Voltage fluctuation control method and system for subway DC power supply system
By performing high-frequency sampling, denoising and time-frequency analysis on the subway DC power supply system, combined with distributed optimization algorithm, the system's voltage fluctuation and energy utilization problems under peak and extreme conditions are solved, and the system's stability and energy are improved. efficiency.
Patent Information
- Application Number
- CN202411197367.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-29
AI Technical Summary
The existing subway DC power supply system is difficult to adapt to complex operating environments during peak hours and extreme weather conditions, resulting in excessive voltage fluctuations or waste of energy, high computational complexity, poor real-time performance, and energy recovery and distribution strategies are difficult to achieve across the entire network. Efficient utilization, lack of precise modeling and analysis of the dynamic characteristics of the system, and lack of robustness.
By sampling node data at the subway DC power supply system at high frequency, perform wavelet threshold denoising processing and Hilbert-yellow transformation analysis, calculate the correspondence between network voltage and electric braking power, perform constant slope optimization analysis, and combine distributed optimization The algorithm conducts energy distribution strategy analysis to form a global voltage stability control scheme.
It improves the accuracy of voltage fluctuation control and the energy utilization efficiency of the system, reduces the adverse impact of voltage fluctuations on system operation, enhances the robustness and voltage stability of the system, and realizes global coordinated control.
Smart Images

Figure CN118713034B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a voltage fluctuation control method and system for a subway direct current power supply system. Background Art
[0002] With the rapid development of urban rail transit, the stability and efficiency of subway DC power supply systems have become key issues. Traditional subway DC power supply systems use fixed voltage rectifiers and switch devices to maintain the stability of the power supply network by adjusting the rectifier output voltage and switch switching. This method has achieved certain results in dealing with voltage fluctuations, energy recovery and load balancing. In recent years, with the development of power electronics technology and intelligent control algorithms, new devices such as controlled rectifiers and bidirectional DC / DC converters have been introduced into power supply systems, improving the system's dynamic response capability and energy utilization efficiency. At the same time, predictive control and distributed optimization algorithms based on artificial intelligence have also begun to be applied in subway power supply systems, providing new ideas for the intelligent management of the system.
[0003] However, the existing technology still has some shortcomings. First, the traditional fixed threshold control strategy is difficult to adapt to the complex and changeable operating environment, especially during peak hours and extreme weather conditions, and is prone to problems such as excessive voltage fluctuations or energy waste. Secondly, although intelligent algorithms have been introduced, most systems still use a centralized control architecture, which has problems such as high computational complexity and poor real-time performance when facing large-scale networks. In addition, the existing energy recovery and distribution strategies often only consider local optimization, making it difficult to achieve efficient energy utilization across the entire network. Finally, due to the lack of accurate modeling and analysis of the dynamic characteristics of the system, the robustness of the existing control scheme in dealing with emergencies and abnormal operating conditions needs to be improved. Summary of the invention
[0004] In view of this, an embodiment of the present invention provides a voltage fluctuation control method and system for a subway DC power supply system, which are used to improve the accuracy of voltage fluctuation control of the subway DC power supply system.
[0005] The present invention provides a voltage fluctuation control method for a subway direct current power supply system, comprising: performing high-frequency sampling of node data of the subway direct current power supply system to obtain an original data set containing voltage, current and position information; performing wavelet threshold denoising processing on the original data set to obtain denoised power data; performing time-frequency analysis on the denoised power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data; calculating the corresponding relationship between the grid voltage data and the electric braking power of the subway direct current power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics to obtain a voltage control characteristic curve; performing constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets a preset time constraint; based on the electric braking force control scheme, performing energy distribution strategy analysis on the system topology structure of the subway direct current power supply system through a distributed optimization algorithm to obtain a global voltage stability control scheme.
[0006] In the present invention, the step of performing high-frequency sampling of node data of a subway DC power supply system to obtain an original data set containing voltage, current and position information includes: analyzing the system topology of the subway DC power supply system to obtain a key node distribution diagram; prioritizing the sampling points according to the key node distribution diagram to obtain a sampling sequence list; setting a sampling frequency parameter for each sampling point in the sampling sequence list to obtain a sampling configuration scheme; synchronously triggering sampling of each sampling point according to the sampling configuration scheme to obtain an original sampling data stream; timestamping the original sampling data stream to obtain time-series associated sampling data; constructing a data packet according to the time-series associated sampling data to obtain a formatted data packet; performing data integrity check on the formatted data packet to obtain a valid data set, and performing time window segmentation on the valid data set to obtain an original data set containing voltage, current and position information.
[0007] In the present invention, the step of performing wavelet threshold denoising on the original data set to obtain denoised power data includes: performing multi-scale wavelet decomposition on the original data set to obtain wavelet coefficient matrices of different frequency bands, and performing inter-layer correlation analysis on the wavelet coefficient matrix to obtain inter-layer correlation coefficients; adaptively selecting wavelet basis functions according to the inter-layer correlation coefficients to obtain optimal wavelet bases, and re-decomposing the original data set according to the optimal wavelet bases to obtain optimized wavelet coefficients; performing statistical characteristic analysis on the optimized wavelet coefficients to obtain noise distribution characteristics of each scale, and constructing a nonlinear mapping function according to the noise distribution characteristics to obtain an initial threshold curve; performing segmented optimization on the initial threshold curve to obtain a multi-level threshold parameter set, and performing adaptive threshold processing on the wavelet coefficients through the multi-level threshold parameter set to obtain denoised wavelet coefficients; performing inverse wavelet transform on the denoised wavelet coefficients to obtain preliminary denoised data, and performing edge-preserving filtering on the preliminary denoised data to obtain denoised power data.
[0008] In the present invention, the step of performing time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data includes: segmenting the noise-reduced power data according to the train running time to obtain voltage and current data segments of multiple time windows; performing Hilbert-Huang transform on the voltage and current data segments to obtain analytical data containing real and imaginary parts; calculating the instantaneous amplitude and instantaneous phase according to the analytical data to obtain the instantaneous characteristics of the voltage and current, and performing empirical mode decomposition on the instantaneous characteristics to obtain embedded modal data; and performing empirical mode decomposition on the embedded The modal data is subjected to Hilbert-Huang transform to obtain time-frequency distribution data of each mode, and the instantaneous frequency is calculated based on the time-frequency distribution data to obtain multi-scale frequency components; the multi-scale frequency components are subjected to time-frequency energy distribution analysis to obtain time-varying characteristics of voltage fluctuations, and the main frequency components are extracted based on the time-varying characteristics to obtain the instantaneous frequency characteristics; the instantaneous frequency characteristics are reconstructed in the time domain to obtain frequency reconstructed data, and voltage distribution analysis is performed on the dynamic voltage distribution data and the spatial information of the subway line through the frequency reconstructed data to obtain the dynamic voltage distribution data.
[0009] In the present invention, the step of calculating the correspondence between the grid voltage data and the electric braking power of the subway DC power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics to obtain the voltage control characteristic curve includes: performing time series segmentation on the dynamic voltage distribution data to obtain voltage change data of multiple time windows, and performing statistical analysis on the voltage change data of the multiple time windows to obtain voltage fluctuation characteristic parameters; classifying the voltage change trend according to the voltage fluctuation characteristic parameters to obtain the voltage fluctuation mode; performing spectral decomposition on the instantaneous frequency characteristics to obtain the main frequency components and amplitude data, and calculating the subway according to the main frequency components and amplitude data. The instantaneous impedance of the DC power supply system is obtained to obtain an impedance characteristic curve; the voltage fluctuation mode is corrected by the impedance characteristic curve to obtain corrected voltage fluctuation data; the corrected voltage fluctuation data is correlated with the real-time collected train operation status data to obtain the corresponding relationship between the grid voltage data and the train position; the electric braking power at different positions is calculated according to the corresponding relationship between the grid voltage data and the train position to obtain spatially distributed electric braking power data, and the spatially distributed electric braking power data is normalized and curve mapped to obtain a standardized electric braking power curve; the standardized electric braking power curve is fitted with the grid voltage data to obtain a voltage control characteristic curve.
[0010] In the present invention, the step of performing constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets the preset time constraint includes: performing piecewise linearization processing on the voltage control characteristic curve to obtain a multi-segment linear curve, and calculating the slope of each segment of the multi-segment linear curve to obtain an initial slope set; performing statistical analysis on the initial slope set to obtain slope distribution characteristics, and determining a target slope range based on the slope distribution characteristics to obtain a slope constraint condition; combining the slope constraint condition with the preset time constraint to obtain an optimization objective function, and solving the optimization objective function by the gradient descent method to obtain a candidate slope value; discretizing the candidate slope value to obtain a discrete slope sequence, and performing interpolation fitting on the discrete slope sequence to obtain a continuous slope curve; according to The continuous slope curve reconstructs the voltage control characteristic curve to obtain an optimized voltage control curve, and performs an electric braking force change rate analysis on the optimized voltage control curve to obtain the electric braking force change rate; a cross-correlation analysis is performed on the electric braking force change rate and the train operation status data to obtain a relationship matrix between the electric braking force and the train speed, and an electric braking force speed curve is constructed based on the relationship matrix to obtain electric braking force control characteristic data; the electric braking force control characteristic data is subjected to piecewise linearization processing to obtain a piecewise control strategy; the braking time of each segment is calculated based on the piecewise control strategy to obtain a cumulative braking time series; the cumulative braking time series is compared with a preset time constraint to obtain a time deviation value; the piecewise control strategy is fine-tuned through an iterative adjustment method to obtain an electric braking force control scheme that meets the preset time constraint.
