A control method of a transformer explosion-proof device integrated with multi-stage pressure relief channels

Abstract

This application relates to the field of power transformer safety protection and fault control technology, specifically disclosing a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels. This method addresses the vaporization of insulating oil and pressure shock in the tank caused by transformer arc faults by acquiring high-frequency fault current sampling signals and dynamic pressure wave sampling signals, extracting the arc energy injection rate and tank pressure time gradient, calculating the insulating oil vaporization expansion volume demand curve, and inverting the target equivalent discharge cross-section time series curve under safe pressure and two-phase flow boundary constraints. It then performs discrete step-wise approximation optimization by combining the fixed aperture of the pressure relief channel and the mechanical action hysteresis characteristics to generate a channel coding sequence. Finally, it triggers the step-wise opening of the channels through a hardware timer array, ensuring that the discharge capacity matches the arc energy changes, thereby reducing the risks of insufficient pressure relief, excessive oil discharge, and tank damage.

Description

Technical Field

[0001] This application relates to the field of power transformer safety protection and fault control technology, and more specifically, to a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels. Background Technology

[0002] During operation, transformers may experience internal arcing faults due to insulation aging, winding short circuits, or external overvoltages. The arcing injects a large amount of heat into the surrounding insulating oil within a short time, causing the oil to rapidly vaporize and expand thousands of times in volume, thus creating a rapidly propagating pressure shock wave inside the tank. When the pressure peak exceeds the tank's pressure limit, it can easily lead to serious accidents such as tank tearing, explosion, insulating oil leakage, and fire.

[0003] Existing transformer explosion-proof devices typically detect abnormal conditions using pressure sensors, gas relays, or other fault sensors. When the pressure reaches a preset threshold, a single pressure relief valve is triggered to open rapidly. While some devices have multiple pressure relief channels, their control methods often employ a crude logic of single-threshold triggering, fixed delay, or simultaneous opening of multiple channels. This type of control relies primarily on open-loop judgment based on the current pressure amplitude, failing to fully utilize the high-frequency fault current change rate, pressure time gradient, and arc energy accumulation process, thus making it difficult to reflect the nonlinear injection characteristics of internal arc energy. Because the energy of the electric arc and the vaporization and expansion of the insulating oil increase rapidly or even abruptly in the early stages of a fault, and the process from receiving the trigger command to forming an effective flow cross-section is limited by the valve core mass, driving force, and displacement inertia, the opening of the mechanical valve usually changes in a relatively gentle linear or gradual manner. This results in the valve not yet forming a sufficient flow area in the early stages of pressure relief, leading to a mismatch between the pressure relief capacity and the energy injection rate, and making it easy to miss the pressure peak suppression window. After the arc energy decays, the valve may reach a large opening, resulting in excessive pressure relief capacity, a large amount of insulating oil being discharged, and even the formation of local negative pressure and collapse of the enclosure due to oil flow inertia and the impact of gas-liquid two-phase flow. In addition, existing hardware timing control usually uses a uniform time quantization method for each channel, without distinguishing the degree of drastic change in the relief demand corresponding to different trigger times. Highly sensitive channels may experience non-negligible action delays due to quantization lag, further reducing the reliability of explosion-proof control.

[0004] Therefore, an optimized control scheme for an integrated multi-stage pressure relief channel transformer explosion-proof protection device is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels, comprising: S1, perform transient feature extraction on the received explosion-proof sensor raw data to obtain the arc energy injection rate characterization vector and the oil tank pressure time gradient sequence; S2, calculate the vaporization expansion volume of insulating oil by performing vaporization expansion volume calculation on the arc energy injection rate characterization vector and the oil tank pressure time gradient sequence. S3, based on the explosion-proof voltage reduction tracking constraint function, performs inverse fluid dynamics solution on the insulating oil vaporization expansion volume demand curve to obtain the target equivalent discharge section time series curve; S4. Based on the fixed aperture area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels, the time series curve of the target equivalent discharge section is subjected to discrete step approximation optimization to obtain the multi-stage channel array combination coding sequence. S5 uses a hardware timer array to count down in parallel, performing frame parsing and delayed triggering of the multi-level channel array combination encoding sequence to drive each physical pressure relief channel to open in a stepwise manner according to the encoding sequence.

[0006] Compared with existing technologies, this application proposes a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels. It synchronously acquires high-frequency fault current sampling signals from the transformer and dynamic pressure wave sampling signals from the tank, extracts the arc energy injection rate and tank pressure time gradient, calculates the insulating oil vaporization expansion volume demand curve, and, under the tank safety pressure constraint and gas-liquid two-phase flow discharge boundary, reversely solves the target equivalent discharge cross-section time-series curve matching the transient expansion demand. Then, combining the fixed aperture area and mechanical action hysteresis characteristics of each pressure relief channel, it performs discrete stepwise approximation optimization on the target curve, generating a multi-stage channel array combination encoding sequence. Through parallel parsing and delayed triggering of each physical pressure relief channel using a hardware timer array, the multi-stage channels are coordinated and opened in a stepwise manner according to the arc energy evolution process. This ensures that the discharge capacity matches the dynamic energy injection process, reducing transformer tank damage and insulating oil loss caused by insufficient pressure relief in the early stages of a fault and excessive pressure relief in the later stages. Attached Figure Description

[0007] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0008] Figure 1 This is a flowchart of a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application; Figure 2This is a data flow diagram illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application; Figure 3 This is a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application. It involves using an explosion-proof voltage reduction tracking constraint function to perform a reverse fluid dynamics solution on the insulating oil vaporization expansion volume demand curve to obtain the target equivalent discharge cross-section time series curve. Figure 4 This is a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multiple pressure relief channels according to an embodiment of this application. It describes a method for obtaining a multi-level channel array combination encoding sequence by performing discrete step approximation optimization on the time series curve of the target equivalent discharge cross-section based on the fixed aperture area and mechanical inertial displacement lag time of each of the multiple pressure relief channels. Figure 5 This is a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multiple pressure relief channels according to an embodiment of this application. The method involves parallel countdown using a hardware timer array to perform frame parsing and delayed triggering execution on the combined encoding sequence of the multi-level channel array to drive each physical pressure relief channel to open in a stepwise manner according to the encoding sequence. Figure 6 This is a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application. The method involves determining the channel timing sensitivity based on the target equivalent discharge cross-section timing curve, and performing heterogeneous adaptive discrete mapping and boundary verification on the absolute trigger timestamps of each channel to obtain the hardware clock tick reload value. Detailed Implementation

[0009] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0010] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0011] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0012] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0013] Currently, transformer explosion-proof protection devices typically control the simultaneous opening of one or more pressure relief channels based on pressure thresholds, making it difficult to dynamically adjust the discharge capacity according to transient changes in internal arc energy and tank pressure. Because the injection of arc energy and the vaporization and expansion of insulating oil exhibit nonlinear changes, and the pressure relief valve is affected by mechanical inertia and lag, the formation rate of its flow area is relatively slow. This easily leads to insufficient pressure relief capacity in the early stages of a fault and excessive discharge cross-section in later stages, thereby increasing the risk of tank rupture, excessive oil discharge, and secondary damage. Therefore, this application proposes a control method for a transformer explosion-proof protection device integrating multiple pressure relief channels. This method first extracts transient features from the high-frequency fault current sampling signal of the transformer and the dynamic pressure wave sampling signal inside the oil tank to obtain the arc energy injection rate characterization vector and the oil tank pressure time gradient sequence, and calculates the vaporization expansion volume demand curve of the insulating oil under the action of the arc. Then, under the constraints of the oil tank safety pressure and the gas-liquid two-phase flow discharge capacity, the vaporization expansion volume demand curve is solved by inverse hydrodynamics to obtain the target equivalent discharge section time series curve that varies with the fault time. Furthermore, by combining the fixed aperture area of ​​each pressure relief channel and the mechanical inertial displacement lag time, the target equivalent discharge section time series curve is optimized by discrete step approximation to generate a multi-level channel array combination encoding sequence. Finally, the encoding sequence is frame parsed and delayed triggered by a hardware timer array to drive the corresponding physical pressure relief channel to open in a step-like manner according to the determined time sequence, thereby compensating for the dynamic mismatch between the response characteristics of a single mechanical valve and the nonlinear energy injection process, and taking into account both the rapid pressure reduction requirement in the early stage of the fault and the appropriate discharge requirement in the later stage of the fault.

