operation. At the same time, the intelligent power distribution terminal can also adjust the output voltage of the box transformer according to the optimization instructions of the system to adapt to the needs of different power-consuming equipment and improve energy utilization efficiency. When an abnormal situation is detected, such as a short circuit fault, the intelligent power distribution terminal can cut off the circuit in a very short time to protect the box transformer and other equipment, and upload the fault information to the monitoring center in time through the communication module, so that the operation and maintenance personnel can respond quickly. By introducing an integrated intelligent power distribution terminal and combining the algorithm optimization of the intelligent power distribution system, the reliability of the new energy box transformer of the wind farm has been significantly improved, the power distribution efficiency has been improved, and the equipment loss has been reduced, which effectively guarantees the stable operation and efficient power generation of the wind farm.

[0065] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0066] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent power distribution method for monitoring and controlling new energy box transformers, characterized in that: include: Obtain real-time data on the operation of new energy box transformers; Processing the real-time data of the new energy box transformer operation to obtain a data feature vector; According to the data feature vector, the box-type transformer monitoring and control power distribution optimization data is obtained based on the intelligent algorithm model; By using the box-type transformer monitoring and control power distribution optimization data, the intelligent power distribution results of the new energy box-type transformer monitoring and control are obtained.

2. According to claim 1, an intelligent power distribution method for monitoring and controlling new energy box transformers is characterized in that: The real-time data of the new energy box transformer operation is processed to obtain the data feature vector including: Determine the number of data types included in the real-time data of new energy box transformer operation; The calculation formula for constructing the data feature vector according to the number of data types is: ； in, is the data feature vector, is the length of data collection time, is the number of data types, is a real number set, which is an array consisting of real numbers.

3. According to claim 2, an intelligent power distribution method for monitoring and controlling new energy box transformers is characterized in that: The step of obtaining the box-type transformer monitoring and control power distribution optimization data based on the intelligent algorithm model according to the data feature vector includes: The data feature vector is subjected to wavelet transformation to obtain an initial feature transformation result, the calculation formula of which is: ； in is the initial feature transformation result, represents the wavelet transform operation; Utilizing the initial feature transformation results, a time series analysis module and a spatial correlation analysis module are respectively constructed based on a deep neural network.

4. According to claim 3, an intelligent power distribution method for monitoring and controlling new energy box transformers is characterized in that: Constructing a time series analysis module and a spatial association analysis module based on a deep neural network respectively includes: slicing and reorganizing the initial feature transformation results to construct a time series analysis module; and performing dimensionality reduction and fusion processing on the initial feature transformation results to construct a spatial association analysis module.

5. According to claim 4, an intelligent power distribution method for monitoring and controlling new energy box transformers is characterized in that: The slicing and reorganizing processing of the initial feature transformation result to construct a time series analysis module comprises: Slice the initial feature transformation result according to the time dimension corresponding to the data collection time length to obtain the initial feature slice; Using the initial feature slices, the reorganization process is performed based on the recurrent neural network, and the calculation formula for the time series analysis result is obtained as follows: ； in, is the result of time series analysis. is a recurrent neural network, To concatenate Q slices along a new dimension, is the number of initial feature slices, is the length of each initial feature slice, which is obtained by dividing the data acquisition time length M by the number of initial feature slices Q and rounding up. is the initial feature transformation result In Dimension On slices, the time range is from arrive ; The time series analysis result is used as the output of the time series analysis module.

6. According to claim 4, an intelligent power distribution method for monitoring and controlling new energy box transformers is characterized in that: The performing dimensionality reduction and fusion processing on the initial feature transformation results to construct a spatial correlation analysis module comprises: The principal component analysis is used to reduce the dimension of the initial feature transformation results to obtain the feature data after dimension reduction. The calculation formula is: ； in, is the feature data after dimension reduction, represents the principal component analysis operation; The feature data after dimensionality reduction is fused with the preset spatial position information to obtain fused feature data. The calculation formula is: ； in, is the preset spatial location information; The fused feature data is convolved to obtain the calculation formula for the spatial correlation analysis result: ； in, is the result of spatial correlation analysis. It is a two-dimensional convolution operation; The spatial correlation analysis result is used as the output of the spatial correlation analysis module.

7. The intelligent power distribution method for monitoring and controlling new energy box transformers according to claim 1 is characterized in that: The intelligent power distribution results obtained by monitoring and controlling the new energy box-type transformer using the box-type transformer monitoring and control power distribution optimization data include: According to the distribution optimization data of the box-type transformer monitoring and control, the calculation formula for establishing the intelligent distribution prediction result is: ； in, For intelligent power distribution prediction results, For the output network, is the feature conversion function, and The spatial correlation analysis module and the time series analysis module are respectively The output of the layer, is the number of network layers; According to the intelligent power distribution prediction results and the corresponding new energy box transformer operation real-time data, the calculation formula of the intelligent power distribution error evaluation function is established as follows: ； in, Provide real-time data for the new power supply box. is the intelligent power distribution error evaluation function, is the number of data samples; The calculated value of the intelligent power distribution error evaluation function is used to adaptively adjust the intelligent algorithm model.

8. The intelligent power distribution method for monitoring and controlling new energy box transformers according to claim 7 is characterized in that: The method of adaptively adjusting the intelligent algorithm model by using the calculated value of the intelligent power distribution error evaluation function includes: Setting the error threshold , the intelligent distribution error evaluation function The value of is compared with the error threshold; like , the simulated annealing algorithm is used to adjust the parameters of the intelligent algorithm model; the initial temperature of the simulated annealing algorithm is set to , the temperature drop coefficient is set to ,in 1; In each iteration, a new parameter combination is randomly generated , calculate the intelligent distribution error evaluation function under the new parameter combination , and calculate the error difference ; like , then accept new parameter combinations ;like , then with probability Accepts new parameter combinations, where is the current temperature; after each iteration, the temperature is updated , until the maximum number of iterations is reached or the temperature is lower than the preset termination temperature .

9. The intelligent power distribution method for monitoring and controlling new energy box transformers according to claim 8 is characterized in that: The method also includes dynamically adjusting the acceptance probability during the process of adjusting the parameters of the intelligent algorithm model using the simulated annealing algorithm, specifically: Introducing adaptive factors , as the number of iterations k increases, the adaptive factor Updated by the following formula: ； in, is the preset iteration number threshold, Parameters that control the rate of change of the adaptive factor; in calculating the acceptance probability When , the acceptance probability is adjusted to ,in The basic acceptance probability.

10. An intelligent power distribution system for monitoring and controlling new energy box transformers, characterized in that: The system comprises: Data acquisition module, used to obtain real-time data of new energy box transformer operation; A data processing module, connected to the data acquisition module, for processing the real-time data of the new energy box transformer operation to obtain a data feature vector; An optimization calculation module, connected to the data processing module, for obtaining distribution optimization data for monitoring and regulating the box-type transformer based on the data feature vector and an intelligent algorithm model; The result output module is connected to the optimization calculation module and is used to obtain the intelligent distribution results of the new energy box transformer monitoring and regulation by using the box transformer monitoring and regulation power distribution optimization data.

Images