[0011] In the present invention, based on the electric braking force control scheme, the system topology structure of the subway DC power supply system is analyzed for energy distribution strategy through a distributed optimization algorithm to obtain a global voltage stability control scheme, including: dividing the system topology structure into nodes to obtain multiple sub-network areas, and calculating the power flow distribution of the multiple sub-network areas to obtain an initial energy distribution scheme; performing load balancing analysis on each sub-network area according to the initial energy distribution scheme to obtain a load imbalance index, and performing threshold judgment on the load imbalance index to obtain a set of key nodes to be adjusted; constructing an energy transfer matrix according to the key node set to obtain an energy transfer relationship between nodes, and setting constraints on the energy transfer relationship to obtain an energy distribution optimization target; solving the energy distribution optimization target through a distributed gradient projection algorithm. , obtain the energy allocation ratio of each node; perform feasibility verification on the energy allocation ratio of each node to obtain a set of alternative energy allocation schemes, and perform multi-objective trade-off analysis based on the set of alternative energy allocation schemes to obtain an optimal energy allocation strategy; discretize the optimal energy allocation strategy to obtain an energy allocation sequence for each time period, and calculate the voltage change trend of each node based on the energy allocation sequence for each time period to obtain a node voltage prediction curve; map the electric braking force control scheme to a current demand curve to obtain a train load characteristic curve; perform time-space matching on the node voltage prediction curve and the train load characteristic curve to obtain a dynamic load distribution diagram, and calculate the voltage margin of each power supply interval based on the dynamic load distribution diagram to obtain a voltage regulation target value; perform real-time correction on the voltage regulation target value to obtain the global voltage stability control scheme.
[0012] The present invention also provides a voltage fluctuation control system for a subway DC power supply system, comprising:
[0013] The sampling module is used to perform high-frequency sampling of node data in the subway DC power supply system to obtain an original data set containing voltage, current and position information;
[0014] A processing module, used for performing wavelet threshold denoising processing on the original data set to obtain denoised power data;
[0015] An analysis module, used for performing time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data;
[0016] A calculation module, used to calculate the corresponding relationship between the grid voltage data and the electric braking power of the subway DC power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics, and obtain a voltage control characteristic curve;
[0017] An optimization module, used for performing a constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets a preset time constraint;
[0018] The distribution module is used to analyze the energy distribution strategy of the system topology of the subway DC power supply system based on the electric braking force control scheme through a distributed optimization algorithm to obtain a global voltage stability control scheme.
[0019] In the technical solution provided by the present invention, by high-frequency sampling of multiple key nodes preset in the system, the original data set containing voltage, current and position information is obtained, which effectively improves the accuracy and real-time performance of voltage fluctuation monitoring. The original data set is subjected to wavelet threshold denoising, which not only effectively removes the noise interference in the signal, but also retains the key characteristics of voltage fluctuation, significantly improving the data quality. The de-noised power data is subjected to time-frequency analysis through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data. This step deeply reveals the time-frequency characteristics of voltage fluctuation and provides a theoretical basis for precise control. According to the dynamic voltage distribution data and the instantaneous frequency characteristics, the corresponding relationship between the grid voltage and the electric braking power is calculated to obtain the voltage control characteristic curve, and a direct connection between the voltage fluctuation and the system operating state is established, laying the foundation for formulating an effective control strategy. The voltage control characteristic curve is subjected to constant slope optimization analysis to obtain an electric braking force control scheme that meets the preset time constraints. This innovative optimization process not only ensures the stability of the electric braking process, but also maximizes the energy recovery efficiency. Finally, based on the electric braking force control scheme, the energy distribution strategy of the system topology is analyzed through the distributed optimization algorithm, and the global voltage stability control scheme is obtained, which fully considers the integrity and complexity of the system, realizes the global coordinated control of voltage fluctuations, significantly improves the voltage stability of the subway DC power supply system, and reduces the adverse effects of voltage fluctuations on system operation. At the same time, by optimizing the electric braking process and energy distribution strategy, the energy utilization efficiency of the system is greatly improved and energy loss is reduced. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0021] Figure 1 The present invention is a flowchart of a method for controlling voltage fluctuations in a subway DC power supply system according to an embodiment of the present invention.
[0022] Figure 2The flowchart of performing wavelet threshold denoising on the original data set in an embodiment of the present invention.
[0023] Figure 3 The figure is a schematic diagram of a voltage fluctuation control system of a subway DC power supply system according to an embodiment of the present invention.
[0024] Reference numerals:
[0025] 301, sampling module; 302, processing module; 303, analysis module; 304, calculation module; 305, optimization module; 306, allocation module. DETAILED DESCRIPTION
[0026] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.
[0027] In the description of the present invention, it should be noted that the terms “first”, “second” and “third” are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0028] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0029] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , Figure 1 is a flow chart of a voltage fluctuation control method for a subway DC power supply system according to an embodiment of the present invention. Figure 1 As shown, the following steps are included:
[0030] S101, performing high-frequency sampling of node data of the subway DC power supply system to obtain an original data set including voltage, current and position information;
[0031] Specifically, high-precision data acquisition equipment is installed at the key nodes of the system. These devices usually include voltage transformers, current transformers and GPS positioning modules, which can capture electrical parameters and spatial information in the power grid in real time. During the installation process, the anti-interference ability and environmental adaptability of the equipment need to be considered to ensure stable operation in the complex electromagnetic environment of the subway. The selection of sampling frequency is crucial, and it usually needs to reach thousands or even tens of thousands of times per second to capture fast transient processes in the power grid. In order to achieve such a high sampling rate, it is necessary to use high-speed analog-to-digital converters and high-performance processors, and large-capacity storage devices are also required to save massive amounts of sampled data. In addition, in order to ensure the synchronization and accuracy of the data, the entire sampling system needs to use a unified clock source, usually using GPS timing or high-precision atomic clocks to provide a time reference. In terms of data transmission, considering the particularity of the subway environment, optical fiber communication networks can be used to achieve high-speed, low-latency data transmission. In the data processing link, it is necessary to design efficient data compression and preprocessing algorithms to reduce storage and transmission pressure. At the same time, in order to ensure the reliability of the data, it is also necessary to implement redundant backup and real-time verification mechanisms.
[0032] In practical applications, take the Metro Line 1 of a certain city as an example. The line is 30 kilometers long and has 15 stations and 3 traction substations. Data acquisition equipment is installed at each station and substation, and the sampling frequency is set to 10kHz. Each sampling point simultaneously records voltage, current and GPS location information to form a data packet containing timestamp, voltage value, current value and spatial coordinates. These data are transmitted to the central control room in real time through the subway's dedicated optical fiber network, where preliminary data filtering and sorting are performed. In this way, the system can collect more than 300,000 data points per second, providing rich raw data support for subsequent voltage fluctuation analysis and control strategy formulation.
[0033] S102, performing wavelet threshold denoising processing on the original data set to obtain denoised power data;
[0034] Specifically, a wavelet basis function suitable for the characteristics of the power signal is selected. Commonly used ones include Daubechies wavelet, Symlet wavelet or Coiflet wavelet, etc. These wavelet functions can effectively capture the mutation and transient characteristics in the power signal. After selecting the wavelet basis, the original data is subjected to multi-scale wavelet decomposition, usually decomposed into 4 to 6 levels to cover the different frequency components of the signal. During the decomposition process, the original signal is divided into low-frequency approximate coefficients and high-frequency detail coefficients, among which the high-frequency part often contains a lot of noise information. Next, the denoising threshold is determined, which is the key step of wavelet threshold denoising. Commonly used threshold selection methods include general threshold method, SURE threshold method and mini-maximum threshold method. Among them, the general threshold method is simple and intuitive, but it may over-smooth the signal; the SURE threshold method is based on Stein unbiased risk estimation, which can achieve a good balance between retaining signal details and removing noise; and the mini-maximum threshold method performs well when processing strong noise signals. In practical applications, it is often necessary to select the most appropriate threshold method according to the characteristics of the signal and the noise level, and even adaptive threshold technology can be used to dynamically adjust the threshold for different decomposition levels and different time windows.
[0035] After determining the threshold, the wavelet coefficients are thresholded. Common processing methods include hard thresholding and soft thresholding. The hard thresholding method directly sets the coefficients less than the threshold to zero and keeps the coefficients greater than the threshold unchanged. This method is simple and effective, but it may introduce discontinuities when reconstructing the signal. The soft thresholding rule shrinks the coefficients greater than the threshold, which can obtain smoother results, but may cause slight attenuation of the signal amplitude. In order to balance the denoising effect and signal fidelity, a hybrid strategy of hard thresholding and soft thresholding is sometimes adopted, or a more complex nonlinear threshold function is used. After completing the thresholding, the signal is reconstructed using the inverse inverse wavelet transform to obtain the denoised power data. During the reconstruction process, it is necessary to pay attention to maintaining the phase information of the signal to ensure that the reconstructed signal is aligned in time with the original signal. In addition, in order to further improve the denoising effect, other filtering techniques, such as Kalman filtering or median filtering, can be combined with wavelet denoising to handle certain types of noise that may not be completely removed by wavelet transform.
[0036] For example, the voltage data collected by the traction substation of a subway line was affected by power frequency interference and random white noise. Daubechies4 wavelet was used for 5-layer decomposition, the SURE threshold method was used to calculate the threshold of each layer, and soft threshold processing was adopted. The processing results show that the 50Hz power frequency interference was effectively suppressed, and the signal-to-noise ratio was improved from the original 20dB to 35dB, while retaining the rapid change characteristics of voltage fluctuations. This high-quality data after denoising provides a reliable basis for subsequent voltage stability analysis and control strategy formulation, effectively improving the operating efficiency and reliability of the entire power supply system.
[0037] S103, performing time-frequency analysis on the noise reduction power data by using Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data;
[0038] Specifically, the de-noised power data is preprocessed, including data standardization and trend item removal, to ensure the accuracy of subsequent analysis. Then, the signal is decomposed into a series of intrinsic mode functions (IMFs) using the empirical mode decomposition (EMD) or ensemble empirical mode decomposition (EEMD) method. The EMD method can adaptively decompose nonlinear and non-stationary signals into a finite number of IMF components, each of which represents an oscillation mode in the signal. In practical applications, EEMD can effectively solve the problem of mode aliasing and improve the stability and reliability of decomposition by adding white noise and performing EMD decomposition multiple times. Hilbert transform is performed on each IMF to obtain an analytical signal. This step can convert the real signal into a complex signal, which is convenient for subsequent phase and amplitude analysis. By calculating the amplitude and phase of the analytical signal, the instantaneous amplitude and instantaneous phase of each IMF can be obtained. The instantaneous frequency is obtained by the time derivative of the instantaneous phase. In this way, we get a complete representation of the signal in the three-dimensional space of time-frequency-energy, namely the Hilbert spectrum. The Hilbert spectrum provides detailed information on the change of signal energy over time and frequency, and can clearly show the nonlinear and non-stationary characteristics in the signal. In order to better understand and analyze the Hilbert spectrum, spectrum analysis and feature extraction are usually required. This includes identifying the main time-frequency components, calculating the statistical characteristics of the instantaneous frequency (such as mean, variance, skewness and kurtosis, etc.), and analyzing the time-varying characteristics of the frequency components. In addition, the overall distribution characteristics of the signal in the frequency domain can be obtained by calculating the marginal spectrum. These analysis results together constitute the instantaneous frequency characteristics, which reflect the dynamic characteristics of voltage and current fluctuations in the power system.
