Remote control method and system for distributed transformer oil treatment equipment

Abstract

The application discloses a remote control method and system of distributed transformer oil treatment equipment, and relates to the technical field of waste transformer oil treatment, which comprises the following steps: establishing a droplet dynamics model, obtaining the initial particle size distribution of oil mist under different ultrasonic frequencies and powers through the model; based on the multiple influencing factors of the oil mist particle size distribution, correcting the initial particle size distribution, generating high-precision real-time particle size prediction values, comparing the prediction values with preset target particle size intervals, and generating a smooth deviation signal; based on the smooth deviation signal, dynamically adjusting the working frequency and power of the ultrasonic generator, and atomizing the waste transformer oil to obtain oil mist particles meeting the target particle size interval; and introducing the atomized oil mist into a corona discharge area, performing a molecular conversion reaction, and feeding back and optimizing; the application combines particle size prediction correction, ultrasonic parameter dynamic adjustment and corona discharge feedback optimization, and realizes efficient, low-energy-consumption and safe treatment of waste transformer oil and distributed remote intelligent management and control.

Description

Technical Field

[0001] This invention relates to the field of waste transformer oil treatment technology, and more specifically, to a remote control method and system for distributed transformer oil treatment equipment. Background Technology

[0002] As a core piece of equipment in the power system, transformers rely on transformer oil for insulation and cooling during operation. Over time, acidic substances, impurities, and aging byproducts gradually accumulate in the transformer oil, leading to a decline in its dielectric and heat dissipation properties, and even affecting the safe operation of the transformer. Therefore, the regeneration and treatment of waste transformer oil has become an urgent problem to be solved in the power industry.

[0003] In existing technologies, the main methods for treating waste transformer oil fall into the following categories: Firstly, refining processes, which regenerate waste oil through physical filtration, chemical adsorption, or hydrorefining. However, these processes are complex, costly, and require strict quality control of the oil. Secondly, high-temperature pyrolysis or incineration converts waste oil into fuel or heat energy. However, high-temperature conditions not only consume enormous amounts of energy but also easily generate secondary pollution, posing environmental and safety hazards. To address these issues, this invention proposes a solution. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a remote control method and system for distributed transformer oil treatment equipment. This method utilizes an ultrasonic generator to uniformly atomize waste transformer oil and achieves stable corona discharge between high-voltage corona plates, converting the atomized oil mist into combustible gases such as acetylene and methane. Simultaneously, by combining particle size prediction correction, dynamic adjustment of ultrasonic parameters, and corona discharge feedback optimization, the method achieves efficient, low-energy, and safe treatment of waste transformer oil, as well as distributed remote intelligent management and control.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a first aspect, this application provides a remote control method for a distributed transformer oil treatment device. The method includes: acquiring the physical property parameters of waste transformer oil and establishing a droplet dynamics model; obtaining the initial particle size distribution of oil mist under different ultrasonic frequencies and powers through the model; correcting the initial particle size distribution based on multiple influencing factors of the oil mist particle size distribution to generate a high-precision real-time particle size prediction value; comparing the high-precision real-time particle size prediction value with a preset target particle size range to generate a smoothing deviation signal; dynamically adjusting the operating frequency and power of the ultrasonic generator based on the smoothing deviation signal, and atomizing the waste transformer oil to obtain an oil mist particle size that conforms to the target particle size range; and introducing the atomized oil mist into a corona discharge zone to perform a molecular conversion reaction and perform feedback optimization.

[0007] In one embodiment, a droplet dynamics model is established, and the initial particle size distribution of oil mist under different ultrasonic frequencies and powers is obtained through the model. Specifically, the preprocessed physical property parameters are used as input variables, and a droplet dynamics model is constructed based on the Rayleigh–Plesset equation. A finite time step is established through the model, and the droplet dynamics model is solved to obtain the oil mist radius data corresponding to each time step. An oscillation curve is plotted based on the oil mist radius data, and a steady-state determination is made based on the difference in oil mist radius data between two adjacent periods. Based on the determination result, statistical analysis is performed on the oil mist radius data within the steady-state time period to obtain the initial particle size distribution.

[0008] In one embodiment, based on the multiple influencing factors of oil mist particle size distribution, the initial particle size distribution is corrected to generate a high-precision real-time particle size prediction value. Specifically, training samples are generated by the deviation between the actual oil mist particle size distribution and the initial particle size distribution; a feedforward neural network model is constructed based on a neural network, and the training samples are used as training input; bubble feature parameters are imported into the trained neural network model, and a first correction factor is output; the first correction factor is applied to the initial particle size distribution to obtain a first corrected particle size distribution; impurity feature parameters are combined with the first corrected particle size distribution and input into the neural network model, and a second correction factor is obtained by learning the role of impurities in nucleation and oil mist growth; the second correction factor is applied to the first corrected particle size distribution to obtain a second corrected particle size distribution; flow field feature parameters and the second corrected particle size distribution are used as inputs to the neural network model to output a third correction factor; the third correction factor is applied to the second corrected particle size distribution to obtain a third corrected particle size distribution; the third corrected particle size distribution is incrementally updated based on real-time operating conditions to obtain the final high-precision real-time particle size prediction value.

