A method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm

Abstract

The present invention provides a method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm, which relates to the technical field of reactor air gaps. The specific steps include: collecting operation data and vibration response data under different air gap parameters, constructing an air gap optimization model based on the TPE-XGBoost algorithm, training with the air gap parameters as input features and the operation and vibration data as output labels, establishing air gap parameter constraint conditions, randomly combining to construct initial population individuals, inputting into the model to obtain data and generating a comprehensive evaluation index, aiming at the optimization of the vibration state and guided by the maximization of the comprehensive evaluation index, using the genetic algorithm to iteratively optimize the initial population under the constraint conditions to obtain the optimal value of the air gap parameters. The present invention can efficiently explore the complex non-linear relationships of various factors, avoid the limitations of traditional algorithms, realize multi-objective collaborative optimization, significantly improve the operation performance and stability of the reactor, and effectively reduce vibration.

Description

Technical Field

[0001] The present invention relates to the technical field of reactor air gaps, and specifically to a method for optimizing reactor air gap parameters based on the TPE-XGBoost algorithm. Background Technique

[0002] As a basic electrical equipment in the power system, reactors are crucial in aspects such as limiting short-circuit current, suppressing harmonic waves of capacitor banks, and reactive power compensation, and can effectively ensure the safe and stable operation of the system. In terms of structure, due to the existence of air gaps between the reactor core laminations and the upper and lower yokes, it is more vulnerable to the action of magnetic forces, causing equipment vibration and then generating noise. Therefore, compared with power transformers, the vibration problem of reactors is more prominent. Equipment vibration will also loosen fasteners such as air gap pads and tie rods, bringing potential safety hazards to the operation of the reactor.

[0003] Traditional algorithms are difficult to capture the complex non-linear relationships among reactor air gap parameters, operating conditions, and vibration response data. Most are limited to simple linear modeling. Restricted by the model complexity, they cannot comprehensively consider the influence of the interaction of various factors on the vibration state. For example, early rule-based methods, due to fixed parameter adjustment rules, cannot adapt to changes in operating conditions and parameter coupling, resulting in poor vibration reduction effects; and most are local searches, prone to falling into local optima, and it is difficult to obtain the global optimal air gap parameter combination. Moreover, they often focus on a single target and ignore the influence of voltage, current, temperature, etc. on the overall operating state of the reactor, leading to problems such as uneven voltage distribution and increased current harmonics during the vibration optimization process of the reactor.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for optimizing reactor air gap vibration reduction based on the TPE-XGBoost algorithm to solve the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A method for optimizing reactor air gap vibration reduction based on the TPE-XGBoost algorithm, the specific steps include:

[0008] S1. Under different reactor air gap parameters, collect operation data and corresponding vibration response data;

[0009] S2. Based on the TPE-XGBoost algorithm, construct an air gap optimization model, use the reactor air gap parameters as input features, and the operation data and vibration response data as output labels to train the air gap optimization model;

[0010] S3. Establish the constraint conditions for the reactor air-gap parameters, randomly combine the reactor air-gap parameters to construct the individuals in the initial population, input the individuals of the initial population into the air-gap optimization model, and obtain the operating data and vibration response data;

[0011] S4. Process the operating data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index for comprehensively evaluating the vibration state of the reactor;

[0012] S5. With the optimization goal of achieving the optimal vibration state of the reactor and the maximization of the comprehensive evaluation index as the quantization guidance, under the constraint conditions of the reactor air-gap parameters, iteratively optimize the individuals of the initial population through the genetic algorithm to obtain the optimal individual, and based on the optimal individual, extract the optimal values of the reactor air-gap parameters.

[0013] Furthermore, the reactor air-gap parameters include the air-gap height, air-gap width, and air-gap magnetic density, the operating data includes voltage, current, and temperature, and the vibration response data includes vibration amplitude and vibration frequency.

[0014] Furthermore, randomly combine the reactor air-gap parameters to construct the individuals in the initial population. The specific process is as follows:

[0015] Calibrate the initial population as , and the initial population , is the th individual in the initial population, is the index of the individual in the initial population, and , is the number of individuals in the initial population, , where are the air-gap height, air-gap width, and air-gap magnetic density of the th individual respectively.