[0039] The acquisition of dynamic voltage distribution data requires the combination of the results of the Hilbert-Huang transform and the spatial information of the system. By spatially interpolating and mapping the instantaneous frequency characteristics of different sampling points, a dynamic voltage distribution map of the entire power supply network can be constructed. This distribution map not only reflects the spatial changes in voltage, but also shows the evolution of this change over time. During the construction process, factors such as network topology and load distribution may need to be considered to ensure that the generated dynamic voltage distribution data accurately reflects the actual state of the system.
[0040] For example, a subway line experienced voltage fluctuations during the morning rush hour. By performing Hilbert-Huang transform analysis on the collected voltage data, it was found that low-frequency oscillations of about 7 Hz occurred in a specific time period. Further analysis of the instantaneous frequency characteristics revealed that this oscillation was highly correlated with the train start-up and braking process. By constructing a dynamic voltage distribution map, the propagation characteristics of voltage fluctuations in space are clearly shown, that is, the fluctuations start from a certain station and gradually spread along the line. This detailed time-frequency analysis result provides valuable information for system operators, enabling them to adjust train operation strategies and power supply parameters in a timely manner, effectively suppress voltage fluctuations, and improve system stability. In addition, these analysis results also provide an important basis for the long-term optimization and upgrading of the power system, such as guiding the reasonable layout and capacity configuration of energy storage devices. In this way, the Hilbert-Huang transform not only plays an important role in short-term operation control, but also provides data support for the long-term improvement of the system.
[0041] S104, calculating the corresponding relationship between the grid voltage data and the electric braking power of the subway DC power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics, and obtaining a voltage control characteristic curve;
[0042] Specifically, the dynamic voltage distribution data is segmented into time series, usually using the sliding window method. The window size is determined according to the system characteristics and analysis requirements, and may range from a few seconds to a few minutes. This segmentation can capture the short-term dynamic characteristics of voltage changes and provide a basis for subsequent analysis. Next, the voltage data in each time window is statistically analyzed to calculate characteristic parameters such as mean, standard deviation, and peak factor, which can describe the intensity and pattern of voltage fluctuations. At the same time, the instantaneous frequency characteristics are used for spectral decomposition, usually using methods such as fast Fourier transform (FFT) or wavelet transform to obtain the main frequency components and their amplitudes. This step can reveal the periodic components in the voltage fluctuations, which is particularly important for identifying harmonic interference and low-frequency oscillations in the system. Combining the voltage statistical characteristics and spectral information, the characteristic vector of the voltage fluctuation can be constructed for subsequent pattern recognition and classification.
[0043] Next, a relationship model between grid voltage and electric brake power is established. This usually involves complex power system modeling, which requires consideration of multiple factors such as train dynamics, traction substation characteristics, and line impedance. A combination of theoretical analysis and data-driven methods can be used, such as using power flow calculations to determine the grid voltage-power relationship under static conditions, and then correcting and optimizing the model through measured data. In this process, advanced machine learning algorithms such as support vector regression (SVR) or deep neural networks (DNN) may be required to capture the nonlinear relationship between grid voltage and electric brake power. In order to improve the accuracy and adaptability of the model, the dynamic characteristics of the system also need to be considered. This can be achieved by introducing instantaneous frequency features into the model. For example, instantaneous frequency features can be used to identify the operating state of the system (such as acceleration, uniform speed, braking, etc.), and then sub-models can be established for different states, or instantaneous frequency features can be used as additional input variables of the model. This method can better describe the dynamic changes during the electric braking process and improve the prediction accuracy of the model.
[0044] After establishing the relationship model between grid voltage and electric brake power, a lot of simulation and verification work needs to be done. This includes using historical data for model training and testing, as well as performance evaluation under different operating scenarios. Through repeated tuning and verification, a model that can accurately describe the relationship between grid voltage and electric brake power is finally obtained. Finally, based on this model, a voltage control characteristic curve can be generated through parameter scanning or optimization algorithm. This curve usually describes the maximum electric brake power that the system can safely absorb at different grid voltage levels. In the generation process, various system constraints need to be considered, such as equipment rated capacity, voltage stability limit, etc., to ensure that the generated characteristic curve is feasible in actual operation.
[0045] For example, a subway line is 30 kilometers long, with 20 stations and 3 traction substations. By analyzing the operation data for one month, it was found that the grid voltage varied between 750V and 900V, and the maximum electric braking power of a single train could reach 3MW. Using the above method, the researchers established a comprehensive model that takes into account factors such as grid voltage, train position, and operating status. The model shows that when the grid voltage is lower than 800V, the absorbable electric braking power drops sharply; when the grid voltage is higher than 880V, the system can absorb almost all the electric braking energy. Based on these findings, a nonlinear voltage control characteristic curve was generated. This curve guides the system to adjust the absorbable electric braking power under different grid voltage conditions. For example, when the grid voltage is 850V, the system can safely absorb 2.5MW of electric braking power. This precise control characteristic curve not only improves the energy recovery efficiency, but also significantly improves the voltage stability of the system, reduces the number of mechanical brakes triggered by overvoltage, and prolongs the life of the equipment, ultimately achieving an overall improvement in the system's operating efficiency.
[0046] S105, performing constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets a preset time constraint;
[0047] Specifically, the obtained voltage control characteristic curve is analyzed in detail, which usually includes segmentation processing and mathematical modeling of the curve. The segmentation processing can adopt the segmentation linearization method to divide the complex nonlinear curve into several linear segments, each of which is represented by a linear equation. This method not only simplifies the subsequent calculation process, but also facilitates the implementation of the actual control system. When segmenting, it is necessary to consider the inflection point of the curve and the area with large slope changes to ensure that the segmented model can accurately reflect the characteristics of the original curve. Next, the slope of each linear segment is calculated to obtain an initial slope set. This slope set reflects the sensitivity of voltage changes to electric braking power. By performing statistical analysis on these slopes, such as calculating the average, standard deviation, maximum and minimum values, the overall characteristics of the system slope distribution can be obtained. Based on these characteristics, combined with the physical limitations and safety boundaries of the system, a target slope range is determined. This range should not only ensure the stability and safety of the system, but also meet the efficiency requirements of electric braking energy recovery.
[0048] Then, the target slope range is combined with the preset time constraint to construct the optimization objective function. This function usually includes two main parts: one is the slope optimization term, which aims to make the slope of the optimized curve as close to the target slope as possible; the other is the time constraint term, which ensures that the electric braking process is completed within a given time. In addition, some penalty terms may also be included, such as penalties for slope mutations, to ensure the smoothness of the optimized curve. The construction of the optimization objective function requires weighing multiple factors, and weight coefficients may need to be introduced to adjust the importance of different objectives. The optimization process can use gradient descent, Newton's method or more complex optimization algorithms, such as genetic algorithms or particle swarm optimization algorithms. These algorithms continuously adjust the curve parameters in an iterative manner to gradually reduce the objective function value and eventually reach the optimal or approximately optimal solution. During the optimization process, special attention should be paid to the handling of constraints, such as upper and lower limits of voltage and power, to ensure that the optimization results are feasible in the actual system.
[0049] After the optimization is completed, a new set of slope values is obtained. These slope values need to be discretized to meet the requirements of the actual control system. The discretization process should take into account the resolution and response speed of the control system to ensure that the discretized control strategy can be accurately executed by the system. Then, the discretized slope sequence is interpolated and fitted to obtain a continuous slope curve. This curve represents the optimized voltage-power relationship, which not only meets the requirements of constant slope, but also completes the electric braking process within the preset time. Finally, based on the optimized slope curve, combined with the dynamic characteristics and actual operating parameters of the system, a detailed electric braking force control plan is formulated. This plan needs to specify how the electric braking force changes over time under different grid voltage conditions, and how the corresponding control parameters are adjusted. The plan should also include strategies for handling abnormal situations, such as a rapid response mechanism when the grid voltage changes suddenly.
[0050] For example, the voltage control characteristic curve of a subway line shows that when the grid voltage changes between 750V and 900V, the electric braking power increases from 0 to 3MW. The system requires the electric braking process to be completed within 5 seconds while keeping the voltage fluctuation within the range of ±50V. Through constant slope optimization analysis, the researchers obtained a piecewise linear control strategy: in the 750V-800V range, a slope of 0.02MW / V is used; in the 800V-850V range, a slope of 0.03MW / V is used; in the 850V-900V range, a slope of 0.04MW / V is used. This strategy not only ensures that the electric braking is completed within 5 seconds, but also effectively controls the voltage fluctuation. In actual applications, when the train enters the electric braking stage, the control system selects the corresponding slope according to the current grid voltage and dynamically adjusts the electric braking power. For example, when the grid voltage is 820V, the system will increase the electric braking power at a slope of 0.03MW / V until the target braking force is reached or the grid voltage reaches the upper limit. This optimized control scheme not only improves the energy recovery efficiency, but also significantly improves the voltage stability of the system, reduces the number of mechanical braking times triggered by overvoltage, and ultimately achieves an overall improvement in system operating efficiency and safety.
[0051] S106. Based on the electric braking force control scheme, an energy distribution strategy analysis is performed on the system topology of the subway DC power supply system through a distributed optimization algorithm to obtain a global voltage stability control scheme.