[0009] In one embodiment, the third corrected particle size distribution is incrementally updated based on real-time operating conditions to obtain the final high-precision real-time particle size prediction value. Specifically, the third corrected particle size distribution is used as the prior state for filtering; real-time operating parameters are collected, and the real-time operating parameters and the actual oil mist particle size distribution are input to the filter as observation inputs for the prior state; based on the prior state and the observation inputs, the extended Kalman filter method is used to recursively estimate the oil mist particle size distribution to obtain the dynamically corrected particle size distribution state; the observation residual is obtained based on the particle size distribution state, and abrupt change detection is performed; if the observation residual exceeds a preset abrupt change threshold, abrupt change occurs, triggering the incremental update mechanism of the filter, dynamically adjusting the filter gain for compensation and optimization, and outputting a high-precision real-time particle size prediction value.

[0010] In one embodiment, a smoothed deviation signal is generated by comparing a high-precision real-time particle size prediction value with a preset target particle size range. Specifically, the deviation signal is obtained by calculating the deviation between the high-precision real-time particle size prediction value and the preset target particle size range; the deviation signal is divided into several continuous subsequences according to a preset time window; a local weighted average value is calculated for each subsequence of the deviation signal within each time window, wherein the weights are adjusted according to the signal change rate of the current sampling point; the local weighted average values ​​are concatenated in chronological order to obtain a preliminary smoothed deviation signal; the preliminary smoothed deviation signal is subjected to adaptive low-pass filtering to generate a smoothed deviation signal, wherein the low-pass filtering includes a filtering coefficient dynamically adjusted according to the change rate of the real-time deviation signal.

[0011] In one embodiment, the operating frequency and power of the ultrasonic generator are dynamically adjusted based on the smoothing deviation signal, and the waste transformer oil is atomized to obtain an oil mist particle size that conforms to the target particle size range. Specifically, the smoothing deviation signal is input into the ultrasonic generator control model to calculate the initial frequency adjustment amount and the initial power adjustment amount; based on the initial frequency adjustment amount and power adjustment amount, the adjustment direction and preliminary adjustment amount are extracted, a limit judgment is performed, and the final adjustment amount is obtained; the final adjustment amount is sent to the ultrasonic generator to modify the operating frequency and power in real time, and the waste transformer oil is atomized to obtain a uniform oil mist, so that the oil mist particle size is continuously maintained within the target particle size range.

[0012] In one embodiment, the atomized oil mist is introduced into a corona discharge region to perform a molecular conversion reaction and undergo feedback optimization. Specifically, the atomized oil mist is introduced into the corona discharge region, allowing it to pass between high-voltage electrode plates. The corona discharge region includes positive and negative electrode plates with an adjustable spacing between them. A conversion efficiency prediction model is established based on the interaction between the oil mist particle size distribution and the corona discharge intensity. During the corona discharge process, first data is collected at different positions between the electrodes, and a weighted electric field uniformity index is calculated. Based on the conversion efficiency prediction model and the electric field uniformity index, a multi-parameter collaborative optimization objective function is established, and a multivariate control method is used to collaboratively adjust the corona voltage, current density, and electrode spacing according to the optimization objective function results. The real-time conversion rate measurement value of the corona discharge region after collaborative adjustment is obtained, and dynamic optimization is performed based on a real-time feedback correction mechanism.

[0013] In one embodiment, the real-time conversion rate measurement of the corona discharge region after coordinated adjustment is obtained, and dynamic optimization is performed based on a real-time feedback correction mechanism. Specifically, the real-time conversion rate measurement is compared with the conversion efficiency prediction model to obtain a conversion deviation signal; a dynamic feedback correction model is constructed based on the conversion deviation signal, and the weight coefficients of the conversion efficiency prediction model are updated through an adaptive weighting factor; based on the updated conversion efficiency prediction model, the oil mist molecule conversion rate and electric field uniformity index are recalculated and input into the multi-parameter coordinated optimization objective function to obtain the updated optimization parameters.

[0014] Secondly, this application provides a remote control system for a distributed transformer oil treatment device, the system comprising:

[0015] The model building module is used to obtain the physical property parameters of waste transformer oil and establish a droplet dynamics model. The initial particle size distribution of oil mist under different ultrasonic frequencies and powers is obtained through the model.

[0016] The data correction module is used to correct the initial particle size distribution based on multiple influencing factors of oil mist particle size distribution, and generate high-precision real-time particle size prediction values.

[0017] The signal generation module is used to compare the high-precision real-time particle size prediction value with the preset target particle size range and generate a smooth deviation signal.

[0018] The atomization execution module is used to dynamically adjust the working frequency and power of the ultrasonic generator based on the smoothing deviation signal, and to atomize the waste transformer oil to obtain oil mist particle size that meets the target particle size range.

[0019] The oil mist treatment module is used to guide the atomized oil mist into the corona discharge region to perform molecular conversion reactions and perform feedback optimization.