[0016] Furthermore, process the operating data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index. The formula is as follows:

[0017] ;

[0018] ;

[0019] ;

[0020] ;

[0021] ;

[0022] where is the comprehensive evaluation index of the th individual. The comprehensive evaluation index is used to evaluate the vibration state of the reactor from five aspects: voltage, current, temperature, vibration amplitude, and vibration frequency.

[0023] In the formula, is the voltage of the th individual, is the lower limit of the ideal voltage value of the th individual, is the upper limit of the ideal voltage value of the th individual, is the magnitude of the voltage distance from the ideal value of the th individual, is the current of the th individual, is the lower limit of the ideal current value of the th individual, is the upper limit of the ideal current value of the th individual, is the magnitude of the current distance from the ideal value of the th individual, is the temperature of the th individual, is the lower limit of the ideal temperature value of the th individual, is the upper limit of the ideal temperature value of the th individual, is the magnitude of the temperature distance from the ideal value of the th individual, is the vibration amplitude of the th individual, is the lower limit of the ideal vibration amplitude value of the th individual, is the upper limit of the ideal vibration amplitude value of the th individual, is the magnitude of the vibration amplitude distance from the ideal value of the th individual;

[0024] Among them, is the vibration frequency of the th individual in the normal operating frequency band, , is the power supply frequency, is a positive integer and , is the vibration frequency of the th individual in the abnormal operating frequency band;

[0025] In the formula, is the weight coefficient of the voltage, is the weight coefficient of the current, is the weight coefficient of the temperature, is the weight coefficient of the vibration amplitude, is the weight coefficient of the vibration frequency. Based on , let or .

[0026] Furthermore, the specific process of step S5 is as follows:

[0027] Taking the optimization of the vibration state of the reactor as the optimization goal and the maximization of the comprehensive evaluation index as the quantization guidance, iterative optimization is performed on the initial population , that is, selection, crossover, and mutation operations are performed on the individuals in the initial population . During the iterative optimization process, constraint conditions need to be set, that is, the maximum and minimum values of the air gap height, air gap width, and air gap magnetic density are set respectively. Within the constraint range of the air gap height, air gap width, and air gap magnetic density, iterative optimization is performed on the initial population . Specifically, individuals with the top comprehensive evaluation index are selected as parents. Through the crossover operation, the genes of the parent individuals are exchanged and combined to generate new individuals. Then, after performing mutation operations on the genes of the air gap height, air gap width, and air gap magnetic density in the newly generated individuals, the selection, crossover, and mutation operations are repeated until the predetermined number of iterations is reached;

[0028] After performing iterative optimization on the initial population , the optimal individual is labeled as . The optimal values of the reactor air gap parameters are the air gap height , the air gap width , and the air gap magnetic density .

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] By constructing an air gap optimization model through the TPE-XGBoost algorithm, the present invention can efficiently search the hyperparameter space. By using the reactor air gap parameters as input features and the operation data and vibration response data as output labels for training, the model can deeply explore the complex nonlinear relationships between various factors, accurately identify how small changes in the air gap parameters affect the vibration response, provide strong support for subsequent precise optimization, and effectively make up for the deficiencies of traditional algorithms in complex relationship modeling;

[0031] By aiming at optimizing the vibration state of the reactor and taking the maximization of the comprehensive evaluation index as the quantization guidance, the influence of multiple factors such as voltage, current, temperature, vibration amplitude and vibration frequency on the vibration state of the reactor is comprehensively considered. At the same time, the genetic algorithm is used to iteratively optimize the individuals of the initial population. Through selection, crossover and mutation operations, it can perform global search in the entire solution space, effectively avoid falling into local optimal solutions, and comprehensively consider the influence of multiple factors such as voltage, current, temperature, vibration amplitude and frequency on the operation state of the reactor, realizing multi-objective collaborative optimization, ensuring that while reducing vibration, the voltage distribution is maintained uniform, the current harmonics are stable, and the overall operation performance and stability of the reactor are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0034] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0035] Embodiment 1:

[0036] Please refer to Figure 1 , the present invention provides a technical solution:

[0037] A method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm, the specific steps include:

[0038] S1. Under different air gap parameters of the reactor, collect operation data and corresponding vibration response data. The air gap parameters of the reactor include air gap height, air gap width and air gap magnetic density. The operation data includes voltage, current and temperature. The vibration response data includes vibration amplitude and vibration frequency;

[0039] On the basis of the above embodiments, data acquisition should be carried out for a sufficient length of time under standard operating conditions, and operating data and vibration response data should be collected multiple times at different time intervals, so as to calculate the average values of the operating data and vibration response data.

