Electricity utilization characteristic analysis method for high-energy-consumption equipment of converter station

Abstract

The invention relates to a converter station high-energy-consumption equipment power consumption characteristic analysis method, and belongs to the field of converter station high-energy-consumption equipment power consumption analysis. The method comprises the steps of firstly collecting historical power consumption data of high-energy-consumption equipment, then performing noise reduction and feature extraction on the data by adopting a noise reduction Bayesian optimization LSTM model, and adjusting model hyper-parameters through a Bayesian optimization algorithm to improve model performance. The method comprises the following steps of: preprocessing time sequence data by using variational mode decomposition, and carrying out signal noise reduction; a load characteristic index library is constructed, and influences of external factors such as temperature and humidity on power utilization of equipment are analyzed, so that power utilization characteristics of the equipment are deeply understood. Through the mode, the method can more accurately understand the use condition of the equipment, discover possible energy efficiency problems and provide scientific basis for energy management so as to achieve the purposes of energy conservation and emission reduction.

Description

Technical Field

[0001] The present invention belongs to the field of power consumption analysis of high-energy-consuming equipment in converter stations, and more specifically relates to a method for analyzing power consumption characteristics of high-energy-consuming equipment in converter stations. Background Art

[0002] In energy management, analyzing the power usage characteristics of high-energy-consuming equipment is crucial for understanding their operation and achieving energy conservation and emission reduction. However, existing methods for analyzing the power usage characteristics of high-energy-consuming equipment primarily rely on empirical or manual analysis, lacking scientific and effective data processing and feature extraction techniques. These methods are unable to deeply explore the characteristics and patterns of equipment power usage, nor can they accurately assess the impact of external environmental factors such as temperature and humidity on equipment power usage. Consequently, they provide limited support for addressing equipment energy efficiency and energy-saving optimization.

[0003] In recent years, deep learning techniques have demonstrated strong performance in time series data processing and feature extraction. Long short-term memory (LSTM) networks, in particular, have proven to be an effective solution for processing data with time-delayed features. However, the performance of LSTM models is often affected by model hyperparameters, making selecting appropriate hyperparameters a challenge.

[0004] Furthermore, traditional data noise reduction methods, such as sliding average and convolution, often lose important information in the data and are not well-suited for processing power consumption data from high-energy-consuming devices. Currently, there is a lack of a data analysis method that can accurately mine and extract the power consumption characteristics of high-energy-consuming devices.

[0005] Therefore, a new method for analyzing the power consumption characteristics of high-energy-consuming equipment in converter stations is urgently needed. Through scientific and effective data processing and feature extraction technology, the power consumption characteristics and patterns of equipment can be deeply explored, and the impact of external environmental factors on equipment power consumption can be accurately evaluated, providing stronger support for the energy efficiency issues and energy-saving optimization of equipment. Summary of the Invention

[0006] The main technical problem to be solved by the present invention is how to use scientific and effective data processing and feature extraction technologies to deeply analyze and understand the power consumption characteristics and patterns of high-energy-consuming equipment, especially converter station equipment, and accurately evaluate and understand the impact of external environmental factors such as temperature and humidity on equipment power consumption, thereby providing stronger support to solve the energy efficiency problems of equipment, optimize energy-saving measures, and achieve efficient operation of converter stations.

[0007] In order to achieve the above object, the present invention is implemented by adopting the following technical solutions: the method comprises:

[0008] Collect historical electricity consumption data of high-energy-consuming equipment, including air conditioners, valve cooling, and converter coolers;

[0009] A denoising Bayesian optimization LSTM model is used to perform denoising and feature extraction on the data, and the hyperparameters of the LSTM model are adjusted using a Bayesian optimization algorithm to improve the performance of the model;

[0010] The time series data is preprocessed using variational mode decomposition and decomposed into several intrinsic mode functions to achieve signal noise reduction;

[0011] Build a load characteristic index library, analyze the impact of external factors such as temperature and humidity on equipment power consumption, and provide an in-depth understanding of equipment power consumption characteristics.

[0012] In one solution, the denoising Bayesian optimization LSTM model is used to perform denoising on the electricity consumption data of high-energy-consuming equipment, thereby improving the accuracy of data feature extraction by filtering out measurement errors and abnormal data. The model can capture the time series characteristics in the data, model and extract features of the time dependence of the electricity consumption data of high-energy-consuming equipment, thereby better understanding the equipment usage and discovering potential energy efficiency problems or optimization opportunities.