[0020] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0021] 1. By employing a comprehensive strategy encompassing "physical property parameter modeling—multi-factor hierarchical correction—dynamic filtering incremental update—closed-loop adaptive control," this approach achieves high-precision, real-time prediction and dynamic control of the atomized particle size of waste transformer oil. Compared to traditional prediction methods relying on single-bubble assumptions or fixed empirical parameters, this scheme not only introduces the Rayleigh–Plesset equation dynamic model but also incorporates neural network correction based on multiple influencing factors such as bubbles, impurities, and flow fields. Furthermore, it utilizes extended Kalman filtering for dynamic updates, effectively overcoming the problems of low prediction accuracy and poor adaptability to complex operating conditions under ideal assumptions. Simultaneously, by combining a closed-loop control mechanism that smooths the deviation signal and uses PID regulation, the frequency and power of the ultrasonic generator can be quickly and stably adjusted, ensuring the oil mist particle size remains within the target range. This guarantees a uniform and stable atomization process, improving the system's accuracy, response speed, robustness, and applicability, providing a reliable guarantee for the efficient atomization and subsequent utilization of waste transformer oil.

[0022] 2. By incorporating key parameters such as oil mist particle size, airflow velocity, electric field strength, and temperature into the conversion efficiency prediction model, and combining electric field uniformity index, multi-parameter collaborative optimization, and real-time feedback correction mechanism, the molecular conversion process of oil mist in the corona discharge region has been transformed from traditional static control to dynamic adaptive control. This scheme can not only significantly improve the conversion efficiency and uniformity of oil mist to acetylene, methane, and other alkane gases, but also enhance the robustness and intelligence of the system under complex operating conditions, enabling continuous optimization of corona voltage, current density, and electrode spacing, ensuring a stable, efficient, safe, and reliable conversion process. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the remote control method for a distributed transformer oil treatment device provided in an embodiment of this application.

[0024] Figure 2 A schematic diagram of the remote control system structure of the distributed transformer oil treatment equipment provided in this application embodiment.

[0025] Figure 3 The oil mist radius varies with time according to the embodiments of this application. Oscillation curve.

[0026] Figure 4 This is a scatter plot of oil mist particle size and ultrasonic parameters provided for an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1 As shown in the schematic diagram, the remote control method for the distributed transformer oil treatment equipment provided by the present invention includes the following steps:

[0029] The idea / objective of this invention is to improve the uniform atomization effect of the ultrasonic generator on the oil by controlling the ultrasonic generator, and to convert the atomized transformer oil through the corona discharge method, thereby improving the conversion efficiency of the corona reaction on the transformer oil.

[0030] S1. Obtain the physical property parameters of waste transformer oil, establish a droplet dynamics model, and obtain the initial particle size distribution of oil mist under different ultrasonic frequencies and powers through the model.

[0031] In this example, the physical properties include oil temperature, viscosity, density, and surface tension. The surface tension is obtained by placing a tensiometer at the oil sampling point or on the oil surface; the viscosity is obtained by placing a viscosity sensor in the oil flow area or oil tank area; the density is obtained by placing an online density meter at the oil sampling point; and the temperature is obtained by placing a thermistor or thermocouple sensor in the oil flow path.

[0032] Specifically, the physical properties of waste transformer oil were obtained, a droplet dynamics model was established, and the initial particle size distribution of oil mist under different ultrasonic frequencies and powers was obtained through the model.

[0033] The physical property parameters were preprocessed and used as input variables. A droplet dynamics model was constructed based on the Rayleigh–Plesset equation to describe the formation and cavitation behavior of oil mist under single bubble or mean field conditions.

[0034] Based on the model, the continuous time interval is discretized to establish a finite time step;

[0035] Based on a finite time step, the droplet dynamics model is solved iteratively step by step using the finite difference method to obtain the oil mist radius data corresponding to each time step;

[0036] like Figure 3 As shown, an oscillation curve of the oil mist radius changing with time is plotted based on the oil mist radius data to reflect the expansion and contraction dynamics of the oil mist under the action of an ultrasonic field.

[0037] The steady-state determination of the oscillation curve is based on the difference in oil mist radius data between two adjacent periods, specifically as follows:

[0038] The specific calculation formula for the difference in oil mist radius data between two adjacent periods is as follows:

[0039]

[0040] In the formula, The difference in oil mist radius between adjacent cycles is represented by T, where T is the ultrasonic oscillation period. This is the start time of the steady-state period. This represents the oil mist radius data at time t.

[0041] like If the value is less than the preset steady-state threshold, it is considered that the steady-state time period has been entered; otherwise, it is considered that it has not been entered.

[0042] Based on the judgment results, the oil mist radius data within the steady-state time period are statistically analyzed to calculate the initial particle size distribution.

[0043] The initial particle size distribution is calculated using the following formula:

[0044]

[0045] In the formula, For the initial particle size distribution, This represents the number of data points where the oil mist radius falls within the j-th particle size range. denoted as the total number of sampling points within the steady-state time period, where j = 1, 2, 3, ..., M, and M represents the total number of particle size intervals to which the oil mist radius is divided within the steady-state time period.

[0046] The specific calculation formula for the droplet dynamics model is as follows:

[0047]

[0048] In the formula, Where is the oil mist radius, The rate of change of oil mist radius. Let the radius of the oil mist be the acceleration. For the density of the oil, For the surface tension of the oil, This refers to the viscosity of the oil. The internal pressure of the bubble, For environmental static pressure, Pressure is exerted by an ultrasonic field.