[0040] Among them, the standard operating conditions refer to the rated speed and rated power.

[0041] On the basis of the above embodiments, the acquisition methods of voltage, current, and temperature are as follows:

[0042] A voltage transformer is used for voltage measurement. The primary side of the voltage transformer is connected in parallel with the power supply side of the reactor, and the secondary side is connected to a high-precision digital voltmeter to measure and record the voltage value in real time.

[0043] A current transformer is used to measure the current. The primary side of the current transformer is connected in series to the current loop of the reactor, and the secondary side is connected to an ammeter to monitor and record the current value in real time.

[0044] The measuring end of the thermocouple is tightly installed on the iron core of the reactor, and the reference end is kept in a constant temperature environment. The thermoelectric potential signal output by the thermocouple is converted into a standard voltage or current signal by a temperature transmitter and then connected to the temperature acquisition system. The temperature acquisition system records the temperature data every 10 seconds.

[0045] On the basis of the above embodiments, the acquisition methods of vibration amplitude and vibration frequency are as follows:

[0046] An acceleration sensor is a commonly used device for measuring vibration. It can convert vibration acceleration into an electrical signal. The acceleration sensor is installed on the iron core of the reactor. After the electrical signal output by the acceleration sensor is processed by an amplifier and a filter, it is input into the data acquisition system for recording. By integrating the acquired acceleration signal, vibration velocity and displacement can be obtained, thereby calculating the vibration amplitude; at the same time, by performing spectral analysis on the signal, the vibration frequency can be obtained.

[0047] On the basis of the above embodiments, the air gap parameters, operating parameters, and vibration response data of the reactor are respectively subjected to maximum-minimum normalization processing, and then the normalized data is used for subsequent analysis processing, so that in the subsequent analysis processing process, various data can be analyzed and processed under the same dimension, avoiding the problem that some data is ignored due to different dimensions.

[0048] Among them, the acceleration sensor can adopt the models in existing equipment, and there is no limitation here.

[0049] S2. Construct an air gap optimization model based on the TPE-XGBoost algorithm, use the air gap parameters of the reactor as input features, and the operating data and vibration response data as output labels to train the air gap optimization model;

[0050] Based on the above embodiments, the process of training the air gap optimization model is as follows:

[0051] First step, optimize the XGBoost hyperparameters using the TPE algorithm

[0052] Define the hyperparameter search range: The XGBoost algorithm has multiple important hyperparameters, including the learning rate, which determines the step size of model updates in each iteration and usually takes values in the range of 0.01 - 0.3; the maximum depth of the tree, which limits the complexity of the decision tree and the value range can be set between 3 - 10; the subsample ratio, that is, the ratio of randomly sampling from the original samples during each training, with a range of 0.5 - 1; the column sampling ratio, which refers to the ratio of randomly sampling from the features when constructing each tree and is also between 0.5 - 1; the number of trees, generally between 50 - 200;

[0053] Determine the objective function: The objective function is used to evaluate the performance of the model under different hyperparameter combinations. Usually, the mean squared error on the validation set is selected as the evaluation metric because it can intuitively reflect the deviation degree between the model's predicted values and the true values. For a given set of hyperparameters, use the training set data to train the XGBoost model, then make predictions on the validation set, and calculate the mean squared error between the predicted values and the true values. This error value is the output result of the objective function under this set of hyperparameters;