[0013] In one solution, the variational mode decomposition is used to decompose the original signal into several intrinsic mode functions and perform selective reconstruction to achieve the effect of data noise reduction. Specifically, it includes two parts: structural analysis and decomposition. By seeking the intrinsic mode function with the smallest bandwidth within a certain range for the original signal and making the sum of the intrinsic mode functions equal to the original signal, the purpose of signal noise reduction is achieved, thereby improving the data processing accuracy.

[0014] In one solution, the temperature factors in the load characteristic index library are refined into high temperature and low temperature states. By constructing an algorithm to associate load with temperature, the range of high and low temperatures is obtained, so as to more accurately evaluate the power consumption of the equipment under different temperature conditions, analyze the degree of load change at different temperatures, characterize the relationship between temperature and equipment load, and obtain the temperature range that has a greater impact on the power consumption characteristics of the equipment.

[0015] In one solution, the humidity factors in the load characteristic index library include high humidity and low humidity states. By constructing corresponding load characteristic indicators, the impact of humidity changes on equipment operating conditions and energy efficiency is analyzed. Considering that high humidity may cause the insulation performance of the equipment to deteriorate, thereby increasing energy consumption and failure risks, the impact of high humidity and low humidity states on equipment power consumption is analyzed separately to optimize the equipment's operating conditions.

[0016] In one solution, the electricity consumption characteristic analysis method systematically analyzes the historical electricity consumption data of high-energy-consuming equipment, extracts key characteristic indicators, including the month-on-month growth rate and year-on-year growth rate of load, and analyzes the electricity consumption characteristics of the equipment in different time periods in combination with external environmental factors. This identifies the electricity consumption patterns of the equipment, provides a scientific basis for energy management, and achieves energy conservation and emission reduction goals.

[0017] In one solution, the Bayesian optimization automatically adjusts the hyperparameters of the LSTM model by introducing a Bayesian optimization algorithm to improve the accuracy and robustness of the model in analyzing and modeling power consumption data features. The optimized model can more effectively predict the power consumption of the equipment, help identify energy efficiency issues, and provide data support for optimizing equipment operation strategies.

[0018] 1. By applying the denoising Bayesian optimization LSTM model, we can efficiently perform noise reduction and feature extraction on the electricity consumption data of high-energy-consuming equipment, improve the accuracy of analysis, better understand the usage of equipment, identify potential energy efficiency issues, and explore optimization opportunities.

[0019] 2. Introduce the variational mode decomposition method to decompose the original signal into multiple intrinsic mode functions, so that the signal can be selectively reconstructed, data noise reduction can be achieved, and thus the accuracy of data processing can be improved.

[0020] 3. By building a load characteristic index library, we can analyze in detail the impact of external factors such as temperature and humidity on equipment power consumption, provide a deeper understanding of equipment power consumption characteristics, and make energy management more scientific.

[0021] 4. Use Bayesian optimization to automatically adjust the hyperparameters of the LSTM model to improve the accuracy and robustness of the model, making the analysis and prediction of electricity consumption data more accurate, thereby effectively identifying energy efficiency issues and providing data support for optimizing equipment operation strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Flow chart of the method of the present invention;

[0023] Figure 2 Flowchart for the Bayesian optimization LSTM method;

[0024] Figure 3 Iteration graphs for four curves;

[0025] Figure 4 Optimize the curve for busbar noise reduction;

[0026] Figure 5 is the maximum voltage deviation rate under different humidity conditions. DETAILED DESCRIPTION

[0027] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate exemplary embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as those understood by those skilled in the art to which the present invention pertains. The terms used in the present specification are for the purpose of describing specific embodiments only and are not intended to limit the present invention. To facilitate understanding of the present invention, a more comprehensive description of the present invention will be provided below with reference to the accompanying drawings. Typical embodiments of the present invention are shown in the drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0029] like Figure 1 As shown, a method for analyzing the power consumption characteristics of high-energy-consuming equipment in a converter station is implemented in the following steps:

[0030] Step 1: Collect historical electricity consumption data of high-energy-consuming equipment, including air conditioners, valve coolers, and converter coolers.