[0049] It should be noted that the aforementioned droplet dynamics model considers the nonlinear dynamic behavior of the oil mist radius changing with time under the action of an ultrasonic field, and incorporates the coupling effects of ultrasonic sound pressure amplitude, operating frequency, and oil viscosity on the dynamic response of the oil mist radius into the model calculation. The ultrasonic field pressure is an external pressure that changes with time, generated by the ultrasonic waves applied to the oil mist surface, and can typically be expressed as a sinusoidal function. ,in The sound pressure level is the amplitude. t represents the ultrasonic angular frequency, and t represents time.

[0050] S2 collects the actual oil mist particle size distribution, and based on the multiple influencing factors of the oil mist particle size distribution, corrects the initial particle size distribution to generate a high-precision real-time particle size prediction value.

[0051] In this embodiment, the actual oil mist particle size distribution is collected. Based on multiple influencing factors of the oil mist particle size distribution, the initial particle size distribution is corrected to generate a high-precision real-time particle size prediction value. The multiple influencing factors include bubble characteristic parameters, impurity characteristic parameters, and flow field characteristic parameters, specifically:

[0052] The actual oil mist particle size distribution is compared with the initial particle size distribution to obtain the deviation between the two and generate training samples.

[0053] Based on neural networks, a feedforward neural network model is constructed, with training samples as training inputs and the output layer as correction factors. The neural network is trained through supervised learning, enabling the model to learn the systematic deviation patterns between theoretical predictions and actual measurements under different working conditions.

[0054] Bubble feature parameters are acquired and imported into a trained neural network model. By learning the relationship between the theoretically predicted particle size and the actual particle size deviation under different bubble conditions, the first correction factor for each particle size interval is output, reflecting the influence of multi-bubble interaction on particle size distribution. The bubble feature parameters include bubble concentration, bursting rate, and merging rate.

[0055] The first correction factor is applied to the initial particle size distribution, and the first corrected particle size distribution is obtained by point-by-point multiplication.

[0056] Among them, the first modified particle size distribution reflects the particle size changes caused by bubble breakage, merging and interaction, which solves the limitations of the single bubble assumption of the Rayleigh–Plesset equation and more accurately predicts the particle size distribution characteristics of oil mist in a multi-bubble environment.

[0057] The impurity characteristic parameters are obtained and combined with the first corrected particle size distribution and input into the neural network model. By learning the role of impurities in nucleation and oil mist growth, they are mapped to the second correction factor for each particle size range. The second correction factor is applied to the first corrected particle size distribution through a point-by-point multiplication method to obtain the second corrected particle size distribution.

[0058] The second corrected particle size distribution reflects the influence of impurities on particle size increase or small particle formation, improving the accuracy of particle size prediction in oil environments containing impurities. The impurity characteristic parameters include impurity concentration, impurity particle size range, and impurity type (solid particles, microparticles). The neural network uses these impurity characteristic parameters as input features to learn the role of impurities in the nucleation process. Higher impurity concentrations result in more large-particle components in the oil mist particle size distribution. The network outputs a correction factor by fitting the relationship between "impurity characteristic parameters and distribution deviation."

[0059] The flow field characteristic parameters are obtained, and the flow field characteristic parameters and the second corrected particle size distribution are used as input to the neural network model. By learning the deviation law between theoretical prediction and actual particle size distribution under different flow field conditions, the third correction factor is output. The third correction factor is applied to the second corrected particle size distribution through the point-by-point multiplication method to obtain the third corrected particle size distribution.

[0060] The third corrected particle size distribution is incrementally updated based on real-time operating conditions to obtain the final high-precision real-time particle size prediction value.

[0061] The third correction of particle size distribution fully considers the dynamic influence of non-ideal flow fields on particle size distribution, overcoming the problem of large prediction deviations in droplet dynamics models in non-ideal flow fields, and providing particle size distribution predictions that are closer to the actual oil mist environment. The flow field characteristic parameters include turbulence intensity, velocity gradient, and vortex scale. The neural network uses these flow field characteristic parameters to correct the particle size distribution deviations caused by turbulent shearing, vortex entrainment, etc. For example, high turbulence intensity often promotes the generation of small-diameter droplets; the network learns the deviation patterns under different flow field parameters and outputs the corrected distribution.

[0062] It should be noted that bubble characteristic parameters, impurity characteristic parameters, and flow field characteristic parameters can be obtained directly through experiments or online sensors. In cases where direct measurement is not possible, they can also be indirectly estimated based on historical operating data, empirical formulas, or numerical simulation results to obtain approximate values ​​of each parameter under the current operating conditions. These approximate values ​​can then be used as input to the neural network to achieve high-precision correction and real-time prediction of particle size distribution.

[0063] By performing hierarchical corrections to multiple influencing factors such as bubble characteristics, impurity characteristics, and flow field characteristics, the limitations of the basic model under the assumptions of single bubble, ideal liquid, and stable flow field can be overcome, significantly improving the accuracy of oil mist particle size prediction. At the same time, the interpretability and applicability of the model are enhanced, enabling it to adapt to particle size variations in complex environments under different working conditions.

[0064] Furthermore, the third corrected particle size distribution is incrementally updated based on real-time operating conditions to obtain the final high-precision real-time particle size prediction value, specifically:

[0065] The third corrected particle size distribution is used as the prior state for filtering. The prior state serves as the initial basis for subsequent dynamic estimation and is used to reflect the distribution of oil mist particle size at the current moment.