[0014] Figure 1 This is a flowchart of a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application. Figure 2 This is a data flow diagram illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application. Figure 1 and Figure 2As shown, a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels according to an embodiment of this application includes: S1, extracting transient features from the received explosion-proof sensor raw data to obtain an arc energy injection rate characterization vector and a tank pressure time gradient sequence; S2, performing vaporization expansion volume calculation on the arc energy injection rate characterization vector and the tank pressure time gradient sequence to obtain an insulating oil vaporization expansion volume demand curve; S3, performing inverse hydrodynamic solution on the insulating oil vaporization expansion volume demand curve based on the explosion-proof voltage reduction tracking constraint function to obtain a target equivalent leakage section time series curve; S4, performing discrete stepwise approximation optimization on the target equivalent leakage section time series curve based on the fixed aperture area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels to obtain a multi-stage channel array combination encoding sequence; S5, performing frame parsing and delayed trigger execution on the multi-stage channel array combination encoding sequence through parallel countdown of a hardware timer array to drive each physical pressure relief channel to open in a stepwise manner according to the encoding sequence.

[0015] Specifically, in step S1, transient feature extraction is performed on the received raw explosion-proof sensor data to obtain the arc energy injection rate characterization vector and the tank pressure time gradient sequence. It should be noted that since the current surge after the formation of the arc inside the transformer and the tank pressure impact both contain steady-state power frequency components, mechanical vibration, and sampling noise, directly using the raw sampled values ​​is difficult to accurately characterize the arc energy release rate and the pressure rise. Based on this, the technical solution of this application first performs transient feature extraction on the received raw explosion-proof sensor data to obtain the arc energy injection rate characterization vector and the tank pressure time gradient sequence. Through the above processing, electrical and pressure characteristics with a unified time-series reference can be provided for the calculation of the insulating oil vaporization expansion volume.

[0016] More specifically, in a specific example of this application, step S1 includes: removing the power frequency component and suppressing vibration noise from the high-frequency fault current sampling signal of the transformer and the dynamic pressure wave sampling signal inside the tank to obtain a transient high-frequency current sequence and a denoised dynamic pressure sequence; calculating the instantaneous rate of change of the current at adjacent sampling times in the transient high-frequency current sequence to obtain the arc energy injection rate characterization vector; and differentiating the rate of change of the pressure values ​​at adjacent sampling times in the denoised dynamic pressure sequence to obtain the tank pressure time gradient sequence.

[0017] More specifically, the raw data from the explosion-proof sensors includes high-frequency fault current sampling signals continuously acquired by a high-frequency current sensor from the transformer, and dynamic pressure wave sampling signals continuously acquired by a pressure sensor inside the tank from the tank. First, based on the timestamps carried by each sample value, the high-frequency fault current sampling signals and the dynamic pressure wave sampling signals inside the tank are aligned to the same sampling time. A notch filter frequency is set according to the transformer's rated power frequency, and a finite impulse response notch filter is applied to the high-frequency fault current sampling signals to remove steady-state power frequency components and their adjacent frequency band components, retaining the high-frequency transient components formed by arc initiation and short-circuit current abrupt changes, thus obtaining a transient high-frequency current sequence. Simultaneously, based on the pre-calibrated mechanical vibration frequency band under normal transformer operation, a band-stop filter is applied to the dynamic pressure wave sampling signals inside the tank, and a median filter is used to suppress isolated pulse noise, obtaining a denoised dynamic pressure sequence that retains the rising edge of the pressure shock wave. Then, adjacent current amplitudes in the transient high-frequency current sequence are read in chronological order of sampling time. The difference between adjacent current amplitudes is divided by the sampling period to calculate the instantaneous rate of change of current corresponding to each sampling time.

[0018] in, Indicates the first Characteristic values ​​of arc energy injection rate at each sampling time. Indicates the first The transient high-frequency current amplitude at each sampling time. This represents the transient high-frequency current amplitude at the previous sampling time. Indicates the sampling period between adjacent sampling times. Indicates the sampling time sequence number; [and] each Arranged chronologically, the arc energy injection rate characterization vector is obtained. Finally, adjacent pressure values ​​in the denoised dynamic pressure sequence are read according to the same sampling period, and the time change rate between adjacent pressure values ​​is calculated:

[0019] in, Indicates the first The time gradient value of tank pressure at each sampling time. Indicates the first The dynamic stress value for denoising at each sampling time. This represents the dynamic pressure value for denoising at the previous sampling time. Indicates the sampling period. Indicates the sampling time sequence number; [and] each Arranged in chronological order, the time gradient sequence of tank pressure is obtained, and it is made to correspond one-to-one with the sampling time of the arc energy injection rate characterization vector.

[0020] Specifically, in step S2, the vaporization expansion volume is calculated using the arc energy injection rate characterization vector and the tank pressure time gradient sequence to obtain the insulating oil vaporization expansion volume demand curve. It should be noted that since the arc energy release intensity and tank pressure changes jointly affect the vaporization quality and expansion volume of the insulating oil, the actual pressure relief demand cannot be determined based solely on a single current or pressure characteristic. Therefore, the technical solution of this application further calculates the vaporization expansion volume using the arc energy injection rate characterization vector and the tank pressure time gradient sequence to obtain the insulating oil vaporization expansion volume demand curve. Through the above processing, a volume demand basis that varies with the fault process can be provided for the subsequent inversion of the target equivalent discharge cross-section.

[0021] More specifically, in a specific example of this application, step S2 includes: accumulating heat energy by weighting the arc energy injection rate characterization vector with heat conversion coefficients based on discrete numerical integration within a sliding time window to obtain a cumulative heat energy equivalent sequence; performing dynamic pressure coupling correction on the atmospheric pressure base latent heat of vaporization based on the tank pressure time gradient sequence to obtain a dynamic latent heat of vaporization sequence; and mapping the cumulative heat energy equivalent sequence and the dynamic latent heat of vaporization sequence to phase change mass and volume based on the gas specific volume constant to obtain the insulating oil vaporization expansion volume demand curve.

[0022] Specifically, firstly, the arc fault initiation time is taken as the starting point of thermal energy accumulation. Elements in the arc energy injection rate representation vector are read in chronological order of sampling times. The absolute value of each arc energy injection rate representation value is taken to eliminate the influence of current change direction on the thermal energy accumulation result. Then, a pre-calibrated arc thermal power equivalent conversion coefficient is used for weighting, converting the representation values ​​at each sampling time into the corresponding thermal power equivalent. Finally, discrete numerical integration is performed from the arc fault initiation time to the current sampling time to obtain the accumulated thermal energy equivalent at the current sampling time. The calculation formula is as follows:

[0023] in, Indicates the first The cumulative thermal energy equivalent at each sampling time. The first element in the vector representing the arc energy injection rate Characteristic values ​​of arc energy injection rate at each sampling time. This represents the arc heat power equivalent conversion factor calibrated through an internal arc fault test of a transformer. Indicates the first Thermal power equivalent at each sampling time Indicates the sampling period. This indicates the sampling time number that participated in the cumulative calculation. Indicates the current sampling time sequence number; calculates the results from each sampling time. Arranged in chronological order, the cumulative thermal energy equivalent sequence is obtained.