[0054] Execute the TPE algorithm search: The TPE algorithm in the hyperparameter search space intelligently selects the next hyperparameter combination to be evaluated by constructing a probability model based on historical evaluation results. It continuously tries different hyperparameter combinations, and gradually approaches the optimal hyperparameter settings according to the evaluation results of the objective function. During the search process, record each evaluated hyperparameter combination and its corresponding objective function value. After a certain number of iterations (such as 100 times), determine the optimal hyperparameter combination;

[0055] Second step, construct and train the XGBoost model

[0056] Initialize the model: Based on the optimal hyperparameter combination obtained by the TPE algorithm search, construct the XGBoost model, set the learning rate, the maximum depth of the tree, the subsample ratio, the column sampling ratio, and the number of trees to the optimal values, and determine the initial structure and parameters of the model;

[0057] Model training: Use the training set data to train the constructed XGBoost model. During the training process, the model continuously adjusts its internal parameters according to the input air gap parameters to minimize the error between the predicted operating data and vibration response data and the actual operating data and vibration response data. The model learns the patterns and rules in the data through an iterative manner, making the prediction results closer and closer to the true values;

[0058] Step 3. Evaluate the model performance

[0059] Use the trained XGBoost model to predict the test set data. Input the air gap parameters of the test set, and the model outputs the predicted values of the corresponding operating data and vibration response data;

[0060] Compare the predicted operating data and vibration response data with the actual operating data and vibration response data of the test set, calculate the mean square error. When the mean square error is within the range, complete the training of the air gap optimization model.

[0061] S3. Establish the constraint conditions for the reactor air gap parameters, randomly combine the reactor air gap parameters to construct the individuals in the initial population, and input the individuals in the initial population into the air gap optimization model to obtain the operating data and vibration response data;

[0062] On the basis of the above embodiments, randomly combine the reactor air gap parameters to construct the individuals in the initial population. The specific process is as follows:

[0063] Label the initial population as , and the initial population , is the th individual in the initial population, is the index of the individual in the initial population, and , is the number of individuals in the initial population, , where are the air gap height, air gap width, and air gap magnetic density of the th individual respectively.

[0064] S4. Process the operating data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index for comprehensively evaluating the vibration state of the reactor;

[0065] On the basis of the above embodiments, process the operating data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index. The formula is as follows:

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] Among them, is the comprehensive evaluation index of the th individual. The comprehensive evaluation index is used to evaluate the vibration state of the reactor from five aspects: voltage, current, temperature, vibration amplitude, and vibration frequency. And the larger the comprehensive evaluation index, the closer the vibration of the reactor is to the ideal state. In the ideal state, the vibration amplitude of the reactor will be reduced to a relatively stable and small level;

[0072] In the formula, is the voltage of the th individual, is the lower limit of the ideal voltage value of the th individual, is the upper limit of the ideal voltage value of the th individual, is the magnitude of the voltage of the th individual from the ideal value, is the current of the th individual, is the lower limit of the ideal current value of the th individual, is the upper limit of the ideal current value of the th individual, is the magnitude of the current of the th individual from the ideal value, is the temperature of the th individual, is the lower limit of the ideal temperature value of the th individual, is the upper limit of the ideal temperature value of the th individual, is the magnitude of the temperature of the th individual from the ideal value, is the vibration amplitude of the th individual, is the lower limit of the ideal vibration amplitude value of the th individual, is the upper limit of the ideal vibration amplitude value of the th individual, is the magnitude of the vibration amplitude of the th individual from the ideal value;

[0073] Among them, is the vibration frequency of the th individual in the normal operation frequency band. The vibration frequency in the normal operation frequency band is composed of the power supply frequency and its harmonic frequencies, , is the power supply frequency, is a positive integer and ;

[0074] is the vibration frequency of the nth individual in the abnormal operating frequency band. The vibration frequencies in the abnormal operating frequency band cover three cases: one is the new frequency components that have no integer multiple relationship with the power frequency and its harmonic frequencies generated by abnormal conditions inside the reactor (loose mechanical components, partial discharge); the second is the frequency after the normal operating frequency shifts due to changes in the internal structure of the reactor (local winding deformation); the third is the frequency range when the vibration frequency of the reactor approaches its natural frequency;

[0075] On this basis, it should be noted that:

[0076] Within the range, as the voltage increases, the internal magnetic field strength of the reactor increases, the magnetic flux distribution becomes more uniform, making the electromagnetic force acting on each component of the reactor more stable, reducing the vibration caused by electromagnetic force fluctuations, making the vibration of the reactor approach the ideal state, and thus the comprehensive evaluation index increases; in other ranges, when increases, if , too low a voltage will cause the reactor to be unable to establish a magnetic field of sufficient strength, making the electromagnetic force insufficient to maintain the stable operation of the components, triggering additional vibrations and noises, making the vibration of the reactor deviate from the ideal state, and thus the comprehensive evaluation index decreases; if , too high a voltage will cause the iron core to saturate, resulting in a sharp change in the magnetic flux, generating a strong electromagnetic force impact, causing the internal structure of the reactor to bear excessive stress, resulting in winding deformation and iron core loosening, making the vibration of the reactor deviate from the ideal state, and thus the comprehensive evaluation index decreases;

[0077] Within the range, as the current increases, the increase in current promotes the enhancement of the self-inductance effect of the reactor, which can better suppress the sudden change of current, stabilize the transmission of current, reduce the impact on the reactor caused by current fluctuations, making the operation of the reactor more stable, the stress fluctuation on its internal mechanical structure reduced, making the vibration of the reactor approach the ideal state, and thus the comprehensive evaluation index increases; in other ranges, when increases, if , too small a current may cause the reactor to malfunction, with insufficient internal magnetic field strength, making it difficult to maintain a stable electromagnetic environment, triggering relative displacement and vibration between components, making the vibration of the reactor deviate from the ideal state, and thus the comprehensive evaluation index decreases; if , an overload current will cause the reactor winding to heat up severely, resulting in a decline in insulation performance and even potentially leading to the burning out of the winding. At the same time, the strong magnetic field generated by the excessive current will subject the components to a huge electromagnetic force, causing damage to the mechanical structure, making the vibration of the reactor deviate from the ideal state, and thus reducing the comprehensive evaluation index;

[0078] Within the range, as the temperature increases, the resistance of components such as the reactor winding will increase appropriately with the rise in temperature. According to Joule's law, its heating power will increase somewhat. However, since within the reasonable temperature range, this increase in heating is within a controllable range and helps to make the thermal expansion of various components inside the reactor tend to be consistent, reducing the internal stress generated due to the difference in thermal expansion and contraction, and further reducing the additional vibration, making the vibration of the reactor approach the ideal state, and thus increasing the comprehensive evaluation index; In other ranges, when increases, the temperature is farther away from the ideal temperature value. If , too low a temperature may cause the material to become brittle, reducing the mechanical strength of the components, and being prone to cracks and damage under the action of electromagnetic force. The mechanical strength of the winding decreases, and it is more likely to undergo displacement and deformation under the action of electromagnetic force, thus triggering abnormal vibration. This abnormal vibration will interfere with the vibration during normal operation, resulting in an increase in the vibration amplitude and a more complex frequency component, making the vibration of the reactor deviate from the ideal state and reducing its comprehensive evaluation index; If , too high a temperature will accelerate the aging of the insulating material, reducing the insulation performance of the winding. At the same time, excessive thermal expansion causes extrusion and deformation between components, triggering serious vibration problems, making the vibration of the reactor deviate from the ideal state, and thus reducing the comprehensive evaluation index;

[0079] Within the range, as the vibration amplitude increases, if this increase is caused by reasonable factors such as load changes under normal operating conditions, it means that the reactor has a certain elastic deformation to adapt to the adjustment of the operating state. During this process, the damping effect of the internal structure of the reactor will consume part of the vibration energy, making the vibration tend to be stable. At the same time, it indicates that the structure of the reactor has a certain toughness and stability, making the vibration of the reactor approach the ideal state and increasing the comprehensive evaluation index; In other ranges, when increases, the vibration amplitude is farther away from the ideal vibration amplitude value. If , too small an amplitude will break the original balance state. For example, if the vibration of the iron core weakens due to component loosening, it will cause uneven distribution of the surrounding electromagnetic force, triggering additional vibration of other components, and even causing a change in the vibration frequency of the entire reactor, making the vibration of the reactor deviate from the ideal state, and thus reducing the comprehensive evaluation index; If , an excessive vibration amplitude indicates that the internal structure of the reactor has been severely unbalanced, with serious problems such as component detachment and loose connections, resulting in a chaotic internal stress distribution, increased vibration, causing the reactor vibration to deviate from the ideal state, and thus the comprehensive evaluation index decreases;