[0052] Specifically, the topological structure of the subway DC power supply system is analyzed and modeled in detail, including the spatial distribution and electrical characteristics of key elements such as traction substations, power supply arms, contact networks, and train locations. Based on this information, an accurate network model can be constructed, which should reflect the dynamic characteristics and constraints of the system. Next, the system is divided into multiple sub-network areas, each of which can be a power supply arm or a specific line. This division lays the foundation for subsequent distributed optimization. Then, based on the previously obtained electric braking force control scheme, the initial energy distribution scheme for each sub-network area is calculated. This initial scheme needs to consider the operating status, location, and expected energy feedback of the train. On this basis, a load balancing analysis is performed on each sub-network area, and the load imbalance index is calculated, which can reflect the fairness and efficiency of energy distribution. By setting thresholds, key nodes that need to be adjusted are identified, which are usually areas with excess or insufficient energy.
[0053] Next, an energy transfer matrix is constructed to describe the energy transfer relationship between nodes. This matrix should not only consider physical connections, but also transmission losses and capacity limitations. Based on this matrix, set the constraints and optimization objectives of energy distribution, such as minimizing total energy loss, balancing voltage levels in each region, maximizing energy utilization, etc. These objectives may be contradictory and need to be balanced through multi-objective optimization methods. In order to solve this complex optimization problem, it is an effective method to use distributed optimization algorithms. Commonly used algorithms include distributed gradient descent, alternating direction multiplier method (ADMM) or consistency-based distributed optimization algorithms. The core idea of these algorithms is to decompose the global optimization problem into multiple local optimization problems, each sub-network area is responsible for solving its own local problem, and coordinate the global goal through information exchange. This method can not only handle large-scale systems, but also has good scalability and robustness.
[0054] During the optimization process, each sub-network area calculates the optimal energy distribution ratio based on local information and information from neighboring areas. This process is iterative. As the number of iterations increases, the solutions of each area will gradually converge, eventually reaching the global optimal or near-optimal solution. The optimization results need to be verified for feasibility to ensure that all physical constraints and operational requirements are met. If an infeasible solution is found, the optimization parameters or constraints need to be adjusted and re-optimized. Finally, the optimized energy distribution strategy is combined with the electric braking force control scheme to form a complete global voltage stability control scheme. This scheme should include the energy distribution ratio of each sub-network area, the voltage control target of the key nodes, and the adjustment strategy under different operating scenarios. At the same time, a real-time control and monitoring system needs to be designed to dynamically adjust the control parameters in actual operation to deal with various emergencies.
[0055] For example, a subway line is 50 kilometers long, including 5 traction substations and 30 stations. The system is divided into 10 sub-network areas. During the morning rush hour, some areas experience energy demand peaks due to intensive train starts, while other areas generate a large amount of feedback energy due to train braking. Through the distributed optimization algorithm, the system calculates the optimal energy distribution plan in real time: the excess energy in areas 3 and 7 is transferred to the adjacent areas 2 and 8 respectively, and the output of traction substations in areas 1 and 10 is adjusted to balance the overall load. This solution not only improves energy utilization efficiency, increasing the energy recovery rate from the original 60% to 85%, but also effectively controls voltage fluctuations, keeping the system voltage within ±5% of the rated value. In addition, the solution also dynamically adjusts the voltage control targets of each area according to the load characteristics of different time periods, such as appropriately increasing the voltage level during periods of dense passenger flow to cope with high power demand. Through this globally coordinated control strategy, the system's operational stability has been significantly improved, the annual number of operational interruptions due to voltage problems has been reduced by 70%, and about 15% of total energy consumption has been saved, demonstrating the powerful advantages of distributed optimization in complex system control.
[0056] By executing the above steps, the original data set containing voltage, current and position information is obtained by high-frequency sampling of multiple key nodes preset in the system, which effectively improves the accuracy and real-time performance of voltage fluctuation monitoring. The original data set is subjected to wavelet threshold denoising, which not only effectively removes the noise interference in the signal, but also retains the key characteristics of voltage fluctuation, significantly improving the data quality. The de-noised power data is subjected to time-frequency analysis through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data. This step deeply reveals the time-frequency characteristics of voltage fluctuation and provides a theoretical basis for precise control. According to the dynamic voltage distribution data and instantaneous frequency characteristics, the corresponding relationship between the grid voltage and the electric braking power is calculated to obtain the voltage control characteristic curve, establish a direct connection between the voltage fluctuation and the system operation state, and lay the foundation for the formulation of an effective control strategy. The voltage control characteristic curve is subjected to constant slope optimization analysis to obtain an electric braking force control scheme that meets the preset time constraints. This innovative optimization process not only ensures the stability of the electric braking process, but also maximizes the energy recovery efficiency. Finally, based on the electric braking force control scheme, the energy distribution strategy of the system topology is analyzed through the distributed optimization algorithm, and the global voltage stability control scheme is obtained, which fully considers the integrity and complexity of the system, realizes the global coordinated control of voltage fluctuations, significantly improves the voltage stability of the subway DC power supply system, and reduces the adverse effects of voltage fluctuations on system operation. At the same time, by optimizing the electric braking process and energy distribution strategy, the energy utilization efficiency of the system is greatly improved and energy loss is reduced.
[0057] In a specific embodiment, the process of executing step S101 may specifically include the following steps:
[0058] (1) Analyze the system topology of the subway DC power supply system and obtain the key node distribution diagram;
[0059] (2) Prioritize the sampling points according to the key node distribution map to obtain a sampling order list;
[0060] (3) Setting the sampling frequency parameters for each sampling point in the sampling sequence list to obtain a sampling configuration scheme;
[0061] (4) Synchronously trigger sampling at each sampling point according to the sampling configuration scheme to obtain the original sampling data stream;
[0062] (5) Timestamp the original sampled data stream to obtain time-series associated sampled data;
[0063] (6) Constructing data packets based on the sampling data associated with the timing to obtain formatted data packets;
[0064] (7) Perform data integrity check on the formatted data packet to obtain a valid data set, and then divide the valid data set into time windows to obtain the original data set containing voltage, current and position information.
[0065] Specifically, a detailed topology diagram including the traction substation, power supply arm, catenary and train location is drawn through the physical layout and electrical connection of the system. In this process, graph theory and network analysis techniques are used to identify the key nodes that have the greatest impact on the operation of the system, such as substation outlets, segmentation points and load concentration areas. These key nodes constitute the key node distribution map, which lays the foundation for subsequent data collection and analysis. According to the key node distribution map, the hierarchical analysis method is used to prioritize the sampling points. This method takes into account multiple factors such as the electrical importance, geographical location and historical fluctuation degree of the node. By establishing a judgment matrix and calculating the weight of each factor, the comprehensive importance score of each node is finally obtained. These scores are arranged in descending order to form a sampling order list to ensure that the most critical nodes obtain priority sampling resources. Next, the sampling frequency parameters are set for each sampling point in the sampling order list. This step adopts an adaptive sampling strategy to dynamically adjust the sampling frequency according to the importance of the node and the fluctuation characteristics of historical data. For nodes with large voltage fluctuations or frequent load changes, a higher sampling frequency is set, such as 1000 times per second; for relatively stable nodes, the sampling frequency can be reduced to 100 times per second to balance data accuracy and system load. These settings are summarized into a sampling configuration plan to guide the subsequent actual sampling process.
[0066] According to the sampling configuration scheme, each sampling point is synchronously triggered for sampling. The synchronous trigger uses GPS timing technology to ensure that all sampling points start data collection at the same time, and the error is controlled at the microsecond level. This high-precision synchronization is essential for analyzing the overall state of the system and the propagation characteristics of voltage fluctuations. During the sampling process, a high-precision analog-to-digital converter is used to convert the analog signal into a digital signal to form the original sampled data stream. Timestamping the original sampled data stream is a key step to ensure data timing. Each data point is assigned a timestamp accurate to milliseconds. This timestamp not only contains the sampling moment, but also records the time of data transmission and processing, providing a time reference for subsequent data analysis and system state reconstruction.
[0067] Data packets are constructed based on the time-series associated sampling data, and multiple node data at the same time are combined into a complete data packet. Each data packet contains sampling time, node identification, voltage value, current value and GPS location information. This structured data format facilitates subsequent data transmission and processing. Finally, the formatted data packet is checked for data integrity, and a cyclic redundancy check (CRC) algorithm is used to detect errors in the data transmission process. The data that passes the check is marked as valid data and constitutes a valid data set. The valid data set is segmented into time windows, and a typical time window may be 100 milliseconds or 1 second, depending on the dynamic characteristics of the system and the analysis requirements. This segmentation method not only retains the timing characteristics of the data, but also provides suitable data units for subsequent time-frequency analysis.
[0068] For example, a subway line is 30 kilometers long, with 5 traction substations and 25 stations. Through topological analysis, 50 key nodes were identified, including substation exits, important segmentation points, and station power supply points. The nodes were prioritized using the hierarchical analysis method, and 30 sampling points were finally selected. The sampling frequency was set to 1000Hz for the top 10 nodes, 500Hz for the middle 10 nodes, and 200Hz for the remaining 10 nodes. In a typical morning peak operation, the system collected about 15,000 raw data points in 1 second. After data packet construction and integrity verification, the effective data rate reached 99.8%. The data was segmented into 100-millisecond time windows, and finally 10 data packets containing complete voltage, current, and location information were obtained, providing a high-quality raw data set for subsequent voltage fluctuation analysis.
[0069] In a specific embodiment, if Figure 2 As shown, the process of executing step S102 may specifically include the following steps:
[0070] S201, performing multi-scale wavelet decomposition on the original data set to obtain wavelet coefficient matrices of different frequency bands, and performing inter-layer correlation analysis on the wavelet coefficient matrix to obtain an inter-layer correlation coefficient;
[0071] S202, adaptively selecting the wavelet basis function according to the inter-layer correlation coefficient to obtain the optimal wavelet basis, and re-decomposing the original data set according to the optimal wavelet basis to obtain the optimized wavelet coefficients;
[0072] S203, performing statistical characteristic analysis on the optimized wavelet coefficients to obtain noise distribution characteristics of each scale, and constructing a nonlinear mapping function according to the noise distribution characteristics to obtain an initial threshold curve;
[0073] S204, performing segmented optimization on the initial threshold curve to obtain a multi-level threshold parameter set, and performing adaptive threshold processing on the wavelet coefficients through the multi-level threshold parameter set to obtain denoised wavelet coefficients;
[0074] S205 , performing inverse wavelet transform on the denoised wavelet coefficients to obtain preliminary denoised data, and performing edge-preserving filtering on the preliminary denoised data to obtain denoised power data.