[0024] Subsequently, the oil tank pressure time gradient sequence was read according to the same sampling time benchmark. The pressure time gradient value from the start of the arc fault to the current sampling time was discretely integrated to reconstruct the oil tank pressure increment relative to the fault start pressure at the corresponding time. Then, based on the pre-calibrated correspondence between the latent heat of vaporization of insulating oil and oil tank pressure, the latent heat of vaporization at atmospheric pressure was corrected using a pressure coupling correction coefficient to obtain the dynamic latent heat of vaporization at each sampling time. The calculation formula is as follows:

[0025] in, Indicates the first Dynamic latent heat of vaporization at each sampling time This indicates the basic latent heat of vaporization of the corresponding insulating oil under normal pressure conditions. This represents the pressure coupling correction factor determined based on the calibration relationship between the type of insulating oil and the change of latent heat of vaporization with pressure. Represents the first time gradient sequence of tank pressure. Pressure time gradient values ​​at each sampling time point Indicates the time from the start of the arc fault to the [number]th [time]. The tank pressure increment at each sampling time; the calculated values ​​at each sampling time. Arranged in chronological order, a dynamic latent heat of vaporization sequence is obtained.

[0026] Finally, corresponding elements in the cumulative thermal energy equivalent sequence and the dynamic latent heat of vaporization sequence are read at the same sampling time. Based on the conversion relationship between heat and latent heat of phase change, the cumulative thermal energy equivalent is divided by the dynamic latent heat of vaporization to obtain the mass of insulating oil that undergoes phase change after absorbing the corresponding heat energy. Then, the mass of insulating oil is mapped to the vaporization expansion volume using a pre-calibrated gaseous specific volume constant. The calculation formula is as follows:

[0027] in, Indicates the first The mass of insulating oil that undergoes a phase change at each sampling time. Indicates the first The required volume of insulating oil vaporization expansion at each sampling time. This represents the specific volume constant of the gaseous insulating oil calibrated under preset arc vaporization temperature and pressure conditions. Indicates the first The cumulative thermal energy equivalent at each sampling time. Represents the dynamic latent heat of vaporization at the same sampling time; [and] the various The volume demand curve for the vaporization expansion of insulating oil is obtained by arranging the samples sequentially according to their sampling times and connecting them in time sequence.

[0028] Specifically, in step S3, based on the explosion-proof voltage reduction tracking constraint function, the insulating oil vaporization expansion volume demand curve is solved using inverse hydrodynamics to obtain the target equivalent discharge cross-section time-series curve. It should be noted that, given that the insulating oil vaporization expansion volume demand curve only reflects the volume expansion demand over time under the action of the internal electric arc of the transformer, it cannot be directly used as the control basis for multi-stage pressure relief channels. Furthermore, the pressure relief process is simultaneously limited by the upper limit pressure difference of the oil tank and the sound velocity congestion boundary of the gas-liquid two-phase flow. If the pressure relief area is configured only according to the expansion volume, it is easy to cause a mismatch between the actual discharge capacity and the transient expansion rate. Based on this, the technical solution of this application further uses the explosion-proof voltage reduction tracking constraint function to solve the insulating oil vaporization expansion volume demand curve using inverse hydrodynamics to obtain the target equivalent discharge cross-section time-series curve. Through the above processing, the vaporization expansion volume flow demand at each moment can be converted into a discharge cross-section demand that meets the oil tank safety pressure difference and the effective discharge velocity constraints of the two-phase flow, thereby providing a time-series control target for the discrete combination and stepped opening of multi-stage pressure relief channels.

[0029] Figure 3 This document presents a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels, based on an explosion-proof voltage reduction tracking constraint function. It describes a process for obtaining the target equivalent discharge cross-section time series curve by performing a reverse hydrodynamic solution on the insulating oil vaporization expansion volume demand curve. (The flowchart is not translated here as it is not part of the document's description.) Figure 3 As shown, step S3 includes: S31, differentiating the growth rate of the volume values ​​at adjacent moments in the vaporization expansion volume demand curve of insulating oil to obtain the vaporization expansion volume flow rate demand sequence; S32, based on the gas-liquid two-phase flow sound velocity congestion boundary, truncating the theoretical discharge velocity under the upper limit pressure difference condition of the oil tank to obtain the restricted two-phase flow discharge velocity sequence; S33, performing flow cross-sectional area inversion on the vaporization expansion volume flow rate demand sequence and the restricted two-phase flow discharge velocity sequence to obtain the target equivalent discharge cross-section time series curve.

[0030] In step S31, the growth rate of the volume values ​​at adjacent moments in the insulating oil vaporization expansion volume demand curve is differentiated to obtain the vaporization expansion volume flow rate demand sequence. It should be noted that since the insulating oil vaporization expansion volume demand curve represents the cumulative volume demand at each sampling moment, it cannot directly reflect the volume of fluid to be discharged per unit time. Therefore, the technical solution of this application further differentiates the growth rate of the volume values ​​at adjacent moments in the insulating oil vaporization expansion volume demand curve to obtain the vaporization expansion volume flow rate demand sequence. Through the above processing, the vaporization expansion volume demand can be converted into an instantaneous discharge demand that changes with the fault process, providing a flow rate basis for subsequent discharge section inversion.

[0031] More specifically, in a concrete example of this application, the processing is performed in the order of curve discretization, adjacent volume difference, and flow rate sequence generation. First, based on the sampling timestamps carried by the insulating oil vaporization expansion volume demand curve, the vaporization expansion volume values ​​at each sampling time are extracted in chronological order, and adjacent volume values ​​are paired for difference calculation. Then, the vaporization expansion volume value at the current sampling time is subtracted from the vaporization expansion volume value at the previous sampling time to obtain the newly added vaporization expansion volume within the adjacent sampling period. Finally, the newly added vaporization expansion volume is divided by the sampling period between adjacent sampling times to obtain the vaporization expansion volume flow rate demand corresponding to the current sampling time.

[0032] in, Indicates the first The vaporization expansion volumetric flow rate requirement at each sampling time. Indicates the first The required volume of insulating oil vaporization expansion at each sampling time. This represents the required volume of insulating oil vaporization and expansion at the previous sampling time. Indicates the sampling period between adjacent sampling times. Indicates the sampling time sequence number. This indicates a non-negative truncation operation, used to correct negative growth rates caused by sampling errors or numerical fluctuations to zero. For the fault initiation time, the vaporization expansion volumetric demand value from the previous sampling time is set to 0 to calculate the volumetric flow rate demand within the initial sampling period. Finally, the values ​​from each sampling time are... Arrange them in chronological order, and keep their timestamps consistent with the corresponding timestamps in the insulating oil vaporization expansion volume demand curve to obtain the vaporization expansion volume flow demand sequence. Each element in the sequence represents the drainage capacity required to offset the increased vaporization expansion volume of the insulating oil within the corresponding sampling period.