[0080] In the normal operating frequency range, when the vibration frequency increases, as long as the frequency change is caused by an increase in the stable harmonic components on the power supply side and within the tolerable frequency range designed for the reactor, as the frequency increases, the acting frequency of the electromagnetic force inside the reactor speeds up, prompting its structure to reach dynamic balance faster and reducing the unstable factors brought about by low-frequency vibration. For example, it reduces the risk of vibration amplitude amplification caused by low-frequency resonance, making the reactor vibration approach the ideal state, and thus the comprehensive evaluation index increases;

[0081] In the abnormal operating frequency range, when the new frequency or offset frequency or natural frequency increases, the increase in the new frequency component often means the intervention of more abnormal excitation sources. These abnormal excitations will break the original electromagnetic and mechanical balance inside the reactor, triggering irregular vibrations; an increase in the offset frequency indicates a deeper degree of abnormality caused by internal structural changes, resulting in uneven electromagnetic force distribution and further exacerbating the vibration; an increase in the proximity to the natural frequency range significantly increases the resonance risk. Once resonance occurs, the vibration amplitude of the reactor will increase sharply, and the internal components will bear huge stresses, all of which will lead to a chaotic internal stress distribution in the reactor, intensified structural vibration, causing the reactor vibration to deviate from the ideal state, and thus the comprehensive evaluation index decreases.

[0082] Therefore, when , the comprehensive evaluation index and voltage , current , temperature , vibration amplitude , the vibration frequency during normal operation are all positively correlated. In addition, the comprehensive evaluation index and voltage , current , temperature , vibration amplitude , the vibration frequency during abnormal operation are all negatively correlated. Therefore, the above weighted summation formula is used to characterize the functional relationship between the comprehensive evaluation index and voltage , current , temperature , vibration amplitude , the vibration frequency during normal operation and the vibration frequency during abnormal operation .

[0083] In the formula, is the weighting coefficient of voltage, is the weighting coefficient of current, is the weighting coefficient of temperature, is the weighting coefficient of vibration amplitude, is the weighting coefficient of vibration frequency;

[0084] The magnitudes of the weighting coefficients are set as follows:

[0085] Both voltage and current are key factors affecting the electromagnetic characteristics of the reactor. Voltage determines the intensity and distribution of the magnetic field inside the reactor. Appropriate voltage can ensure the stability of the magnetic field, thereby reducing vibrations caused by electromagnetic force fluctuations. Changes in current will change the self-inductance effect of the reactor, affect the magnitude of the electromagnetic force, and thus have an impact on the running stability. In the power system, the two are interrelated and have a relatively close degree of influence on the operating state of the reactor. For example, during the normal operation of the power grid, fluctuations in voltage and current will directly affect the performance of the reactor. Therefore, their weighting coefficients are set equal, ;

[0086] Within the normal operating temperature range, the influence of temperature changes on the vibration stability of the reactor is relatively mild. Although an increase in temperature will increase the winding resistance and cause changes in thermal expansion, as long as it is within the normal range, these changes have a gradual and relatively small impact on vibration stability. Only when the temperature exceeds the normal range will it have a serious impact on insulation performance, mechanical structure, etc., and thus affect vibration stability. In contrast, changes in voltage and current have a more direct and immediate impact on the operating state of the reactor. Therefore, ;

[0087] Vibration amplitude is a key indicator directly reflecting the vibration stability of the reactor. It directly reflects the mechanical vibration state of the reactor during operation. Whether it is caused by electromagnetic force changes or other factors, the vibration will ultimately be manifested through the vibration amplitude. For example, when a fault occurs inside the reactor, such as winding looseness or core displacement, it will first be reflected in an increase in the vibration amplitude. Moreover, excessive vibration amplitude will directly threaten the structural integrity and operating reliability of the reactor, and its impact on stability is comparable to that of voltage and current. Therefore, in terms of the setting of the weighting coefficient, it is equivalent to that of voltage and current. Thus ;