[0066] Real-time operating parameters are collected, and the real-time operating parameters and actual oil mist particle size distribution are input to the filter as observation input for the prior state;

[0067] The real-time operating parameters refer to process quantities that reflect the current operating status of the equipment or environment. These parameters are closely related to changes in oil mist particle size and include equipment operating parameters such as rotational speed, load, current, and torque; lubricating oil temperature, pressure, flow rate, viscosity, and parameters directly related to the oil mist generation mechanism; and external conditions affecting oil mist diffusion and condensation such as ambient temperature, humidity, and airflow speed.

[0068] Based on the prior state and the observed input, the extended Kalman filter method is used to recursively estimate the oil mist particle size distribution and obtain the dynamically corrected particle size distribution state.

[0069] The observation residuals are obtained based on the particle size distribution, and abrupt changes are detected.

[0070] If the observed residual exceeds the preset mutation threshold, a mutation occurs, triggering the incremental update mechanism of the filter to dynamically adjust the filter gain for compensation and optimization.

[0071] Outputs optimized, high-precision real-time particle size predictions.

[0072] The specific formula for calculating the observation residual is as follows:

[0073]

[0074] In the formula, To observe the residuals, This represents the actual oil mist particle size distribution. H represents the corrected particle size distribution, and H is the observation matrix.

[0075] The incremental update mechanism is specifically calculated using the following formula:

[0076]

[0077] in, The Kalman gain matrix is ​​calculated using the following formula:

[0078]

[0079] In the formula, This refers to the real-time particle size distribution prediction value updated by the filtering increment, i.e., the high-precision real-time particle size prediction value. To predict the covariance matrix, To observe the noise covariance matrix, This is the transpose of the observation matrix.

[0080] It should be noted that, based on the particle size prediction obtained by correcting for multiple influencing factors, Kalman filtering is introduced for incremental updates. This can dynamically combine real-time observations and prior predictions, filter out measurement noise, quickly respond to sudden operating conditions, and ensure high accuracy, stability, and continuity of particle size prediction in actual operation.

[0081] The two-step strategy of combining multi-influencing factor correction with Kalman filter incremental update takes into account both static correction and dynamic adaptation. It not only compensates for the systematic deviation of the basic model, but also corrects the observed disturbance in real time, achieving high-precision and reliable real-time particle size prediction. This provides a robust basis for closed-loop control and improves the stability and reliability of oil mist atomization particle size control.

[0082] S3 generates a smooth deviation signal by comparing the high-precision real-time particle size prediction value with the preset target particle size range.

[0083] In this embodiment, a smooth deviation signal is generated by comparing the high-precision real-time particle size prediction value with a preset target particle size range, specifically as follows:

[0084] The high-precision real-time particle size prediction value is compared with the preset target particle size range, and the deviation is calculated point by point to obtain the deviation signal;

[0085] The specific calculation formula for the deviation signal is as follows:

[0086]

[0087] In the formula, This is a deviation signal. This is the center value of the target particle size range.

[0088] Smooth the deviation signal to remove the effects of short-term fluctuations;

[0089] Deviation signal Divide into several continuous subsequences according to a preset time window W. , used for local signal processing, where N is the number of sampling points within the time window;

[0090] For each time window, calculate the local weighted average of the bias signal subsequence, where the weights are... Adjust according to the rate of change of the signal at the current sampling point;

[0091] The weight The specific calculation formula is as follows:

[0092]

[0093] The local weighted average The specific calculation formula is as follows:

[0094]

[0095] In the formula, For a moment The deviation signal.

[0096] By concatenating the local weighted averages in chronological order, a preliminary smoothed deviation signal is obtained. ;

[0097] The initial smoothed deviation signal is subjected to adaptive low-pass filtering to generate a smoothed deviation signal. , where the filter coefficients The filter coefficient is dynamically adjusted based on the rate of change of the real-time deviation signal: when the signal changes slowly, the filter coefficient is increased to improve the response speed; when the signal has short-term spikes or fluctuations, the filter coefficient is decreased to suppress sudden fluctuations.

[0098] The smoothing deviation signal The specific calculation formula is as follows:

[0099]

[0100] In the formula, This is the smoothed deviation signal from the previous moment.

[0101] Among them, the filter coefficients The specific calculation formula is as follows:

[0102]

[0103] In the formula, This is the adjustment coefficient.

[0104] It should be noted that by combining local weighted averaging with adaptive low-pass filtering, short-term spikes and sudden fluctuations in the deviation signal are effectively suppressed, while the slow trend of deviation change is preserved. This achieves continuous, smooth, and stable processing of the oil mist particle size deviation signal, enabling the closed-loop control unit to obtain reliable and real-time control input, improving the accuracy and response speed of ultrasonic generator adjustment, and enhancing stability and robustness.

[0105] S4, based on the smoothing deviation signal, dynamically adjusts the working frequency and power of the ultrasonic generator to atomize the waste transformer oil and obtain oil mist particle size that meets the target particle size range.