[0033] In step S32, based on the acoustic congestion boundary of the gas-liquid two-phase flow, the theoretical discharge velocity under the upper limit pressure difference condition of the oil tank is truncated by a minimum constraint to obtain a restricted two-phase flow discharge velocity sequence. It should be noted that since the theoretical discharge velocity corresponding to the upper limit pressure difference of the oil tank may exceed the acoustic congestion boundary of the gas-liquid two-phase flow, directly using this theoretical velocity will overestimate the actual discharge capacity of the pressure relief channel. Therefore, the technical solution of this application further truncates the theoretical discharge velocity under the upper limit pressure difference condition of the oil tank based on the acoustic congestion boundary of the gas-liquid two-phase flow to obtain a restricted two-phase flow discharge velocity sequence. Through the above processing, a velocity boundary that conforms to the actual flow capacity of the pressure relief port can be provided for the inversion of the target equivalent venting cross section.

[0034] More specifically, in a particular example of this application, the processing generally follows the order of theoretical discharge velocity calculation, sound velocity congestion boundary determination, and minimum value truncation. First, the upper limit pressure of the oil tank (pre-determined based on the tank's structural strength), the external atmospheric pressure at the pressure relief port, and the equivalent two-phase flow density calibrated based on the insulating oil-gas-liquid mixing state are read. The difference between the upper limit pressure of the oil tank and the external atmospheric pressure is used as the pressure relief driving differential, and the theoretical discharge velocity of the oil tank under the upper limit pressure differential condition is calculated using the equivalent orifice flow relationship.

[0035] in, This represents the theoretical discharge velocity under the safe upper limit pressure differential condition of the fuel tank. This represents the upper limit of absolute pressure determined based on the yield strength of the fuel tank body. This indicates the atmospheric pressure outside the pressure relief port. This represents the equivalent density of the gas-liquid two-phase flow formed by insulating oil vapor and liquid insulating oil. This indicates the maximum safe pressure difference allowed by the fuel tank for pressure relief. Subsequently, the sound velocity and critical velocity coefficient of the mixed two-phase flow, pre-calibrated through a gas-liquid two-phase flow discharge test, are read. Multiplying these two values ​​determines the critical discharge velocity permissible before sound velocity congestion occurs at the pressure relief port.

[0036] in, This represents the critical discharge velocity corresponding to the sound velocity congestion boundary in a gas-liquid two-phase flow. This represents the critical velocity coefficient, which is less than 1. This represents the sound velocity of the gas-liquid two-phase flow mixture under the preset explosion-proof pressure relief condition. This represents the upper limit of the discharge velocity set to prevent the pressure relief port from entering a state of sonic congestion. Finally, for each sampling moment in the gasification expansion volumetric flow rate demand sequence, the theoretical discharge velocity is compared with the minimum value of the critical discharge velocity. When the theoretical discharge velocity does not exceed the critical discharge velocity, the theoretical discharge velocity is retained; when the theoretical discharge velocity exceeds the critical discharge velocity, it is truncated to the critical discharge velocity.

[0037] in, Indicates the first The constrained two-phase flow discharge velocity at each sampling time. This represents the operation of finding the minimum value. Indicates the sampling time sequence number; [and] each Arrange the timestamps of the gasification expansion volumetric flow rate demand sequence sequentially to obtain the restricted two-phase flow discharge velocity sequence, and ensure that the sampling time of this sequence corresponds one-to-one with that of the gasification expansion volumetric flow rate demand sequence.

[0038] In step S33, the flow cross-sectional area is inverted between the vaporization expansion volumetric flow rate demand sequence and the restricted two-phase flow discharge velocity sequence to obtain the time-series curve of the target equivalent discharge cross-section. It should be noted that since the vaporization expansion volumetric flow rate demand only characterizes the volume of fluid to be discharged per unit time, it cannot directly determine the flow area that the multi-stage pressure relief channels should form. Based on this, the technical solution of this application further inverts the flow cross-sectional area between the vaporization expansion volumetric flow rate demand sequence and the restricted two-phase flow discharge velocity sequence to obtain the time-series curve of the target equivalent discharge cross-section. Through the above processing, the transient discharge demand can be converted into a time-series discharge cross-sectional demand that satisfies the two-phase flow discharge velocity boundary.

[0039] More specifically, in a specific example of this application, the processing generally follows the order of timestamp registration, flow area inversion, and time series curve generation. First, based on a unified sampling clock, the vaporization expansion volumetric flow rate demand sequence and the restricted two-phase flow discharge velocity sequence are registered item by item according to the sampling timestamp, so that the vaporization expansion volumetric flow rate demand value at each sampling moment corresponds to the restricted two-phase flow discharge velocity value at the same moment. Then, based on the correspondence between volumetric flow rate, discharge velocity, and effective flow area during orifice discharge, the vaporization expansion volumetric flow rate demand value at the k-th sampling moment is divided by the restricted two-phase flow discharge velocity value at the same sampling moment. A pre-calibrated pressure relief channel comprehensive flow coefficient is used to correct for the flow reduction caused by orifice contraction, fluid viscosity, and local flow resistance, resulting in the target equivalent discharge cross-sectional area at the k-th sampling moment. The calculation formula is as follows:

[0040] in, Indicates the first The target equivalent discharge cross-sectional area at each sampling time This represents the first volumetric flow rate in the gasification expansion demand sequence. The volumetric flow rate requirement at each sampling time. This represents the discharge velocity value at the same sampling time in the discharge velocity sequence of a confined two-phase flow. This represents the comprehensive flow coefficient calibrated based on the orifice structure of the multi-stage pressure relief channel and the gas-liquid two-phase flow discharge test. This represents the effective volumetric flow rate that can be generated per unit flow cross-section under the current restricted discharge velocity. Indicates the sampling time sequence number. When When it equals 0, the corresponding Set it to 0. Finally, calculate the results at each sampling time. Arranged in chronological order and connected according to the corresponding timestamps, the time-series curve of the target equivalent discharge cross section is obtained. Each area value on the curve represents the equivalent flow area that needs to be formed at the corresponding sampling time to discharge the newly added vaporized expansion volume. This provides a target basis for subsequent discrete step approximation by combining the fixed aperture area of ​​each pressure relief channel and the mechanical inertial displacement lag time.

[0041] Specifically, in step S4, based on the fixed orifice area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels, a discrete stepwise approximation optimization is performed on the time series curve of the target equivalent discharge cross-section to obtain a multi-stage channel array combination encoding sequence. It should be noted that, given that the target equivalent discharge cross-section time series curve represents a continuously changing theoretical area requirement, while each stage of the pressure relief channel can only provide a fixed orifice area, and the valve's formation of an effective flow area from receiving the trigger command is limited by the mechanical inertial displacement lag time, it is impossible to directly and continuously adjust the discharge cross-section according to the target curve. Therefore, the technical solution of this application further performs a discrete stepwise approximation optimization on the time series curve of the target equivalent discharge cross-section based on the fixed orifice area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels to obtain a multi-stage channel array combination encoding sequence. Through the above processing, the continuous discharge section requirements can be converted into executable control data including channel combinations and trigger timing, so that the stepped discharge section formed by the superposition of multiple channels matches the timing curve of the target equivalent discharge section, thereby reducing the insufficient pressure relief capacity in the early stage of the arc fault and the excessive discharge capacity in the later stage.

[0042] Figure 4 This document describes a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels, based on an embodiment of this application. The method involves discretizing and optimizing the time-series curve of the target equivalent discharge cross-section using a stepped approximation based on the fixed aperture area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels to obtain a multi-stage channel array combined encoding sequence. (See flowchart for example.) Figure 4 As shown, step S4 includes: S41, based on the fixed aperture area of ​​each channel and the mechanical inertial displacement lag time, performing single-channel action area model derivation on each candidate trigger timescale combination corresponding to the target equivalent discharge section time series curve to obtain candidate multi-channel action area tensors; S42, fitting the channel summation result of the candidate multi-channel action area tensor to the target equivalent discharge section time series curve and performing residual cost quantization to obtain candidate combination residual mapping set; S43, performing global minimum optimization and discrete control protocol encoding on the candidate combination residual mapping set to obtain multi-level channel array combination encoding sequence.