[0075] It should be noted that an initial wavelet basis function, such as Daubechies wavelet or Symlet wavelet, is selected to perform multi-layer decomposition on the original data. Each layer of decomposition will divide the signal into approximate coefficients and detail coefficients. The approximate coefficients represent the low-frequency part of the signal, and the detail coefficients represent the high-frequency part. Through multiple iterations, the wavelet coefficient matrix of different frequency bands is finally obtained. This matrix contains the characteristics of the signal at different scales and frequencies. Next, the inter-layer correlation analysis of the wavelet coefficient matrix is performed to calculate the correlation coefficient of the coefficients between adjacent levels. The purpose of this step is to evaluate the information redundancy between different decomposition levels and provide a basis for the subsequent wavelet basis selection. The inter-layer correlation coefficient is calculated using the Pearson correlation coefficient method. For the two adjacent layers of coefficients a and b, their correlation is calculated.
[0076] According to the inter-layer correlation coefficient, the wavelet basis function is adaptively selected. The selection process takes into account factors such as the size of the correlation coefficient, energy concentration, and computational complexity. By comparing the performance of different wavelet bases on these indicators, the wavelet base that best suits the current signal characteristics is selected, which is called the optimal wavelet base. Subsequently, the original data set is re-decomposed using this optimal wavelet base to obtain the optimized wavelet coefficients, which better reflect the essential characteristics of the signal. Statistical analysis of the optimized wavelet coefficients is an important part of the denoising process. By calculating the mean, variance, skewness, and kurtosis of each scale coefficient, the noise distribution characteristics are obtained. These characteristics reflect the distribution of noise at different decomposition levels and provide a basis for subsequent threshold setting. Based on these noise distribution characteristics, a nonlinear mapping function is constructed to map the statistical characteristics to the initial threshold to form an initial threshold curve. This mapping process takes into account the local and global characteristics of the signal and can more accurately distinguish between signals and noise.
[0077] The initial threshold curve is optimized in segments, with the goal of using different threshold strategies in different frequency bands and at different signal strengths. The optimization process uses a segmented linear or nonlinear fitting method to adjust the threshold according to the local characteristics of the signal, and finally obtains a multi-level threshold parameter set. This parameter set contains thresholds for different decomposition levels and different signal strengths, and can handle complex noise conditions more flexibly. This multi-level threshold parameter set is used to perform adaptive threshold processing on the wavelet coefficients, and the coefficients less than the threshold are set to zero or shrunk, thereby removing noise and obtaining denoised wavelet coefficients. Finally, the denoised wavelet coefficients are subjected to an inverse wavelet transform to reconstruct the signal and obtain preliminary denoised data. In order to further improve the signal quality, especially to protect the edge information of the signal, the preliminary denoised data is subjected to edge-preserving filtering. This filtering method can retain the mutation characteristics of the signal while removing residual noise, which is particularly important for detecting transient changes in voltage fluctuations. After this series of processing, high-quality denoised power data is finally obtained.
[0078] For example, a key node of a subway line collected 60,000 voltage data points within 1 minute. First, these data were decomposed by 5 layers of wavelets, and Daubechies4 wavelet was used initially. Through inter-layer correlation analysis, it was found that the correlation coefficients of the third and fourth layers were the highest, reaching 0.85, indicating that these two layers contained a lot of redundant information. After comparison, Symlet6 wavelet was selected as the optimal wavelet basis and decomposed again. Statistical analysis of the optimized wavelet coefficients showed that the variance of the first layer (highest frequency) was 0.02 and the variance of the fifth layer (lowest frequency) was 0.001. Based on this, a nonlinear mapping function was constructed to generate the initial threshold curve. Through segmented optimization, a 5-level threshold parameter set was obtained, in which the threshold of the first layer was set to 0.05 and the threshold of the fifth layer was set to 0.002. After applying these thresholds, high-frequency noise was effectively suppressed, and the signal-to-noise ratio of the signal was increased from the original 25dB to 38dB. Finally, through edge-keeping filtering, the characteristics of voltage mutations, such as a 5V voltage drop within 0.1 seconds, are successfully retained. This sophisticated denoising process provides a high-quality data basis for subsequent voltage fluctuation analysis and control, significantly improving the system's detection capability and control accuracy for tiny voltage fluctuations.
[0079] In a specific embodiment, the process of executing step S103 may specifically include the following steps:
[0080] (1) Segment the noise reduction power data according to the train running time to obtain voltage and current data segments in multiple time windows;
[0081] (2) Performing Hilbert-Huang transform on the voltage and current data segments to obtain analytical data containing real and imaginary parts;
[0082] (3) Calculate the instantaneous amplitude and instantaneous phase based on the analytical data to obtain the instantaneous characteristics of voltage and current, and perform empirical mode decomposition on the instantaneous characteristics to obtain embedded modal data;
[0083] (4) Perform Hilbert-Huang transform on the embedded modal data to obtain the time-frequency distribution data of each mode, and calculate the instantaneous frequency based on the time-frequency distribution data to obtain multi-scale frequency components;
[0084] (5) Perform time-frequency energy distribution analysis on multi-scale frequency components to obtain the time-varying characteristics of voltage fluctuations, and extract the main frequency components based on the time-varying characteristics to obtain the instantaneous frequency characteristics;
[0085] (6) Reconstruct the instantaneous frequency characteristics in the time domain to obtain frequency reconstruction data, and use the frequency reconstruction data to perform voltage distribution analysis on the dynamic voltage distribution data and the spatial information of the subway line to obtain dynamic voltage distribution data.
[0086] Specifically, according to the train schedule and real-time location information, the power data is divided into multiple time windows, each window corresponding to different stages of train operation, such as starting, acceleration, constant speed and braking. This segmentation method can capture the voltage and current characteristics under different operating conditions. Next, the Hilbert-Huang transform (HHT) is performed on the voltage and current data segments of each time window. HHT is a method suitable for nonlinear and non-stationary signal analysis, which includes two main steps: empirical mode decomposition (EMD) and Hilbert spectrum analysis. First, the signal is decomposed into a series of intrinsic mode functions (IMFs) through EMD. Then, each IMF is Hilbert transformed to obtain an analytical signal containing real and imaginary parts. The expression of the analytical signal is:
[0087] ;
[0088] in, is the original signal, yes The Hilbert transform of .
[0089] Based on the analytical signal, calculate the instantaneous amplitude and instantaneous phase. and instantaneous phase The calculation formula is as follows:
[0090] ;
[0091] ;
[0092] These calculation results constitute the transient characteristics of voltage and current. Subsequently, the transient characteristics are subjected to empirical mode decomposition to obtain a series of embedded mode functions (IMFs). The EMD process repeatedly extracts the local extreme points of the signal, constructs the upper and lower envelopes, calculates the mean envelope, and subtracts the mean envelope from the original signal until an IMF is obtained. Repeating this process, multiple IMFs are obtained, which constitute the embedded mode data.
[0093] The embedded modal data is transformed again with Hilbert-Huang transform to obtain the time-frequency distribution data of each mode. The multi-scale frequency components are obtained by calculating the instantaneous frequency, that is, the time derivative of the phase:
[0094] ;
[0095] The time-frequency energy distribution of multi-scale frequency components is analyzed, and the energy density of each time-frequency point is calculated to obtain the time-varying characteristics of voltage fluctuations. By setting the energy threshold, the main frequency components are extracted to form instantaneous frequency characteristics. Finally, the instantaneous frequency characteristics are reconstructed in the time domain, and the frequency domain information is converted back to the time domain to obtain frequency reconstruction data. Combined with the spatial information of the subway line, such as station location, power supply segmentation point, etc., the dynamic voltage distribution data is spatially mapped and interpolated to finally obtain complete dynamic voltage distribution data.
[0096] For example, a subway line is 30 kilometers long and contains 20 stations. During the morning rush hour, voltage data of a train from the starting station to the terminal station is collected with a sampling frequency of 1kHz, and a total of 1.8 million data points are obtained. First, the data is divided into 60 time windows according to the train operation stage, and each window is 30 seconds long. The data of the 10th time window (corresponding to the train acceleration stage) is analyzed by Hilbert-Huang transform. Five IMFs are obtained by EMD decomposition, of which the frequency range of the first IMF is 10-50Hz, reflecting the switching frequency characteristics of the traction system. The instantaneous amplitude is calculated and it is found that the voltage fluctuation amplitude is within the range of ±20V. Through time-frequency energy distribution analysis, a major low-frequency oscillation component with a frequency of about 2Hz is identified, corresponding to the load fluctuation during the train acceleration process. Combining these features with spatial information, it is found that the voltage fluctuation amplitude reaches a maximum of 25V in an uphill section 5 kilometers away from the traction substation. This detailed spatiotemporal analysis result provides an accurate basis for the formulation of voltage fluctuation control strategies, such as adding reactive power compensation devices at the location or adjusting the output voltage of the traction substation, thereby effectively suppressing voltage fluctuations and improving the stability and energy efficiency of the power supply system.
[0097] In a specific embodiment, the process of executing step S104 may specifically include the following steps:
[0098] (1) Perform time series segmentation on the dynamic voltage distribution data to obtain voltage change data in multiple time windows, and perform statistical analysis on the voltage change data in multiple time windows to obtain voltage fluctuation characteristic parameters;
[0099] (2) Classify the voltage change trend according to the voltage fluctuation characteristic parameters to obtain the voltage fluctuation pattern;
[0100] (3) Perform spectral decomposition on the instantaneous frequency characteristics to obtain the main frequency components and amplitude data, and calculate the instantaneous impedance of the subway DC power supply system based on the main frequency components and amplitude data to obtain the impedance characteristic curve;
[0101] (4) Correcting the voltage fluctuation pattern through the impedance characteristic curve to obtain corrected voltage fluctuation data;
[0102] (5) Correlation analysis is performed on the corrected voltage fluctuation data and the real-time collected train operation status data to obtain the corresponding relationship between the grid voltage data and the train position;
[0103] (6) Calculate the electric brake power at different positions based on the corresponding relationship between the grid voltage data and the train position to obtain the spatially distributed electric brake power data, and perform normalization and curve mapping on the spatially distributed electric brake power data to obtain a standardized electric brake power curve;
[0104] (7) Fit the grid voltage data to the standardized electric brake power curve to obtain the voltage control characteristic curve.