[0106] In one exemplary embodiment, Figure 4 The scatter plot of oil mist particle size and ultrasonic parameters provided in the embodiments of this application is as follows: Figure 4 As shown, based on the smoothed deviation signal, the operating frequency and power of the ultrasonic generator are dynamically adjusted to obtain an oil mist particle size that conforms to the target particle size range, specifically:

[0107] The smoothing deviation signal is input into the ultrasonic generator control model to calculate the initial frequency adjustment and initial power adjustment. The control model adopts the PID control algorithm.

[0108] The specific calculation formula for the initial frequency adjustment is as follows:

[0109]

[0110] The initial power adjustment amount is calculated using the following formula:

[0111]

[0112] In the formula, , , , , , For adjustable controllable gain, The sampling time interval, This is the initial frequency adjustment amount. This is the initial power adjustment amount.

[0113] Based on the initial frequency adjustment and power adjustment, the adjustment direction and preliminary adjustment are extracted, a limit judgment is performed, and the final adjustment is obtained. The final adjustment includes the final frequency adjustment and the final power adjustment.

[0114] The final adjustment amount is calculated using the following formula:

[0115]

[0116]

[0117] In the formula, For the final frequency adjustment amount, This is a sign function used to determine the adjustment direction. , These are the proportional coefficients for frequency and power regulation, respectively. This represents the maximum allowable adjustment range of the frequency. This represents the maximum allowable adjustment range of power.

[0118] The final adjustment is sent to the ultrasonic generator to modify the operating frequency and power in real time, and the waste transformer oil is atomized to obtain a uniform oil mist, so that the oil mist particle size is continuously maintained within the target particle size range.

[0119] The specific calculation formula for the modified operating frequency is as follows:

[0120]

[0121] The specific calculation formula for the power modification is as follows:

[0122]

[0123] In the formula, Let be the operating frequency of the ultrasonic generator at the current time t. Let be the operating frequency of the ultrasonic generator at the previous moment t−1. Let be the operating power of the ultrasonic generator at the current time t. This represents the operating power of the ultrasonic generator at the previous moment t-1.

[0124] It should be noted that the adjustment direction is determined by the smoothing deviation signal, ensuring that the direction of the adjustment is consistent with the deviation direction. A positive value in the smoothing deviation signal indicates a larger deviation, i.e., reducing the frequency / power; a smaller deviation indicates an increase. The target particle size range is the optimal oil mist particle size distribution that conforms to the subsequent corona reaction.

[0125] S5 introduces the atomized oil mist into the corona discharge zone to undergo molecular conversion reactions to generate alkane gases and then performs feedback optimization.

[0126] In this embodiment, the atomized oil mist is introduced into the corona discharge region to undergo a molecular conversion reaction to generate alkane gas, and feedback optimization is performed. Specifically:

[0127] The atomized oil mist is introduced into the corona discharge area, so that the oil mist passes between the high-voltage electrode plates. The corona discharge area includes positive and negative electrode plates, and the spacing between the plates is adjustable to ensure that the corona discharge uniformly covers the oil mist. The electrode plates are connected to a high-voltage DC power supply to form a stable corona discharge field.

[0128] Based on the interaction between oil mist particle size distribution and corona discharge intensity, a conversion efficiency prediction model is established. The conversion efficiency prediction model takes oil mist particle size, airflow velocity, electric field intensity and temperature that meet the target particle size range as input variables and oil mist molecule conversion rate as output variable.

[0129] The specific calculation formula for the conversion efficiency prediction model is as follows:

[0130]

[0131] In the formula, For oil mist molecule conversion rate, Based on the conversion rate, Let v be the average oil mist particle size, and v be the oil mist airflow velocity. For electric field strength, Temperature of the reaction zone , , , These are the weighting coefficients.

[0132] During the corona discharge process, first data is collected at different positions between the electrodes, and a weighted electric field uniformity index is calculated to evaluate the overall reaction uniformity of the corona region. The first data includes local electric field strength, oil mist particle residence time, and local flow field disturbance.

[0133] The specific formula for calculating the weighted electric field uniformity index is as follows:

[0134]

[0135] In the formula, As an index of electric field uniformity, To measure the number of points, The average electric field strength is Let be the local electric field intensity at the i-th position.

[0136] Based on the conversion efficiency prediction model and electric field uniformity index, a multi-parameter collaborative optimization objective function is established, and a multi-variable control method is used to collaboratively adjust the corona voltage, current density, and electrode spacing according to the results of the optimization objective function.

[0137] The real-time conversion rate of the corona discharge region after coordinated adjustment is obtained, and dynamic optimization is performed based on the real-time feedback correction mechanism.

[0138] The specific calculation formula for the multi-parameter collaborative optimization objective function is as follows:

[0139]

[0140] In the formula, To achieve multi-parameter collaborative optimization of the objective function, For the target conversion rate, , These are the weighting coefficients. Corona voltage, d is the current density and d is the distance between the plates.

[0141] Furthermore, the real-time conversion rate measurement of the corona discharge region after coordinated adjustment is obtained, and dynamic optimization is performed based on a real-time feedback correction mechanism, specifically as follows:

[0142] The real-time conversion rate measurement is compared with the conversion efficiency prediction model to obtain the conversion deviation signal;

[0143] A dynamic feedback correction model is constructed based on the conversion deviation signal, and the weight coefficients of the conversion efficiency prediction model are updated by an adaptive weighting factor.