[0043] In step S41, based on the fixed orifice area of ​​each channel and the mechanical inertial displacement lag time, a single-channel action area model is derived for each candidate trigger timescale combination corresponding to the target equivalent discharge section time series curve to obtain the candidate multi-channel action area tensor. It should be noted that since the orifice area of ​​each pressure relief channel is fixed, and the process of forming an effective flow area after triggering is affected by the mechanical inertial displacement lag time, the actual area response cannot be determined solely based on the candidate trigger timescales. Therefore, the technical solution of this application further derives a single-channel action area model for each candidate trigger timescale combination corresponding to the target equivalent discharge section time series curve based on the fixed orifice area of ​​each channel and the mechanical inertial displacement lag time to obtain the candidate multi-channel action area tensor. Through the above processing, the time series action area of ​​each channel under different trigger combinations can be obtained, providing a basis for the subsequent calculation of the fitting residual of the combined discharge section.

[0044] More specifically, in a specific example of this application, the processing generally follows the order of constructing candidate trigger timescale combinations, deducing single-channel area response, and arranging multi-dimensional action area. First, the fault start time, prediction termination time, and sampling period of the target equivalent discharge cross-section time series curve are read, and the sampling time is used as an optional trigger timescale. Simultaneously, the fixed orifice area of ​​each pressure relief channel and the mechanical inertial displacement lag time calibrated by valve action tests are read. For each pressure relief channel, one timescale is selected from the optional trigger timescales, and the selected timescales for each channel are arranged to form a candidate trigger timescale combination, which also includes candidate states where the corresponding channel remains closed within the prediction time range. Subsequently, for each candidate trigger timescale combination, the candidate trigger timescale, fixed orifice area, and mechanical inertial displacement lag time of each channel are substituted into the first-order inertial action area model to calculate the actual flow area formed by each channel at each sampling time.

[0045] in, Indicates candidate trigger time stamp combinations Next The first channel in the The area of ​​action at each sampling time Indicates the first The fixed aperture area of ​​each channel, Indicates the sampling period. Indicates candidate trigger time stamp combinations The Middle The trigger timestamps for each channel, Indicates the first The mechanical inertial displacement lag time of each channel, calibrated by the action test, is used as the time constant of the first-order inertial response. Represents the natural constant. The unit step function is represented; when the current sampling time is earlier than the candidate trigger time, the unit step function is 0, and the corresponding channel action area is 0. When the current sampling time reaches or is later than the candidate trigger time, the unit step function is 1, and the corresponding channel action area gradually approaches the fixed aperture area according to the mechanical inertial response. Finally, all calculation results are arranged according to three dimensions: candidate trigger time combination, pressure relief channel, and sampling time, so that each tensor element records the action area of ​​a specific channel under a specific candidate combination at a specific sampling time, thus obtaining the candidate multi-channel action area tensor.

[0046] In step S42, the channel summation result of the candidate multi-channel action area tensor is fitted with the target equivalent discharge cross-section time series curve to obtain a candidate combination residual cost quantization to obtain a candidate combination residual mapping set. It should be noted that since the candidate multi-channel action area tensor records the action area of ​​each channel separately, it cannot directly reflect the deviation between the overall discharge cross-section formed by the combination of candidate triggering time scales and the target requirement. Based on this, the technical solution of this application further performs residual cost quantization on the channel summation result of the candidate multi-channel action area tensor and the target equivalent discharge cross-section time series curve to obtain a candidate combination residual mapping set. Through the above processing, a unified cost can be used to characterize the degree of approximation of each candidate triggering time scale combination to the target discharge cross-section requirement.

[0047] More specifically, in a particular example of this application, the processing generally follows the order of channel action area summation, time-series area residual calculation, and candidate combination mapping. First, the candidate multi-channel action area tensor is read based on the candidate trigger time-stamp combination identifier and sampling timestamp. At each sampling time, the action areas of all pressure relief channels are summed to obtain the candidate equivalent discharge cross-section sequence corresponding to each candidate trigger time-stamp combination.

[0048] in, Indicates candidate trigger time stamp combinations In the The candidate equivalent discharge cross-sectional area formed at each sampling time. Indicates candidate trigger time stamp combinations Next The first pressure relief channel is in the... The area of ​​action at each sampling time This indicates the number of multi-stage pressure relief channels. Indicates the pressure relief channel number. This indicates the sampling time sequence number. Subsequently, according to a unified sampling timestamp, the candidate equivalent spillway cross-section sequence is registered item by item with the target equivalent spillway cross-section time series curve. The difference between the area of ​​the candidate equivalent spillway cross-section and the area of ​​the target equivalent spillway cross-section at the same sampling time is calculated. The squares of each area difference within the prediction time range are accumulated to obtain the fitting residual cost value of the corresponding candidate trigger time-stamp combination.

[0049] in, Indicates candidate trigger time stamp combinations The fitting residual algebraic value, Indicates candidate trigger time stamp combinations In the The candidate equivalent discharge cross-sectional area formed at each sampling time. The time series curve of the target equivalent discharge section is shown in the first... The target equivalent discharge cross-sectional area at each sampling time This represents the sequence number of the last sampling moment within the prediction time range. The squared term is used to uniformly quantify the deviation of the candidate equivalent spillway cross section from the target equivalent spillway cross section. Finally, a one-to-one correspondence is established between the combination identifier of each candidate trigger timescale combination and the corresponding fitted residual algebraic value, and these are collected according to the traversal order of the candidate trigger timescale combinations to obtain the candidate combination residual mapping set. This allows subsequent processing to select the candidate trigger timescale combination with the smallest deviation from the time series curve of the target equivalent spillway cross section based on the fitted residual algebraic value.

[0050] In step S43, the candidate combination residual mapping set is subjected to global minimum optimization and discrete control protocol encoding to obtain a multi-level channel array combination encoding sequence. It should be noted that since the candidate combination residual mapping set only represents the fitting error between each candidate triggering timescale combination and the target equivalent discharge cross-section timing curve, it cannot directly form control data that can be executed by the multi-level pressure relief channels. Based on this, the technical solution of this application further performs global minimum optimization and discrete control protocol encoding on the candidate combination residual mapping set to obtain a multi-level channel array combination encoding sequence. Through the above processing, the channel triggering scheme with the highest matching degree to the target equivalent discharge cross-section timing curve can be determined, and this scheme can be converted into resolvable discrete control data.

[0051] More specifically, in a particular example of this application, the processing generally follows the order of residual traversal, optimal combination determination, and control protocol encoding. First, the identifiers of each candidate trigger timescale combination in the candidate combination residual mapping set and their corresponding fitted residual algebraic values ​​are read sequentially. All fitted residual algebraic values ​​are numerically compared, and the candidate trigger timescale combination corresponding to the minimum fitted residual algebraic value is determined as the optimal trigger timescale combination. The calculation formula is as follows:

[0052] in, Indicates the optimal trigger time stamp combination identifier. This represents the operation of finding the independent variable that minimizes the objective function. Indicates the candidate trigger time stamp combination identifier. Represents a finite set of all candidate trigger timescale combinations. Indicates candidate trigger time stamp combinations The corresponding fitting residual algebraic value is then calculated. Subsequently, based on the optimal trigger time stamp combination identifier, the channel address, on / off status, and trigger time stamp relative to the arc fault initiation time of each pressure relief channel are extracted from the corresponding candidate trigger time stamp combinations, and arranged in order of channel address to form the optimal channel trigger record. Finally, according to the preset discrete control protocol, the protocol start identifier, number of channels, channel address, on / off status, and relative trigger time stamp are sequentially written into the control frame, and a verification field is generated based on the control frame content. The encapsulated control frames are then arranged in order of trigger time stamp to obtain a multi-level channel array combination encoding sequence for subsequent frame parsing and hardware timing triggering.