[0105] Specifically, the sliding window technology is used to divide the continuous voltage data into multiple overlapping time windows. The length of each window is usually a few seconds to a few minutes, depending on the specific analysis requirements. The voltage change data in each time window is statistically analyzed to calculate statistical parameters such as the mean, standard deviation, and peak-to-peak value. These parameters constitute the voltage fluctuation characteristic parameter set. The calculation formula for the voltage fluctuation characteristic parameters is as follows:
[0106] ;
[0107] ;
[0108] in, is the mean, is the standard deviation, For the i The voltage value of the sampling point, Nis the number of sampling points in the window. Based on the calculated voltage fluctuation characteristic parameters, a clustering algorithm (such as K-means or hierarchical clustering) is used to classify the voltage change trend to obtain different voltage fluctuation modes. These modes may include stable mode, fluctuation mode, mutation mode, etc. Next, the instantaneous frequency characteristics are spectrally decomposed, and the main frequency components and their amplitude data are obtained by fast Fourier transform (FFT) or wavelet transform. Based on these data, the instantaneous impedance of the subway DC power supply system is calculated. The calculation formula for instantaneous impedance is:
[0109] ;
[0110] in, For frequency The impedance at V(f) and I(f) are the voltage and current amplitudes at this frequency, respectively. By calculating the impedance values at different frequency points, an impedance characteristic curve is drawn. The previously identified voltage fluctuation pattern is corrected using the obtained impedance characteristic curve. The correction process takes into account the frequency response characteristics of the system and can more accurately reflect the actual voltage fluctuation. The corrected voltage fluctuation data is correlated with the real-time collected train operation status data (such as position, speed, acceleration, etc.) and a corresponding relationship model between the network voltage data and the train position is established using methods such as multivariate regression or neural network.
[0111] Based on the established corresponding relationship model, the electric braking power at different positions is calculated. The calculation formula of the electric braking power is:
[0112] ;
[0113] in, P is the electric braking power, For network pressure, is the braking current. The calculated electric brake power data forms a spatially distributed electric brake power data set. This data set is normalized, and the power value is mapped to the [0,1] interval to obtain a standardized electric brake power curve. Finally, the standardized electric brake power curve is fitted with the grid voltage data. The fitting process uses methods such as polynomial regression or spline interpolation to obtain the voltage control characteristic curve. This curve describes the relationship between grid voltage and electric brake power, and is an important basis for the formulation of subsequent control strategies.
[0114] For example, a subway line is 25 kilometers long and contains 15 stations. In a typical morning peak operation, voltage data is collected from the starting station to the terminal station of a train, with a sampling frequency of 1kHz, and a total of 9 million data points are obtained. First, the data is divided into 30-second time windows to obtain 500 time windows. The voltage mean and standard deviation are calculated for each window, and it is found that the mean fluctuates within the range of DC 1500V±50V, and the standard deviation varies between 0.5V and 15V. Through the K-means clustering algorithm, these windows are divided into three types of voltage fluctuation modes: stable mode (standard deviation <2V), fluctuation mode (2V≤standard deviation <10V) and violent fluctuation mode (standard deviation ≥10V). FFT analysis is performed on the instantaneous frequency characteristics to identify three main frequency components: 1.5Hz (corresponding to vehicle mechanical vibration), 50Hz (grid base frequency) and 300Hz (rectifier switching frequency). The system impedance is calculated based on these frequency components to obtain the impedance characteristic curve. The curve is used to correct the voltage fluctuation pattern, especially the high-frequency fluctuation around 300Hz is described more accurately.
[0115] The corrected voltage fluctuation data was correlated with the train position data, and it was found that the voltage drop was most obvious in a long uphill section 7 kilometers away from the traction substation, reaching a minimum of DC 1420V. Based on this relationship, the maximum electric brake power at this location was calculated to be 2.5MW. The electric brake power data of the entire line was normalized to obtain a standardized electric brake power curve. Finally, the voltage control characteristic curve was obtained through cubic spline interpolation, which clearly showed that in the range of DC 1450V-1550V, the electric brake power and the grid voltage showed a nonlinear positive correlation. This detailed analysis result provides an important basis for formulating accurate voltage control strategies, such as adding energy storage devices at key locations or adjusting the output characteristics of traction substations, so as to effectively control voltage fluctuations and improve the energy utilization efficiency and operation stability of the system.
[0116] In a specific embodiment, the process of executing step S105 may specifically include the following steps:
[0117] (1) Performing piecewise linearization processing on the voltage control characteristic curve to obtain multiple linear curves, and calculating the slope of each segment of the multiple linear curves to obtain an initial slope set;
[0118] (2) Perform statistical analysis on the initial slope set to obtain slope distribution characteristics, and determine the target slope range based on the slope distribution characteristics to obtain slope constraints;
[0119] (3) The slope constraint condition is combined with the preset time constraint to obtain the optimization objective function, and the optimization objective function is solved by the gradient descent method to obtain the candidate slope value;
[0120] (4) Discretize the candidate slope values to obtain a discrete slope sequence, and perform interpolation fitting on the discrete slope sequence to obtain a continuous slope curve;
[0121] (5) Reconstructing the voltage control characteristic curve according to the continuous slope curve to obtain an optimized voltage control curve, and performing an electric braking force change rate analysis on the optimized voltage control curve to obtain the electric braking force change rate;
[0122] (6) Perform cross-correlation analysis on the electric braking force change rate and the train running status data to obtain the relationship matrix between the electric braking force and the train speed. Based on the relationship matrix, the electric braking force speed curve is constructed to obtain the electric braking force control characteristic data.
[0123] (7) Performing piecewise linearization processing on the electric brake force control characteristic data to obtain a piecewise control strategy;
[0124] (8) Calculate the braking time of each section according to the segmented control strategy to obtain the cumulative braking time series;
[0125] (9) Compare the accumulated braking time series with the preset time constraint to obtain the time deviation value;
[0126] (10) The segmented control strategy is fine-tuned through the iterative adjustment method to obtain an electric braking force control scheme that meets the preset time constraints.
[0127] Specifically, the piecewise linearization of the voltage control characteristic curve is a key step in the voltage fluctuation control method of the subway DC power supply system. First, the nonlinear voltage control characteristic curve is divided into multiple linear segments using a piecewise linear fitting algorithm. Each linear segment can be represented by a straight line equation:
[0128] ;
[0129] in, is the slope, is the intercept. Calculate the slope of each linear curve to get the initial slope set ; Perform statistical analysis on the initial slope set and calculate the mean and standard deviation :
[0130] ;
[0131] ;
[0132] Determine the target slope range based on the slope distribution characteristics , this range is usually set to , as the slope constraint. Combine the slope constraint with the preset time constraint Combined, construct the optimization objective function:
[0133] ;
[0134] in, is the slope value to be optimized, is the braking time under the corresponding slope, is the weight coefficient. The optimization problem is solved by the gradient descent method to obtain the candidate slope value. The candidate slope value is discretized to obtain a discrete slope sequence, and then the discrete slope sequence is fitted using the cubic spline interpolation method to obtain a continuous slope curve. The voltage control characteristic curve is reconstructed according to the continuous slope curve to obtain the optimized voltage control curve. The electric braking force change rate is analyzed on the optimized voltage control curve to calculate the change rate of the electric braking force over time. The electric braking force change rate is cross-correlated with the train operation status data (such as speed and acceleration) to obtain the relationship matrix between the electric braking force and the train speed. Based on this matrix, the electric braking force speed curve is constructed to form the electric braking force control characteristic data. These data are piecewise linearized again to obtain the segmented control strategy. According to the segmented control strategy, the braking time of each segment is calculated and accumulated to obtain the cumulative braking time series. The cumulative braking time series is compared with the preset time constraint to calculate the time deviation value. Finally, the segmented control strategy is fine-tuned by the iterative adjustment method to continuously reduce the time deviation until the preset time constraint is met to obtain the final electric braking force control scheme.
[0135] For example, a subway line is 20 kilometers long. During the process of the train decelerating from 80 km / h to stopping, the original voltage control characteristic curve is divided into 5 sections, and the initial slope set is {-0.5, -0.8, -1.2, -1.5, -1.8}. Statistical analysis shows that the mean slope is -1.16, the standard deviation is 0.49, and the target slope range is determined to be [-2.14, -0.18]. The preset braking time constraint is 25 seconds. The optimization objective function is constructed, and the candidate slope value -1.1 is obtained by 50 iterations of the gradient descent method. Discretization and interpolation fitting are performed to obtain a continuous slope curve. The reconstructed optimized voltage control curve shows that in the 80-60 km / h range, the electric braking force change rate is -15 kN / s; in the 60-40 km / h range, it is -12 kN / s; in the 40-20 km / h range, it is -8 kN / s; and in the 20-0 km / h range, it is -5 kN / s. The data were cross-correlated with the train running status to obtain the electric braking force-speed relationship matrix, and a segmented control strategy was constructed based on it. The cumulative braking time calculated initially was 26.5 seconds, which exceeded the preset constraint by 1.5 seconds. After 5 iterative adjustments, the electric braking force control scheme that met the 25-second constraint was finally obtained: 80-55km / h uses the maximum electric braking force; 55-30km / h linearly reduces the electric braking force; 30-0km / h maintains a constant low electric braking force.