[0144] Based on the updated conversion efficiency prediction model, the oil mist molecule conversion rate and electric field uniformity index are recalculated and input into the multi-parameter collaborative optimization objective function to obtain the updated optimization parameters, thereby realizing dynamic feedback optimization of the oil mist molecule conversion process. The optimization parameters include corona voltage, current density and electrode spacing.

[0145] The specific calculation formula for the dynamic feedback correction model is as follows:

[0146]

[0147] In the formula, The weight coefficients for the j-th input variable in the conversion efficiency prediction model during the g-th iteration are: For learning rate, To convert the deviation signal, The input variables for the conversion efficiency prediction model at the g-th iteration include oil mist particle size, airflow velocity, electric field strength, and temperature.

[0148] It should be noted that, through a closed-loop mechanism of predictive modeling, multi-parameter collaborative optimization, and real-time feedback correction, the molecular conversion process of oil mist in the corona discharge region is transformed from traditional static control to dynamic adaptive control. First, a conversion efficiency prediction model is established using oil mist particle size distribution, electric field strength, airflow velocity, and temperature to achieve feedforward adjustment of the conversion trend. Second, an electric field uniformity index and a multi-parameter collaborative optimization objective function are introduced to ensure that the control not only pursues conversion rate improvement but also considers the stability and uniformity of the reaction region. Finally, through real-time conversion rate measurement and dynamic feedback correction, an adaptive weighted update mechanism is used to continuously correct the prediction model parameters, enabling it to evolve in response to fluctuations in operating conditions, thereby achieving optimal adjustment of corona voltage, current density, and electrode spacing. This scheme not only improves the efficiency and stability of oil mist molecules converting into alkane gases but also enhances the system's robustness and intelligence, maintaining efficient and reliable conversion results under complex operating conditions.

[0149] Reference Figure 2 As shown in the diagram, the remote control system structure of the distributed transformer oil treatment equipment provided by this invention includes: a model building module, a data correction module, a signal generation module, an atomization execution module, and an oil mist treatment module, with connections between the modules.

[0150] The model building module is used to obtain the physical property parameters of waste transformer oil and establish a droplet dynamics model. The initial particle size distribution of oil mist under different ultrasonic frequencies and powers is obtained through the model.

[0151] The data correction module is used to correct the initial particle size distribution based on multiple influencing factors of oil mist particle size distribution, and generate high-precision real-time particle size prediction values.

[0152] The signal generation module is used to compare the high-precision real-time particle size prediction value with the preset target particle size range and generate a smooth deviation signal.

[0153] The atomization execution module is used to dynamically adjust the working frequency and power of the ultrasonic generator based on the smoothing deviation signal, and to atomize the waste transformer oil to obtain oil mist particle size that meets the target particle size range.

[0154] The oil mist treatment module is used to guide the atomized oil mist into the corona discharge region to perform molecular conversion reactions and perform feedback optimization.

[0155] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0156] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other 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.

[0157] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0158] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0159] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0160] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A remote control method for distributed transformer oil treatment equipment, characterized in that, include: The physical properties of waste transformer oil were obtained and a droplet dynamics model was established. The initial particle size distribution of oil mist under different ultrasonic frequencies and powers was obtained through the model. The initial particle size distribution is corrected to generate high-precision real-time particle size prediction values, specifically as follows: A training sample set is generated by the deviation between the actual oil mist particle size distribution and the initial particle size distribution. Build a neural network model and train the model using a training sample set; The bubble feature parameters are imported into the trained neural network model, and the first correction factor is output. The initial particle size distribution is corrected based on the first correction factor to obtain the corrected particle size distribution; The third correction particle size distribution is incrementally updated based on real-time operating parameters to generate the final high-precision real-time oil mist particle size prediction result. The process of correcting the initial particle size distribution based on the first correction factor to obtain the corrected particle size distribution is as follows: Applying the first correction factor to the initial particle size distribution yields the first corrected particle size distribution; The impurity characteristic parameters and the first corrected particle size distribution are input into the neural network model, the second correction factor is output, and the first corrected particle size distribution is corrected to obtain the second corrected particle size distribution. The flow field characteristic parameters and the second corrected particle size distribution are used as inputs to the neural network model, and the third correction factor is output. Applying the third correction factor to the second corrected particle size distribution yields the third corrected particle size distribution. A smoothed deviation signal is generated by comparing the predicted value with a preset target particle size range. The working frequency and power of the ultrasonic generator are dynamically adjusted based on the smooth deviation signal, and the waste transformer oil is atomized to obtain oil mist particle size that meets the target particle size range. The atomized oil mist is introduced into the corona discharge region to perform molecular conversion reactions and feedback optimization.

2. The remote control method for the distributed transformer oil treatment equipment according to claim 1, characterized in that, The establishment of a droplet dynamics model, and the acquisition of the initial particle size distribution of oil mist under different ultrasonic frequencies and powers through the model, specifically involves: The physical property parameters are preprocessed, and a droplet dynamics model is constructed based on the processed parameters; Solve the droplet dynamics model to obtain the oil mist radius data for each time step; Oscillation curves were plotted based on oil mist radius data, and steady-state determination was performed on the oscillation curves. Based on the judgment results, statistical analysis was performed on the oil mist radius data within the steady-state time period to obtain the initial particle size distribution.