[0031] Regarding the analysis of the power consumption characteristics of high-energy-consuming equipment, the present invention first collects historical power consumption data of high-energy-consuming electrical appliances, including air conditioners, valve cooling, and converter coolers within the station. Taking into account internal and external influencing factors and the load data flow of the equipment, the historical data of high-energy-consuming equipment is systematically analyzed and feature extracted. The influencing factors analyzed by the present invention specifically include:

[0032] Internal and external influencing factors

[0033]

[0034] Regarding the systematic analysis and feature extraction of historical data, due to the difficulties of the huge and redundant historical data as well as some missing values and outliers, the characteristics of the historical energy consumption data of high-energy-consuming equipment are shown in the following table:

[0035] Characteristics of energy consumption data of high-energy-consuming equipment

[0036]

[0037]

[0038] Step 2: Use the denoising Bayesian optimization LSTM model to perform denoising and feature extraction on the data, and adjust the hyperparameters of the LSTM model through the Bayesian optimization algorithm to improve the performance of the model.

[0039] Based on the characteristics and difficulties of historical data, this paper uses a noise reduction Bayesian optimization LSTM model to analyze and process historical data. The specific operations are as follows:

[0040] The LSTM model with Bayesian optimization for noise reduction has the following benefits for feature analysis and extraction of electricity usage data from high-energy-consuming devices: Noise reduction: Electricity usage data often contains noise, such as measurement errors or anomalies. The LSTM model can be used to filter out this noise, thereby more accurately extracting the true features of the data. Feature extraction: The LSTM model captures time series features in the data, modeling and extracting features of the temporal dependencies of electricity usage data from high-energy-consuming devices. Analyzing these features can better understand device usage and identify potential energy efficiency issues or optimization opportunities. Bayesian optimization: By introducing the Bayesian optimization algorithm, LSTM model hyperparameters can be more effectively adjusted to improve model performance. This helps improve the accuracy and robustness of electricity data feature analysis and modeling.

[0041] LSTM neural network is a special recurrent neural network (RNN) that has stronger learning ability for long time series data. It effectively solves the gradient vanishing and gradient exploding problems that occur when RNN models sequences with longer time spans, and is widely used in the field of time series prediction.

[0042] In LSTM neural networks, hyperparameters such as the number of hidden layers, the number of neurons in each hidden layer, and the initial learning rate must be manually configured. Appropriate parameter settings directly impact the neural network's topology and predictive performance. Bayesian optimization algorithms, however, can quickly and accurately find the optimal solution to complex objective functions with fewer objective function evaluations. Setting LSTM model parameters based on Bayesian optimization can effectively improve LSTM model performance.

[0043] The number of hidden layers in the LSTM neural network is set to 1. The number of hidden layer neurons, L2 regularization strength, and initial learning rate are used as hyperparameters to be optimized. The optimization goal is to minimize the prediction error of the validation set and continuously iterate the optimization. The Gaussian process is used as the proxy model for the actual problem, and the acquisition function is the PI function based on the lifting strategy, that is,

[0044]

[0045] Where: p(·) is the probability function; f(x) is the target value obtained by the Gaussian process; f(x + ) is the current optimal target value; is the cumulative density function of the Gaussian distribution; θ is the parameter that balances the relationship between global and local search; μ(x) and σ(x) are the mean and variance of the objective function, respectively.

[0046] The Bayesian optimization LSTM method process is as follows Figure 2 As shown in the figure. i 、y i are the historical output value and failure rate value of the high-energy-consuming equipment at the i-th moment; D is the data training set consisting of a total of n moments.

[0047] The noise reduction Bayesian optimization LSTM model has good accuracy, stability and applicability in the analysis of power consumption characteristics of high-energy-consuming equipment, and can better help analyze and predict the energy consumption of equipment. Figure 3 This is a comparison chart of the effects of several related models.

[0048] Step 3: Use variational mode decomposition to preprocess the time series data and decompose it into several intrinsic mode functions to achieve signal noise reduction.

[0049] VDM (Variational Mode Decomposition) is a signal processing method that effectively preprocesses time series data. It decomposes the original curve into several intrinsic mode functions and then selectively reconstructs them to achieve data noise reduction. The process of using VDM to reduce bus load noise consists of two parts: construction and decomposition.