[0053] Specifically, in step S5, a hardware timer array is used for parallel countdown to perform frame parsing and delayed triggering execution of the multi-level channel array combination encoding sequence, thereby driving each physical pressure relief channel to open in a step-by-step manner according to the encoding sequence. It should be noted that, given that the multi-level channel array combination encoding sequence only records the channel address and relative trigger timestamp of each pressure relief channel, it cannot directly drive the physical pressure relief channel action, and the trigger timing of each channel needs to be consistent with the discrete step-by-step approximation optimization result. Based on this, the technical solution of this application further uses a hardware timer array for parallel countdown to perform frame parsing and delayed triggering execution of the multi-level channel array combination encoding sequence, thereby driving each physical pressure relief channel to open in a step-by-step manner according to the encoding sequence. Through the above processing, the channel combination and trigger timing can be converted into physical driving pulses for the corresponding pressure relief channel, enabling each channel to form a step-by-step equivalent discharge cross-section according to a determined time interval, thereby compensating for the lag in valve mechanical action and making the actual discharge capacity match the dynamic pressure relief requirements during arc faults.

[0054] Figure 5 This is a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multi-level pressure relief channels according to an embodiment of this application. The method involves parallel countdown using a hardware timer array to perform frame parsing and delayed triggering execution of the multi-level channel array combined encoding sequence, thereby driving each physical pressure relief channel to open in a step-by-step manner according to the encoding sequence. Figure 5As shown, step S5 includes: S51, by parsing the communication frame protocol, stripping the channel address and relative delay time stamp of the multi-level channel array combination encoding sequence, and superimposing the initial zero point time of the arc fault to obtain the absolute trigger timestamp set of each channel; S52, based on the high-frequency clock frequency of the programmable logic device, performing discrete clock down rounding mapping on the difference between the absolute trigger timestamp set of each channel and the current system time to obtain the hardware clock clock reload value; S53, by using the parallel countdown timer zeroing interrupt and solid-state power switch drive, performing square wave pulse synthesis output on the hardware clock clock reload value to drive each physical pressure relief channel to open in a stepwise manner according to the encoding sequence.

[0055] In step S51, the channel address and relative delay time stamp are separated from the multi-level channel array combined encoding sequence through communication frame protocol parsing, and the initial zero-point time of the arc fault is superimposed to obtain the absolute trigger timestamp set for each channel. It should be noted that since the multi-level channel array combined encoding sequence records the channel address and relative delay time stamp in the form of communication frames, it cannot be directly used as the trigger time basis for the hardware timer. Based on this, the technical solution of this application further separates the channel address and relative delay time stamp from the multi-level channel array combined encoding sequence through communication frame protocol parsing, and superimposes the initial zero-point time of the arc fault to obtain the absolute trigger timestamp set for each channel. Through the above processing, the relative timing encoding can be converted into the channel trigger time under a unified clock reference, providing a time basis for the parallel timing execution of each pressure relief channel.

[0056] More specifically, in a specific example of this application, the processing generally follows the order of communication frame verification, field parsing, and absolute time conversion. First, based on a preset communication frame protocol, the frame header, data length, payload, and check field of each communication frame in the multi-level channel array combined encoding sequence are identified. The integrity of the communication frame is determined through data length comparison and cyclic redundancy check. Then, the channel address code and relative delay code are extracted from the verified payload according to preset field offsets. Subsequently, based on the channel address mapping table, each channel address code is converted to the channel address of the corresponding physical pressure relief channel. Finally, the relative delay code is converted to a delay time relative to the initial zero-point of the arc fault according to the time scale resolution specified in the protocol. The initial zero-point of the arc fault is taken from the unified clock timestamp latched when the high-frequency fault current sampling signal of the transformer undergoes a transient change. This time is added to the relative delay time of each channel to obtain the absolute trigger timestamp of each channel. The calculation formula is as follows:

[0057] in, Indicates the first The absolute trigger timestamp of each physical pressure relief channel Indicates the initial zero-point time of the arc fault. This represents the first digit obtained from the parsing of the multi-level channel array combined coding sequence. The relative delay code value of each channel, This indicates the time-stamp resolution specified by the communication frame protocol. Indicates the first The delay time of each channel relative to the initial zero point of the arc fault. This represents the physical pressure relief channel number. Finally, each channel address is associated with its corresponding absolute trigger timestamp, and arranged in order of channel address to obtain a set of absolute trigger timestamps for each channel, so that each timestamp in the set can correspond to the physical pressure relief channel to be driven.

[0058] In step S52, based on the high-frequency clock frequency of the programmable logic device, the difference between the absolute trigger timestamp set of each channel and the current system time is discretized and rounded down to obtain the hardware clock tick reload value. It should be noted that since the absolute trigger timestamps of each channel are continuous time quantities, while the hardware timer can only load integer count values ​​in units of high-frequency clock cycles, the two cannot be directly correlated. Therefore, the technical solution of this application further uses the high-frequency clock frequency of the programmable logic device to discretize and round down the difference between the absolute trigger timestamp set of each channel and the current system time to obtain the hardware clock tick reload value. Through the above processing, the remaining waiting time of each channel can be converted into an integer tick that the hardware timer can execute, and the time quantization error is limited to one high-frequency clock cycle, providing a counting basis for the parallel countdown and stepped triggering of multi-level pressure relief channels. It should be noted that rounding down will ensure that the actual countdown length is no greater than the theoretical remaining time, meaning the trigger time is less than one clock cycle ahead of the theoretical time, thus not producing a delayed triggering effect.

[0059] More specifically, in a specific example of this application, the processing is performed in the order of remaining latency calculation, discrete clock tick mapping, and reload value arrangement. First, the set of absolute trigger timestamps for each channel is read, and the current system time of the programmable logic device under the same clock reference is obtained. The current system time is then subtracted from the absolute trigger timestamps of each channel to obtain the remaining latency for each physical pressure relief channel. Subsequently, each remaining latency is multiplied by the high-frequency clock frequency of the programmable logic device to convert it into a continuous clock tick count. This continuous clock tick count is then rounded down to obtain the integer tick count value that the hardware timer can load. The calculation formula is as follows:

[0060] in, Indicates the first Hardware clock tick reload value for each physical pressure relief channel. Indicates the first The absolute trigger timestamp of each physical pressure relief channel This represents the current system time when the discrete mapping is performed. This indicates the high-frequency clock frequency of the programmable logic device. Indicates the first The number of consecutive clock ticks corresponding to each physical pressure relief channel from the current system time to the absolute trigger timestamp. This indicates the floor function. This indicates the physical pressure relief channel number. Finally, match each channel address with its corresponding... The data is associated and arranged according to the channel address order to obtain the hardware clock tick reload value, which is then used by the subsequent parallel hardware timer to execute the countdown.