[0088] In the normal operating frequency range, frequency changes (caused by an increase in the stable harmonic components on the power supply side and within the designed tolerance frequency range) have a positive impact on vibration stability. However, this impact is relatively indirect. It mainly changes the electromagnetic force acting frequency, enabling the reactor structure to reach dynamic equilibrium faster and reducing the unstable factors of low-frequency vibration. Nevertheless, compared with direct influencing factors such as voltage, current, and vibration amplitude, the impact of normal operating frequency changes on vibration stability is weaker. For example, within a certain range of normal operating frequencies, changes do not immediately cause obvious changes in the vibration state of the reactor like fluctuations in voltage, current, or changes in vibration amplitude. Therefore, the weight coefficient is relatively small. Thus, ;

[0089] In the abnormal operating frequency range, the appearance of abnormal frequencies indicates serious problems inside the reactor, such as loose mechanical components, partial discharge, winding deformation, etc. These abnormal conditions introduce new excitation sources, disrupt the original electromagnetic and mechanical balance, trigger irregular vibrations, and may even lead to resonance, greatly reducing the vibration stability of the reactor. Compared with other factors, the destructive effect of abnormal frequencies on the stability of the reactor is the most serious. For example, once the vibration frequency approaches the natural frequency and resonance occurs, the vibration amplitude of the reactor will increase sharply, and the internal components will bear huge stresses, which may cause equipment damage in a short time. Therefore, in the abnormal operating frequency range, it is necessary to significantly increase value to highlight the serious negative impact of abnormal frequencies on the comprehensive evaluation index. Thus, .

[0090] To sum up, on the basis of , let or .

[0091] As an implementation method, when , the value range of is 0.2 - 0.33, the value range of is 0.1 - 0.2, the value range of is 0.2 - 0.33,

[0092] When , the value range of is 0.15 - 0.25, the value range of is 0 - 0.15, the value range of The value range of is 0.25 - 1, and the specific value is set by technicians according to the actual situation, which is not limited here.

[0093] S5. Taking the optimization of the vibration state of the reactor as the optimization goal and the maximization of the comprehensive evaluation index as the quantization orientation, under the constraint conditions of the reactor air-gap parameters, the individuals of the initial population are iteratively optimized through the genetic algorithm to obtain the optimal individual, and based on the optimal individual, the optimal values of the reactor air-gap parameters are extracted.

[0094] Based on the above embodiments, the specific process of step S5 is as follows:

[0095] Taking the optimization of the vibration state of the reactor as the optimization goal and the maximization of the comprehensive evaluation index as the quantization orientation, the initial population is iteratively optimized, that is, the individuals in the initial population are selected, crossed, and mutated. During the iterative optimization process, constraint conditions need to be set, that is, the maximum and minimum values of the air-gap height, air-gap width, and air-gap magnetic density are set respectively. Within the constraint range of the air-gap height, air-gap width, and air-gap magnetic density, the initial population is iteratively optimized. Specifically, the individuals with the comprehensive evaluation index in the forefront are selected as the parents. The forefront refers to the individuals in the top 50% of the comprehensive evaluation index. Through the crossing operation, the genes of the parent individuals are exchanged and combined to generate new individuals. Then, after mutating the genes of the air-gap height, air-gap width, and air-gap magnetic density in the newly generated individuals, the selection, crossing, and mutation operations are repeated until the predetermined number of iterations is reached;

[0096] After the initial population is iteratively optimized, the optimal individual is labeled as , and the optimal values of the reactor air-gap parameters are the air-gap height , air-gap width , and air-gap magnetic density .

[0097] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by technicians in this field according to the actual situation.

[0098] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0099] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0100] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application.