[0050] S301, VDM structure analysis

[0051] In order to find the intrinsic mode function uk(t) with the minimum bandwidth for the original signal X(t) within a certain range and to make the sum of the intrinsic mode functions equal to the original function, the construction constraints of the intrinsic mode function are as follows. Let δ(t) be the Dirichlet function, * be the convolution symbol, and the established constraints are as follows:

[0052]

[0053] S302, ADM decomposition analysis

[0054] After assuming the constraints of the intrinsic mode function, the Lagrange multiplier method is used to unconstrain the constrained problem and the following function expression is constructed:

[0055]

[0056] Initialize the object to be solved iteratively On this basis, the iterative function is used to iteratively update ω k and

[0057]

[0058]

[0059] Until

[0060] S303, VMD noise reduction evaluation index

[0061] After the noisy signal X(t) is decomposed by variational mode, the reconstructed signal can be expressed as

[0062]

[0063] Where: r n (t) is the remainder; n is the number of IMFs. Suppose the noisy signal is represented by x=(x1,x2,…,x m ) defines the mean square error of the noise reduction deviation as

[0064]

[0065] Define the correlation as

[0066]

[0067] Define standard deviation δ A , δ B for

[0068]

[0069] Among the above indicators, MSE is a relative error indicator, and the correlation coefficient ρ is used to characterize the accuracy of the reconstructed intrinsic mode function curve after denoising and the bus load curve before denoising. The closer the correlation coefficient is to 1, the higher the degree of overlap with the bus load curve before denoising.

[0070] On this basis, the noise reduction deviation mean square error and the correlation coefficient are combined to form the judgment criterion:

[0071]

[0072] Where: α is the impact factor of the algorithm approximation, and its value range is [0, 1]. The judgment criteria can be adjusted by adjusting the impact factor: if the impact factor is increased, the proportion of the mean square error of the deviation in the judgment criteria can be adjusted, so that the judgment criteria pay more attention to the impact of the mean square error on the degree of noise reduction. Figure 4 This is a comparison chart before and after noise reduction.

[0073] Step 4: Build a load characteristic index library to analyze the impact of external factors such as temperature and humidity on equipment power consumption, providing an in-depth understanding of equipment power consumption characteristics.

[0074] Building a load characteristic index library is crucial when studying the power usage characteristics of high-energy-consuming equipment. Since the power usage characteristics of high-energy-consuming equipment are often influenced by multiple factors, including temperature, humidity, and month-on-month and year-on-year load growth rates, building a load characteristic index library can help systematically organize, record, and analyze these influencing factors, resulting in a comprehensive load characteristic index system. Secondly, building a load characteristic index library can help us gain a deeper understanding of the power usage patterns of high-energy-consuming equipment. By recording and analyzing load characteristic indicators, we can uncover the correlations and influences between different factors, revealing the complex patterns and characteristics of power usage in high-energy-consuming equipment. This facilitates in-depth research and understanding of the mechanisms and patterns of power usage in high-energy-consuming equipment. Finally, building a load characteristic index library is crucial for practical energy management and control. By analyzing and monitoring load characteristic indicators, we can better understand and predict the power usage of high-energy-consuming equipment, providing a scientific basis for energy management and control decisions. This helps enterprises optimize energy consumption, reduce costs, and improve energy efficiency. The following table lists the specific items in the load characteristic index library constructed by the present invention.

[0075] Load characteristic table

[0076]

[0077]

[0078] For the load characteristics in the table above, the present invention constructs specific characteristic indicators, which can provide a more comprehensive understanding of the power consumption characteristics of high-energy-consuming equipment. These characteristic indicators can be used for model training and prediction, helping us better manage and optimize the power consumption of equipment and achieve the goal of energy conservation and emission reduction. In actual application, further adjustment and verification are required based on specific data and business needs to ensure the accuracy and effectiveness of the model. The following is the specific construction content:

[0079] Temperature is a key factor influencing the power consumption of high-energy-consuming equipment. Different temperature conditions can have varying impacts on equipment operating efficiency and energy consumption. To further refine the impact of temperature, we can categorize it into different states, such as high and low temperatures. By developing an algorithm that correlates load and temperature, we can determine the ranges for high and low temperatures, enabling more accurate assessment of equipment power consumption in hot and cold conditions.