[0061] It should be noted that the above embodiments use a uniform down-rounding method to convert the continuous remaining waiting time of all physical pressure relief channels into an integer clock count value that can be loaded by the hardware timer. This can meet the basic execution requirements of parallel countdown, but this processing method assumes that the tolerance of each channel to time quantization error is the same. The timing curve of the target equivalent leakage cross section changes non-linearly with the arc energy release process. In the initial 5 to 10 milliseconds of the arc fault, the insulating oil vaporizes and expands rapidly, and the target equivalent leakage cross section is in a steep rise range. At this time, even a shift of less than one clock cycle in the channel triggering time will correspond to a large change in the target leakage cross section. In the arc energy decay stage, the timing curve of the target equivalent leakage cross section tends to be flat, and the change in leakage cross section corresponding to the same time shift is smaller. It can be seen that the triggering time of different channels implicitly contains a timing sensitivity related to the local change rate of the target equivalent leakage cross section timing curve. The uniform rounding method cannot allocate time quantization accuracy according to the pressure relief task undertaken by each channel. Furthermore, when the absolute trigger timestamp has arrived or is earlier than the current system time, the number of consecutive ticks will be non-positive; when the remaining waiting time exceeds the timer's counting range, the number of consecutive ticks will exceed the register's load limit, and direct loading will affect the effectiveness of trigger execution. Therefore, in a preferred embodiment, the channel timing sensitivity is further extracted based on the rate of change of the target equivalent discharge cross-section at the trigger time of each channel. The number of consecutive ticks is then differentially and discretely mapped according to the timing sensitivity, and the mapping results are zeroed out for non-positive values ​​and truncated by register bit width saturation, so that the hardware tick mapping takes into account both the pressure relief timing requirements and the timer's executable boundary.

[0062] Figure 6 This document describes a flowchart illustrating a control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels, based on an embodiment of this application. The method involves determining the channel timing sensitivity based on the target equivalent discharge cross-section timing curve, and performing heterogeneous adaptive discrete mapping and boundary verification on the absolute trigger timestamps of each channel to obtain the hardware clock tick reload value. Figure 6As shown, step S52 includes: S521, based on the target equivalent discharge cross-section timing curve, performing timing sensitivity quantification extraction on the absolute trigger timestamp set of each channel to obtain the timing sensitivity weight vector of each channel; S522, based on a preset sensitivity threshold, determining the high and low sensitivity channels of the timing sensitivity weight vector of each channel, and using a high-frequency clock frequency to perform heterogeneous adaptive rounding discrete mapping on the difference between the absolute trigger timestamp set of each channel and the current system time to obtain the original hardware clock tick count value; S523, based on the high and low determination results of each element in the timing sensitivity weight vector of each channel, performing negative overflow correction for high-sensitivity channels and bit width saturation truncation verification on the original hardware clock tick count value to obtain the hardware clock tick reload value.

[0063] In step S521, based on the target equivalent discharge cross-section time-series curve, the time-sensitivity quantification of the absolute trigger timestamp set of each channel is extracted to obtain the time-sensitivity weight vector of each channel. It should be noted that, given that the first embodiment uses a uniform clockwise discretization method for all pressure relief channels and does not differentiate the tolerance of different channels to trigger time quantization errors, and that the target equivalent discharge cross-section time-series curve changes rapidly in the early stage of arc eruption and tends to flatten out in the energy decay stage, the assigned activation times of each independent channel are not equivalent. Therefore, the technical solution of this application further extracts the time-sensitivity quantification of the absolute trigger timestamp set of each channel based on the target equivalent discharge cross-section time-series curve to obtain the time-sensitivity weight vector of each channel. Through the above processing, the degree of change in the discharge cross-section corresponding to the trigger time of each channel can be quantified, providing a basis for subsequent differentiated clockwise mapping.

[0064] More specifically, in a specific example of this application, the target equivalent spillway cross-section timing curve and the set of absolute trigger timestamps for each channel are first read. The absolute trigger timestamps for each channel are then converted to the fault time reference used in the target equivalent spillway cross-section timing curve. Based on the converted time coordinates, the corresponding position of each channel's trigger time in the target equivalent spillway cross-section timing curve is determined. The degree of local change at each channel's trigger time is extracted from the target equivalent spillway cross-section timing curve as a sensitivity criterion. For trigger times located inside the curve, the local first derivative is calculated using the center difference of adjacent sampling points. For trigger times located at the start or end of the curve, the local first derivative is calculated using one-sided difference. The absolute value of the first derivative is taken as the area demand change rate for the corresponding channel. Subsequently, the maximum value is selected from the area demand change rates of all channels as a normalization reference. The area demand change rate of each channel is divided by this maximum value to obtain the normalized timing sensitivity weight for each channel.

[0065] in, Indicates the first The rate of change of the target equivalent discharge cross section corresponding to each channel This represents the time series curve of the target equivalent discharge section at a relative time. The target equivalent discharge cross-sectional area, This represents the first derivative of the target equivalent discharge cross-sectional area with respect to time. Indicates the first The absolute trigger timestamp of each channel Indicates the initial zero-point time of the arc fault. Indicates the first The triggering time of each channel relative to the initial zero point of the arc fault. Indicates the first The timing sensitivity weight value for each channel ranges from 0 to 1. This indicates the total number of physical pressure relief channels. Indicates the current channel number. The value indicates the channel number participating in the maximum value comparison. The absolute value operation is used to characterize the local change of the target equivalent discharge cross section at the channel trigger time. The maximum value term represents the rate of change of the maximum target equivalent discharge cross section corresponding to all channels. When the rate of change of the target equivalent discharge cross section corresponding to all channels is 0, the time sensitivity weight value of each channel is set to 0 to avoid division by zero in the normalization calculation.

[0066] For example, four pressure relief channel trigger times are configured at 4 milliseconds, 7 milliseconds, 20 milliseconds, and 36 milliseconds after an internal arc fault occurs in the transformer. The corresponding area demand change rates, calculated using the local first derivative, are 0.80 square meters per second, 0.64 square meters per second, 0.16 square meters per second, and 0.04 square meters per second, respectively. Using 0.80 square meters per second as the normalization benchmark, the time sensitivity weight vectors for each channel are 1, 0.8, 0.2, and 0.05. When the weight of a channel approaches 1, it indicates that its trigger time is located in the steep rise range of the target curve at the initial stage of the arc outbreak, and this channel undertakes the urgent pressure relief task during the peak pressure phase. When the channel weight approaches 0, it indicates that its trigger time is located in the flat range after energy decay, with a higher tolerance for time quantization errors. This transforms the trigger timestamp, which originally only contained time information, into quantitative data that includes pressure relief safety sensitivity.

[0067] In step S522, the timing sensitivity weight vectors of each channel are used to determine high and low sensitivity channels based on a preset sensitivity threshold. Then, a heterogeneous adaptive rounding discrete mapping is performed on the difference between the absolute trigger timestamp set of each channel and the current system time using a high-frequency clock frequency to obtain the original hardware clock tick count value. It should be noted that, given that the timing sensitivity differences of each channel have been quantized in step S521, this difference needs to be further reflected in the tick discrete mapping process. Based on this, the technical solution of this application further determines high and low sensitivity channels based on a preset sensitivity threshold and performs a heterogeneous adaptive rounding discrete mapping on the difference between the absolute trigger timestamp set of each channel and the current system time using a high-frequency clock frequency to obtain the original hardware clock tick count value. Through the above processing, additional advance trigger margin can be configured for high-sensitivity channels while maintaining the original quantization accuracy of low-sensitivity channels.