[0136] In a specific embodiment, the process of executing step S106 may specifically include the following steps:
[0137] (1) Divide the system topology into nodes to obtain multiple sub-network areas, calculate the power flow distribution of multiple sub-network areas, and obtain the initial energy allocation plan;
[0138] (2) Perform load balancing analysis on each sub-network area according to the initial energy allocation scheme to obtain a load imbalance index, and make a threshold judgment on the load imbalance index to obtain a set of key nodes to be adjusted;
[0139] (3) Construct an energy transfer matrix based on the key node set to obtain the energy transfer relationship between nodes, set constraints on the energy transfer relationship, and obtain the energy allocation optimization target;
[0140] (4) Solve the energy allocation optimization objective through the distributed gradient projection algorithm to obtain the energy allocation ratio of each node;
[0141] (5) Verify the feasibility of the energy allocation ratio of each node, obtain a set of alternative energy allocation schemes, and perform multi-objective trade-off analysis based on the set of alternative energy allocation schemes to obtain the optimal energy allocation strategy;
[0142] (6) Discretize the optimal energy allocation strategy to obtain the energy allocation sequence for each time period, and calculate the voltage change trend of each node based on the energy allocation sequence for each time period to obtain the node voltage prediction curve;
[0143] (7) Map the electric braking force control scheme to the current demand curve to obtain the train load characteristic curve;
[0144] (8) Performing time-space matching on the node voltage prediction curve and the train load characteristic curve to obtain a dynamic load distribution diagram, and calculating the voltage margin of each power supply section based on the dynamic load distribution diagram to obtain the voltage regulation target value;
[0145] (9) The voltage regulation target value is corrected in real time to obtain the global voltage stability control solution.
[0146] Specifically, according to the physical structure of the power supply network, the entire system is divided into multiple sub-network areas, each of which usually includes a traction substation and the lines and stations within its power supply range. The power flow distribution of these sub-network areas is calculated, and the node voltage method is used to establish the power flow equation:
[0147] ;
[0148] in, is the node admittance matrix, is the node voltage vector, is the injected current vector. Solving this equation yields the initial energy distribution scheme.
[0149] According to the initial energy allocation plan, calculate the load rate of each sub-network area and define the load imbalance index :
[0150] ;
[0151] in, For the i The load rate of each subnetwork. Set the threshold ,when When the load rate is the highest and the lowest, the areas are identified and their key nodes are included in the set to be adjusted. Based on the set of key nodes to be adjusted, the energy transmission matrix is constructed. , whose elements Represents a slave node To Node Energy transmission capacity. Set constraints such as transmission capacity and line loss to form the energy allocation optimization objective function:
[0152] ;
[0153] in, For Node arrive The transmission power, is the unit transmission cost, is the balance factor, is the average load rate. The distributed gradient projection algorithm is used to solve this optimization problem. k Update its energy distribution ratio according to local information:
[0154] ;
[0155] in, is the projection operator, which projects the solution into the feasible domain Inside, is the energy allocation ratio of node k after the t+1th iteration. For Node k Energy allocation ratio at the tth iteration. is the step size of the tth iteration, which controls the magnitude of each update and usually decreases as the number of iterations increases.
[0156] is the objective function exist The gradient at , indicating the node k The optimal adjustment direction based on the current state. is the local objective function of node k, usually the component of the global objective function at this node. The feasibility of the obtained energy allocation ratio is verified to check whether all constraints are met. Through multi-objective trade-off analysis, considering factors such as energy utilization efficiency, voltage stability and equipment life, the optimal energy allocation strategy is selected from the alternatives.
[0157] The optimal energy allocation strategy is discretized into a time series, and the voltage change trend of each node in each time period is calculated to obtain the node voltage prediction curve. At the same time, the electric braking force control scheme is converted into a current demand curve to obtain the train load characteristic curve. The node voltage prediction curve and the train load characteristic curve are matched in time and space to generate a dynamic load distribution diagram. Based on this distribution diagram, the voltage margin of each power supply interval is calculated, which is defined as the difference between the actual voltage and the minimum allowable voltage. According to the voltage margin, the voltage regulation target value is set and corrected through real-time feedback, and finally a global voltage stability control scheme is formed.
[0158] For example, a subway line is 30 kilometers long, including 3 traction substations and 15 stations. The system is divided into 3 subnetwork areas, each of which corresponds to the power supply range of a traction substation. Through power flow calculation, the initial energy distribution plan is obtained: the load rates of areas 1, 2, and 3 are 65%, 80%, and 50%, respectively. Calculate the load imbalance index η =0.3, which exceeds the preset threshold of 0.2, so the key nodes in areas 2 and 3 are included in the set to be adjusted.
[0159] Construct a 3×3 energy transmission matrix and set the upper limit of transmission capacity to 5MW. In the optimization objective function, set the unit transmission cost to 0.1 yuan / kWh and the balance factor = 0.5. Using the distributed gradient projection algorithm, after 50 iterations, the energy allocation ratio of each node was obtained. After feasibility verification and multi-objective trade-off analysis, the final energy allocation strategy was determined as: transfer 2MW power from area 3 to area 2. This strategy was discretized into an energy allocation sequence with a 5-minute interval, and the node voltage prediction curve was calculated. At the same time, the train load characteristic curve was generated according to the train schedule and the electric braking force control scheme. Through time and space matching, it was found that during the morning peak period of 7:30-8:00, the voltage margin of the section from the 8th station to the 10th station in area 2 was the lowest, only 50V. Based on this, the voltage regulation target value of this section was set to increase by 30V. Finally, these regulation targets were incorporated into the global control scheme, and the voltage stability control of the entire line was achieved by adjusting the output of the traction substation and the charging and discharging strategy of the mobile energy storage device in real time, and the voltage fluctuation amplitude was controlled within ±5% of the rated value. At the same time, the utilization rate of regenerative braking energy was improved to more than 95%.
[0160] The embodiment of the present invention also provides a voltage fluctuation control system for a subway DC power supply system, such as Figure 3 As shown, the voltage fluctuation control system of the subway DC power supply system specifically includes:
[0161] The sampling module 301 is used to perform high-frequency sampling of node data of the subway DC power supply system to obtain an original data set including voltage, current and position information;
[0162] The processing module 302 is used to perform wavelet threshold denoising on the original data set to obtain denoised power data;
[0163] An analysis module 303 is used to perform time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data;
[0164] A calculation module 304 is used to calculate the corresponding relationship between the grid voltage data and the electric braking power of the subway DC power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics, and obtain a voltage control characteristic curve;
[0165] An optimization module 305 is used to perform a constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets a preset time constraint;
[0166] The allocation module 306 is used to perform energy allocation strategy analysis on the system topology of the subway DC power supply system based on the electric braking force control scheme through a distributed optimization algorithm to obtain a global voltage stability control scheme.
[0167] Through the collaborative work of the above modules, the original data set containing voltage, current and position information is obtained by high-frequency sampling of multiple key nodes preset in the system, which effectively improves the accuracy and real-time performance of voltage fluctuation monitoring. The original data set is subjected to wavelet threshold denoising, which not only effectively removes the noise interference in the signal, but also retains the key characteristics of voltage fluctuation, significantly improving the data quality. The de-noised power data is subjected to time-frequency analysis through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data. This step deeply reveals the time-frequency characteristics of voltage fluctuation and provides a theoretical basis for precise control. According to the dynamic voltage distribution data and instantaneous frequency characteristics, the corresponding relationship between the grid voltage and the electric braking power is calculated to obtain the voltage control characteristic curve, establish a direct connection between the voltage fluctuation and the system operation status, and lay the foundation for the formulation of an effective control strategy. The constant slope optimization analysis of the voltage control characteristic curve is performed to obtain an electric braking force control scheme that meets the preset time constraints. This innovative optimization process not only ensures the stability of the electric braking process, but also maximizes the energy recovery efficiency. Finally, based on the electric braking force control scheme, the energy distribution strategy of the system topology is analyzed through the distributed optimization algorithm, and the global voltage stability control scheme is obtained, which fully considers the integrity and complexity of the system, realizes the global coordinated control of voltage fluctuations, significantly improves the voltage stability of the subway DC power supply system, and reduces the adverse effects of voltage fluctuations on system operation. At the same time, by optimizing the electric braking process and energy distribution strategy, the energy utilization efficiency of the system is greatly improved and energy loss is reduced.
[0168] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the embodiments, a person skilled in the art should understand that the specific implementation modes of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.
Claims
1. A method for controlling voltage fluctuations in a subway DC power supply system, characterized in that: include: Perform high-frequency sampling of node data in the subway DC power supply system to obtain the original data set containing voltage, current and location information; Performing wavelet threshold denoising processing on the original data set to obtain denoised power data; Performing time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data; According to the dynamic voltage distribution data and the instantaneous frequency characteristics, the corresponding relationship between the grid voltage data and the electric braking power of the subway DC power supply system is calculated to obtain a voltage control characteristic curve, which specifically includes: performing time series segmentation on the dynamic voltage distribution data to obtain voltage change data of multiple time windows, and performing statistical analysis on the voltage change data of the multiple time windows to obtain voltage fluctuation characteristic parameters; classifying the voltage change trend according to the voltage fluctuation characteristic parameters to obtain a voltage fluctuation mode; performing spectral decomposition on the instantaneous frequency characteristics to obtain main frequency components and amplitude data, and calculating the subway DC power supply according to the main frequency components and amplitude data. The instantaneous impedance of the system is obtained to obtain an impedance characteristic curve; the voltage fluctuation mode is corrected by the impedance characteristic curve to obtain corrected voltage fluctuation data; the corrected voltage fluctuation data is correlated with the real-time collected train operation status data to obtain the corresponding relationship between the grid voltage data and the train position; the electric braking power at different positions is calculated according to the corresponding relationship between the grid voltage data and the train position to obtain the spatially distributed electric braking power data, and the spatially distributed electric braking power data is normalized and curve mapped to obtain a standardized electric braking power curve; the standardized electric braking power curve is fitted with the grid voltage data to obtain a voltage control characteristic curve; The voltage control characteristic curve is subjected to constant slope optimization analysis to obtain an electric braking force control scheme that satisfies a preset time constraint, specifically comprising: performing piecewise linearization processing on the voltage control characteristic curve to obtain a plurality of linear curves, and calculating the slope of each segment of the plurality of linear curves to obtain an initial slope set; performing statistical analysis on the initial slope set to obtain slope distribution characteristics, and determining a target slope range according to the slope distribution characteristics to obtain a slope constraint condition; combining the slope constraint condition with a preset time constraint to obtain an optimization objective function, and solving the optimization objective function by a gradient descent method to obtain a candidate slope value; discretizing the candidate slope values to obtain a discrete slope sequence, and performing interpolation fitting on the discrete slope sequence to obtain a continuous slope curve; and determining a target slope range according to the continuous slope constraint condition. The slope curve reconstructs the voltage control characteristic curve to obtain an optimized voltage control curve, and performs an electric braking force change rate analysis on the optimized voltage control curve to obtain the electric braking force change rate; the electric braking force change rate and the train operation status data are cross-correlatedly analyzed to obtain a relationship matrix between the electric braking force and the train speed, and an electric braking force speed curve is constructed according to the relationship matrix to obtain electric braking force control characteristic data; the electric braking force control characteristic data is subjected to piecewise linearization processing to obtain a piecewise control strategy; the braking time of each section is calculated according to the piecewise control strategy to obtain a cumulative braking time series; the cumulative braking time series is compared with a preset time constraint to obtain a time deviation value; the piecewise control strategy is fine-tuned by an iterative adjustment method to obtain an electric braking force control scheme that meets the preset time constraint; Based on the electric braking force control scheme, an energy distribution strategy analysis is performed on the system topology of the subway DC power supply system through a distributed optimization algorithm to obtain a global voltage stability control scheme.