[0080]

[0081] Where: X and Y represent temperature and equipment load respectively, r is the correlation coefficient, and S is the covariance of X and Y variables; X is the standard deviation of the X variable; S Y is the standard deviation of the Y variable. This formula can be used to analyze the degree of load change at different temperatures, thereby describing the relationship between temperature and equipment load, and deriving the temperature range that has a greater impact on the equipment's power consumption characteristics, as follows Figure 5 shown.

[0082] Humidity also affects the electricity consumption of high-energy-consuming equipment. High humidity can degrade the insulation performance of equipment, increasing energy consumption and the risk of failure. Similarly, humidity can be categorized into different states, such as high and low, and corresponding load characteristic indicators can be constructed. Humidity fluctuations can affect equipment operating conditions and energy efficiency, so the impact of high and low humidity states on electricity consumption needs to be considered separately.

[0083] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0084] It should be understood that the detailed description of the technical solutions of the present invention using the preferred embodiments above is illustrative and not restrictive. A person skilled in the art, after reading the present specification, may modify the technical solutions described in the embodiments or replace some of the technical features therein with equivalents; such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing the power consumption characteristics of high-energy-consuming equipment in a converter station, characterized by: The method comprises: Collect historical electricity consumption data of high-energy-consuming equipment, including air conditioners, valve cooling, and converter coolers; A denoising Bayesian optimization LSTM model is used to perform denoising and feature extraction on the data, and the hyperparameters of the LSTM model are adjusted using a Bayesian optimization algorithm to improve the performance of the model; The time series data is preprocessed using variational mode decomposition and decomposed into several intrinsic mode functions to achieve signal noise reduction; Build a load characteristic index library, analyze the impact of temperature, humidity, and external factors on equipment power consumption, and provide an in-depth understanding of equipment power consumption characteristics.

2. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The denoising Bayesian optimization LSTM model is used to denoise the electricity consumption data of high-energy-consuming equipment. By filtering out measurement errors and abnormal data, the accuracy of data feature extraction is improved. The model can capture the time series characteristics in the data, model the time dependency of the electricity consumption data of high-energy-consuming equipment, and extract features, thereby better understanding the equipment usage and discovering potential energy efficiency issues or optimization opportunities.

3. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The variational mode decomposition is used to decompose the original signal into several intrinsic mode functions and perform selective reconstruction to achieve the effect of data noise reduction. It specifically includes two parts: construction analysis and decomposition. By seeking the intrinsic mode function with the smallest bandwidth within a certain range of the original signal and making the sum of the intrinsic mode functions equal to the original signal, the purpose of signal noise reduction is achieved, thereby improving data processing accuracy.

4. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The temperature factors in the load characteristic index library are refined into high temperature and low temperature states. By constructing an algorithm to associate load with temperature, the range of high and low temperatures is obtained, so as to more accurately evaluate the power consumption of the equipment under different temperature conditions, analyze the degree of load change at different temperatures, characterize the relationship between temperature and equipment load, and obtain the temperature range that has a greater impact on the power consumption characteristics of the equipment.

5. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The humidity factors in the load characteristic index library include high humidity and low humidity states. By constructing corresponding load characteristic indicators, the impact of humidity changes on equipment operating conditions and energy efficiency is analyzed. Considering that high humidity may cause the insulation performance of equipment to deteriorate, thereby increasing energy consumption and failure risks, the impact of high humidity and low humidity states on equipment power consumption is analyzed separately to optimize the equipment's operating conditions.

6. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The electricity consumption characteristic analysis method systematically analyzes the historical electricity consumption data of high-energy-consuming equipment, extracts key characteristic indicators, including the month-on-month load growth rate and the year-on-year load growth rate, and analyzes the electricity consumption characteristics of the equipment in different time periods in combination with external environmental factors. This method identifies the electricity consumption patterns of the equipment, provides a scientific basis for energy management, and achieves energy conservation and emission reduction goals.

7. The method for analyzing power consumption characteristics of high-energy-consuming equipment in a converter station according to claim 1, characterized in that: The Bayesian optimization method automatically adjusts the hyperparameters of the LSTM model by introducing a Bayesian optimization algorithm to improve the accuracy and robustness of the model in analyzing and modeling electricity consumption data features. The optimized model can more effectively predict the electricity consumption of the equipment, help identify energy efficiency issues, and provide data support for optimizing equipment operation strategies.

Images