Claims

1. A method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm, characterized in that The specific steps are as follows: S1. Collect operation data and corresponding vibration response data under different reactor air-gap parameters; S2. Build an air-gap optimization model based on the TPE-XGBoost algorithm. Use the reactor air-gap parameters as input features, and the operation data and vibration response data as output labels to train the air-gap optimization model; S3. Establish the constraint conditions for the reactor air-gap parameters, randomly combine the reactor air-gap parameters to construct individuals in the initial population, input the individuals of the initial population into the air-gap optimization model, and obtain the operation data and vibration response data; S4. Process the operation data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index for comprehensively evaluating the vibration state of the reactor; S5. With the optimization goal of achieving the optimal vibration state of the reactor and the maximization of the comprehensive evaluation index as the quantization guide, under the constraint conditions of the reactor air-gap parameters, iteratively optimize the individuals of the initial population through the genetic algorithm to obtain the optimal individual, and based on the optimal individual, extract the optimal value of the reactor air-gap parameters; Process the operation data and vibration response data and perform correlation analysis to generate a comprehensive evaluation index. The basis formula is as follows: Among them, is the comprehensive evaluation index of the th individual. The comprehensive evaluation index is used to evaluate the vibration state of the reactor from five aspects: voltage, current, temperature, vibration amplitude, and vibration frequency. Wherein, is the voltage of the th individual, is the lower limit of the ideal voltage value of the th individual, is the upper limit of the ideal voltage value of the th individual, is the magnitude of the voltage distance from the ideal value of the th individual, is the current of the th individual, is the lower limit of the ideal current value of the th individual, is the upper limit of the ideal current value of the th individual, is the magnitude of the current distance from the ideal value of the th individual, is the temperature of the th individual, is the lower limit of the ideal temperature value of the th individual, is the upper limit of the ideal temperature value of the th individual, is the magnitude of the temperature distance from the ideal value of the th individual, is the vibration amplitude of the th individual, is the lower limit of the ideal vibration amplitude value of the th individual, is the upper limit of the ideal vibration amplitude value of the th individual, is the magnitude of the vibration amplitude distance from the ideal value of the th individual; Among them, is the vibration frequency of the th individual in the normal operating frequency band, , is the power supply frequency, is a positive integer and , is the vibration frequency of the th individual in the abnormal operating frequency band; In the formula, is the weight coefficient of voltage, is the weight coefficient of current, is the weight coefficient of temperature, is the weight coefficient of vibration amplitude, is the weight coefficient of vibration frequency. On the basis of , let or .

2. The reactor air gap parameter optimization method based on the TPE-XGBoost algorithm according to claim 1, characterized in that: The reactor air-gap parameters include air-gap height, air-gap width, and air-gap magnetic density. The operation data includes voltage, current, and temperature. The vibration response data includes vibration amplitude and vibration frequency.

3. The method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm according to claim 2, wherein: Randomly combine the reactor air-gap parameters to construct individuals in the initial population. The specific process is as follows: Calibrate the initial population as , and the initial population , is the th individual in the initial population, is the index of the individual in the initial population, and , is the number of individuals in the initial population, , where are respectively the air-gap height, air-gap width and air-gap magnetic flux density of the th individual.

4. The method for optimizing the air gap parameters of a reactor based on the TPE-XGBoost algorithm according to claim 3, wherein: The specific process of step S5 is as follows: Taking the optimization of the vibration state of the reactor as the optimization goal and the maximization of the comprehensive evaluation index as the quantization guidance, iterative optimization is carried out on the initial population , that is, selection, crossover, and mutation operations are performed on the individuals in the initial population . During the iterative optimization process, constraint conditions need to be set, that is, the maximum and minimum values of the air-gap height, air-gap width, and air-gap magnetic density are set respectively. Within the constraint ranges of the air-gap height, air-gap width, and air-gap magnetic density, iterative optimization is carried out on the initial population . Specifically, individuals with comprehensive evaluation indexes in the forefront are selected as parents. Through the crossover operation, the genes of the parent individuals are exchanged and combined to generate new individuals. Then, after mutation operations are performed on the genes of the air-gap height, air-gap width, and air-gap magnetic density in the newly generated individuals, the selection, crossover, and mutation operations are repeated until the predetermined number of iterations is reached; After iteratively optimizing the initial population the optimal individual is labeled as , and the optimal values of the reactor air-gap parameters are the air-gap height , the air-gap width and the air-gap magnetic flux density .

Images