2. The voltage fluctuation control method of a subway DC power supply system according to claim 1, characterized in that: The step of performing high-frequency sampling of node data of the subway DC power supply system to obtain an original data set containing voltage, current and position information includes: Analyze the system topology of the subway DC power supply system to obtain a key node distribution diagram; Prioritize the sampling points according to the key node distribution map to obtain a sampling order list; Setting a sampling frequency parameter for each sampling point in the sampling sequence list to obtain a sampling configuration scheme; Perform synchronous trigger sampling on each sampling point according to the sampling configuration scheme to obtain an original sampling data stream; Performing time stamp marking on the original sampled data stream to obtain time-series associated sampled data; Constructing a data packet according to the sampling data associated with the time sequence to obtain a formatted data packet; The formatted data packet is subjected to a data integrity check to obtain a valid data set, and the valid data set is subjected to a time window segmentation to obtain an original data set containing voltage, current and position information.
3. The voltage fluctuation control method of a subway DC power supply system according to claim 1, characterized in that: The step of performing wavelet threshold denoising on the original data set to obtain denoised power data comprises: Performing multi-scale wavelet decomposition on the original data set to obtain wavelet coefficient matrices of different frequency bands, and performing inter-layer correlation analysis on the wavelet coefficient matrix to obtain an inter-layer correlation coefficient; Adaptively selecting a wavelet basis function according to the inter-layer correlation coefficient to obtain an optimal wavelet basis, and re-decomposing the original data set according to the optimal wavelet basis to obtain optimized wavelet coefficients; Performing statistical characteristic analysis on the optimized wavelet coefficients to obtain noise distribution characteristics of each scale, and constructing a nonlinear mapping function according to the noise distribution characteristics to obtain an initial threshold curve; The initial threshold curve is optimized in sections to obtain a multi-level threshold parameter set, and the wavelet coefficients are subjected to adaptive threshold processing by using the multi-level threshold parameter set to obtain denoised wavelet coefficients; The denoised wavelet coefficients are subjected to inverse wavelet transform to obtain preliminary denoised data, and the preliminary denoised data are subjected to edge-preserving filtering to obtain denoised power data.
4. The voltage fluctuation control method of a subway DC power supply system according to claim 1, characterized in that: The step of performing time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data includes: Segmenting the noise reduction power data according to the train running time to obtain voltage and current data segments of multiple time windows; Performing a Hilbert-Huang transform on the voltage and current data segments to obtain analytical data including real and imaginary parts; Calculating the instantaneous amplitude and the instantaneous phase according to the analytical data to obtain the instantaneous characteristics of the voltage and current, and performing empirical mode decomposition on the instantaneous characteristics to obtain embedded mode data; Performing Hilbert-Huang transform on the embedded modal data to obtain time-frequency distribution data of each mode, and calculating the instantaneous frequency according to the time-frequency distribution data to obtain multi-scale frequency components; Performing time-frequency energy distribution analysis on the multi-scale frequency components to obtain time-varying characteristics of voltage fluctuations, and extracting main frequency components according to the time-varying characteristics to obtain the instantaneous frequency characteristics; The instantaneous frequency characteristics are reconstructed in the time domain to obtain frequency reconstruction data, and voltage distribution analysis is performed on the dynamic voltage distribution data and the spatial information of the subway line through the frequency reconstruction data to obtain the dynamic voltage distribution data.
5. The voltage fluctuation control method of a subway DC power supply system according to claim 1, characterized in that: The step of performing energy distribution strategy analysis on the system topology of the subway DC power supply system based on the electric braking force control scheme by using a distributed optimization algorithm to obtain a global voltage stability control scheme includes: Performing node division on the system topology structure to obtain a plurality of sub-network areas, and calculating the power flow distribution of the plurality of sub-network areas to obtain an initial energy allocation scheme; Performing load balancing analysis on each sub-network area according to the initial energy allocation scheme to obtain a load imbalance index, and performing threshold judgment on the load imbalance index to obtain a set of key nodes to be adjusted; An energy transfer matrix is constructed according to the key node set to obtain the energy transfer relationship between the nodes, and constraint conditions are set on the energy transfer relationship to obtain an energy allocation optimization target; The energy allocation optimization target is solved by a distributed gradient projection algorithm to obtain the energy allocation ratio of each node; Performing feasibility verification on the energy allocation ratio of each node to obtain a set of alternative energy allocation solutions, and performing multi-objective trade-off analysis based on the set of alternative energy allocation solutions to obtain an optimal energy allocation strategy; Discretizing the optimal energy allocation strategy to obtain an energy allocation sequence for each time period, and calculating a voltage change trend of each node according to the energy allocation sequence for each time period to obtain a node voltage prediction curve; Mapping the electric braking force control scheme to a current demand curve to obtain a train load characteristic curve; Performing time-space matching on the node voltage prediction curve and the train load characteristic curve to obtain a dynamic load distribution diagram, and calculating the voltage margin of each power supply section according to the dynamic load distribution diagram to obtain a voltage regulation target value; The voltage regulation target value is corrected in real time to obtain the global voltage stabilization control solution.
6. A voltage fluctuation control system for a subway DC power supply system, used to execute the voltage fluctuation control method for a subway DC power supply system according to any one of claims 1 to 5, characterized in that: include: The sampling module is used to perform high-frequency sampling of node data in the subway DC power supply system to obtain an original data set containing voltage, current and position information; A processing module, used for performing wavelet threshold denoising processing on the original data set to obtain denoised power data; An analysis module, used for performing time-frequency analysis on the noise-reduced power data through Hilbert-Huang transform to obtain instantaneous frequency characteristics and dynamic voltage distribution data; The calculation module is used to calculate the corresponding relationship between the grid voltage data and the electric braking power of the subway DC power supply system according to the dynamic voltage distribution data and the instantaneous frequency characteristics, and obtain the voltage control characteristic curve, which specifically includes: performing time series segmentation on the dynamic voltage distribution data to obtain voltage change data of multiple time windows, and performing statistical analysis on the voltage change data of the multiple time windows to obtain voltage fluctuation characteristic parameters; classifying the voltage change trend according to the voltage fluctuation characteristic parameters to obtain the voltage fluctuation mode; performing spectrum decomposition on the instantaneous frequency characteristics to obtain the main frequency components and amplitude data, and calculating the subway according to the main frequency components and amplitude data. The instantaneous impedance of the DC power supply system is obtained to obtain an impedance characteristic curve; the voltage fluctuation mode is corrected by the impedance characteristic curve to obtain corrected voltage fluctuation data; the corrected voltage fluctuation data is correlated with the real-time collected train operation status data to obtain the corresponding relationship between the grid voltage data and the train position; the electric braking power at different positions is calculated according to the corresponding relationship between the grid voltage data and the train position to obtain spatially distributed electric braking power data, and the spatially distributed electric braking power data is normalized and curve mapped to obtain a standardized electric braking power curve; the standardized electric braking power curve is fitted with grid voltage data to obtain a voltage control characteristic curve; The optimization module is used to perform constant slope optimization analysis on the voltage control characteristic curve to obtain an electric braking force control scheme that meets the preset time constraint, specifically comprising: performing piecewise linearization processing on the voltage control characteristic curve to obtain a multi-segment linear curve, and calculating the slope of each segment of the multi-segment linear curve to obtain an initial slope set; performing statistical analysis on the initial slope set to obtain slope distribution characteristics, and determining a target slope range based on the slope distribution characteristics to obtain a slope constraint condition; combining the slope constraint condition with the preset time constraint to obtain an optimization objective function, and solving the optimization objective function by a gradient descent method to obtain a candidate slope value; discretizing the candidate slope value to obtain a discrete slope sequence, and performing interpolation fitting on the discrete slope sequence to obtain a continuous slope curve; The continuous slope curve is used to reconstruct the voltage control characteristic curve to obtain an optimized voltage control curve, and the electric braking force change rate is analyzed on the optimized voltage control curve to obtain the electric braking force change rate; the electric braking force change rate and the train operation status data are cross-correlatedly analyzed to obtain a relationship matrix between the electric braking force and the train speed, and an electric braking force speed curve is constructed according to the relationship matrix to obtain electric braking force control characteristic data; the electric braking force control characteristic data is subjected to piecewise linearization processing to obtain a piecewise control strategy; the braking time of each section is calculated according to the piecewise control strategy to obtain a cumulative braking time series; the cumulative braking time series is compared with a preset time constraint to obtain a time deviation value; the piecewise control strategy is fine-tuned by an iterative adjustment method to obtain an electric braking force control scheme that meets the preset time constraint; The distribution module is used to analyze the energy distribution strategy of the system topology of the subway DC power supply system based on the electric braking force control scheme through a distributed optimization algorithm to obtain a global voltage stability control scheme.