3. The remote control method for the distributed transformer oil treatment equipment according to claim 1, characterized in that, The incremental update of the corrected particle size distribution based on real-time operating parameters generates the final high-precision real-time oil mist particle size prediction result, specifically as follows: The third corrected particle size distribution is used as the prior state for filtering; Real-time operating parameters are collected, and the real-time operating parameters and actual oil mist particle size distribution are input to the filter as observation input for the prior state; Based on the prior state and the observed input, the oil mist particle size distribution is recursively estimated to obtain the dynamically corrected particle size distribution state. The observation residuals are obtained based on the particle size distribution, and abrupt changes are detected. If the observed residual exceeds the preset mutation threshold, a mutation occurs, triggering the incremental update mechanism of the filter. The filter gain is dynamically adjusted for compensation and optimization, and a high-precision real-time particle size prediction value is output.

4. The remote control method for the distributed transformer oil treatment equipment according to claim 1, characterized in that, Based on the comparison between the predicted value and the preset target particle size range, a smoothed deviation signal is generated, specifically as follows: The deviation signal is obtained by calculating the deviation between the high-precision real-time particle size prediction value and the preset target particle size range; The deviation signal is divided into several continuous subsequences according to a preset time window; For each time window, calculate the local weighted average of the deviation signal subsequence, where the weights are adjusted according to the signal change rate at the current sampling point; By concatenating the local weighted averages in chronological order, a preliminary smoothed deviation signal is obtained; An adaptive low-pass filter is applied to the initial smoothed deviation signal to generate a smoothed deviation signal. The low-pass filter includes a filter coefficient that is dynamically adjusted according to the rate of change of the real-time deviation signal.

5. The remote control method for the distributed transformer oil treatment equipment according to claim 4, characterized in that, The process involves dynamically adjusting the operating frequency and power of the ultrasonic generator based on a smoothed deviation signal, and atomizing the waste transformer oil to obtain an oil mist particle size that meets the target particle size range. Specifically: The smoothing deviation signal is input into the ultrasonic generator control model to calculate the initial frequency adjustment and the initial power adjustment. Based on the initial frequency and power adjustment values, the adjustment direction and preliminary adjustment value are extracted, a limit judgment is performed, and the final adjustment value is obtained. The final adjustment is sent to the ultrasonic generator to modify the operating frequency and power in real time, and the waste transformer oil is atomized to obtain a uniform oil mist, so that the oil mist particle size is continuously maintained within the target particle size range.

6. The remote control method for the distributed transformer oil treatment equipment according to claim 1, characterized in that, The process of introducing the atomized oil mist into the corona discharge region to perform molecular conversion reactions and feedback optimization is as follows: The atomized oil mist is introduced into the corona discharge region, allowing the oil mist to pass between the high-voltage electrode plates. The corona discharge region includes positive and negative electrode plates, and the spacing between the plates is adjustable. A conversion efficiency prediction model is established based on the interaction between oil mist particle size distribution and corona discharge intensity. During the corona discharge process, first data are collected at different positions between the electrodes, and the weighted electric field uniformity index is calculated. Based on the conversion efficiency prediction model and electric field uniformity index, a multi-parameter collaborative optimization objective function is established, and a multi-variable control method is used to collaboratively adjust the corona voltage, current density, and electrode spacing according to the optimization objective function results. The real-time conversion rate of the corona discharge region after coordinated adjustment is obtained, and dynamic optimization is performed based on the real-time feedback correction mechanism.

7. The remote control method for the distributed transformer oil treatment equipment according to claim 6, characterized in that, The acquisition of the real-time conversion rate measurement value of the corona discharge region after coordinated adjustment is dynamically optimized based on a real-time feedback correction mechanism, specifically as follows: The real-time conversion rate measurement is compared with the conversion efficiency prediction model to obtain the conversion deviation signal; A dynamic feedback correction model is constructed based on the conversion deviation signal, and the weight coefficients of the conversion efficiency prediction model are updated by adaptive weighting factors. Based on the updated conversion efficiency prediction model, the oil mist molecule conversion rate and electric field uniformity index were recalculated and input into the multi-parameter collaborative optimization objective function to obtain the updated optimization parameters.

8. A system for remote control of a distributed transformer oil treatment device as described in any one of claims 1-7, characterized in that, include: The model building module is used to obtain the physical property parameters of waste transformer oil and establish a droplet dynamics model. The initial particle size distribution of oil mist under different ultrasonic frequencies and powers is obtained through the model. The data correction module is used to correct the initial particle size distribution based on multiple influencing factors of oil mist particle size distribution, and generate high-precision real-time particle size prediction values. The signal generation module is used to compare the high-precision real-time particle size prediction value with the preset target particle size range and generate a smooth deviation signal. The atomization execution module is used to dynamically adjust the working frequency and power of the ultrasonic generator based on the smoothing deviation signal, and to atomize the waste transformer oil to obtain oil mist particle size that meets the target particle size range. The oil mist treatment module is used to guide the atomized oil mist into the corona discharge region to perform molecular conversion reactions and perform feedback optimization.

Images