[0068] More specifically, in a concrete example of this application, firstly, based on the channel address, the weight elements in the timing sensitivity weight vector of each channel are associated item by item with the absolute trigger timestamps in the absolute trigger timestamp set of each channel, and a preset sensitivity threshold is set. This threshold is calibrated based on tank pressure response tests and pressure relief channel action tests under different arc energy levels. Subsequently, each element in the timing sensitivity weight vector of each channel is compared and judged against the threshold. When the weight value of a channel is greater than or equal to the threshold, the channel is determined to be a high-sensitivity channel, indicating that the trigger time of the channel is located in the steep rising range of the timing curve of the target equivalent discharge section; when the weight value is less than the threshold, the channel is determined to be a low-sensitivity channel, indicating that the trigger time of the channel is located in the flat range of the timing curve of the target equivalent discharge section. Further, subtract the current system time from the absolute trigger timestamp of each channel to obtain the remaining waiting time for each channel. Multiply the remaining waiting time by the high-frequency clock frequency of the programmable logic device to obtain the number of consecutive clock ticks. For highly sensitive channels, round down the number of consecutive clock ticks and subtract one clock tick, causing the corresponding channel to trigger one clock cycle earlier than the uniformly rounded result. For low-sensitive channels, maintain the rounding method and map the number of consecutive clock ticks to the nearest integer not greater than this consecutive value, thereby generating the original hardware clock tick count value. The calculation formula is as follows:

[0069] in, Indicates the first The raw hardware clock tick count values ​​for each channel, Indicates the first The absolute trigger timestamp of each channel This indicates the current time when the discrete mapping is performed. This indicates the high-frequency clock frequency of the programmable logic device. Indicates the first The number of consecutive clock ticks corresponding to each channel. Indicates the first The timing sensitivity weight value of each channel, This indicates the preset sensitivity threshold. This indicates the floor function. This indicates the physical pressure relief channel number. The subtracted term for the high-sensitivity channel is an early trigger bias, used to add a safety margin of one high-frequency clock cycle to the quantization result formed by uniform down-rounding; the low-sensitivity channel retains the uniform down-rounding result to control unnecessary early pressure relief.

[0070] For example, if the preset sensitivity threshold is set to 0.6, the high-frequency clock frequency of the programmable logic device is 1 MHz, and the timing sensitivity weights of the four pressure relief channels are 1, 0.8, 0.2, and 0.05 respectively, then the first two channels are determined to be high-sensitivity channels, and the last two channels are determined to be low-sensitivity channels. When the remaining waiting time of the four channels relative to the current time corresponds to 4.6, 3004.6, 16004.6, and 32004.6 consecutive clock cycles respectively, the first two channels, after being rounded down and subtracted by 1, yield 3 and 3003 respectively, and the last two channels, after being rounded down, yield 16004 and 32004 respectively. Thus, the first two channels, located in the steep rise range of the target curve in the early stage of the arc fault, each have an increased advance trigger margin of 1 clock cycle compared to the first embodiment, while the last two channels, in the energy decay stage, maintain the original rounding method, making the pressure relief timing control correspond to the safety sensitivity of different channels.

[0071] In step S523, based on the high / low determination results of each element in the timing sensitivity weight vector of each channel, the original hardware clock tick count value is corrected for negative overflow of high-sensitivity channels and truncated for bit width saturation of low-sensitivity channels to obtain the hardware clock tick reload value. It should be noted that, given that heterogeneous adaptive rounding may cause high-sensitivity channels nearing the current time to generate negative count values, and may cause low-sensitivity channels with longer waiting times to generate count values ​​exceeding the timer register bit width, the unverified result cannot be directly loaded. Therefore, the technical solution of this application further corrects for negative overflow of high-sensitivity channels and truncates for bit width saturation of low-sensitivity channels based on the high / low determination results of each element in the timing sensitivity weight vector of each channel to obtain the hardware clock tick reload value. Through the above processing, it can be ensured that the generated original hardware clock tick count value is legal and valid within the hardware executable range.

[0072] More specifically, in a specific example of this application, firstly, based on the channel address, the high / low determination results in the timing sensitivity weight vector of each channel are associated with the corresponding original hardware clock tick count value item by item, and the maximum loadable count value is determined according to the bit width of the hardware timer register. Then, for high-sensitivity channels, it is checked whether the original hardware clock tick count value is less than 0; when the difference between the absolute trigger timestamp and the current time is less than one high-frequency clock cycle, the aforementioned early trigger bias may cause the count value to become negative. Therefore, for high-sensitivity channels, it is checked whether their count value has a negative overflow. If so, the count value of that channel is forcibly reset to zero, so that the corresponding hardware timer outputs a trigger signal on the next clock edge after loading. For low-sensitivity channels, the original hardware clock tick count value is compared with the maximum loadable count value of the timer. For low-sensitivity channels, it is checked whether their count value exceeds the upper limit of the maximum loadable bit width of the timer. If it does, saturation truncation is performed to the maximum value. The corresponding calculation formula is:

[0073] in, Indicates the first The hardware clock tick reload value is obtained after valid interval verification for each channel. Indicates the first The original hardware clock tick count values ​​are obtained after heterogeneous adaptive rounding discrete mapping of each channel. Indicates the effective bit width of the hardware timer register. This indicates the maximum loadable count value of the hardware timer register. This indicates that the original hardware clock tick count value, which is less than 0, will be reset to zero. This indicates that results exceeding the maximum loadable count will be saturated and truncated to... This ensures that the hardware clock tick reload value of each channel is between 0 and... Within the scope of hardware executable.

[0074] For example, with a preset sensitivity threshold of 0.6 and an effective bit width of 16 bits for the hardware timer register, the maximum loadable count value is 65535. When a high-sensitivity channel with a timing sensitivity weight of 0.8 obtains a raw hardware clock tick count of -1 after heterogeneous rounding, it is reset to zero, allowing the channel to trigger immediately on the next clock edge. When a low-sensitivity channel with a timing sensitivity weight of 0.2 obtains a raw hardware clock tick count of 70000, it is saturated and truncated to 65535. When the raw count values ​​of another high-sensitivity channel and a low-sensitivity channel are 3003 and 32004 respectively, both are within their respective boundary ranges, and their original count values ​​remain unchanged. Therefore, high-sensitivity channels near the pressure peak window will not miss actions due to negative count values, and low-sensitivity channels with longer waiting times will not trigger erroneously due to register overflow.

[0075] In step S53, the hardware clock tick reload value is synthesized into a square wave pulse and output through a parallel countdown timer reset interrupt and a solid-state power switch drive to drive each physical pressure relief channel to open in a stepped manner according to the coded sequence. It should be noted that since the hardware clock tick reload value only represents the number of waiting ticks for each physical pressure relief channel, it cannot directly drive the electromagnetic tripping mechanism of the pressure relief channel to complete the mechanical opening. Based on this, the technical solution of this application further uses a parallel countdown timer reset interrupt and a solid-state power switch drive to synthesize a square wave pulse and output the hardware clock tick reload value to drive each physical pressure relief channel to open in a stepped manner according to the coded sequence. Through the above processing, discrete count values ​​can be converted into power drive pulses for the corresponding channels, enabling multi-level pressure relief channels to form a stepped discharge cross-section according to the set trigger timing.

[0076] More specifically, in a specific example of this application, the processing generally follows the order of parallel loading of overload values, generation of zero-reset interrupts, and output of square wave power pulses. First, based on the channel address, the overload values ​​of each hardware clock cycle are loaded into the corresponding parallel countdown timers, and the current time latched during discrete clock cycle mapping is used as the common timing starting point, causing each countdown timer to begin counting down at the same high-frequency clock edge. When a hardware clock cycle overload value is 0, the corresponding countdown timer directly generates a zero-reset interrupt on the next high-frequency clock edge. When the hardware clock cycle overload value is greater than 0, the count value is decremented by 1 every high-frequency clock cycle until the count value reaches zero, thus obtaining the actual pulse start time of the corresponding physical pressure relief channel.

[0077] in, Indicates the first The pulse start time of each physical pressure relief channel This indicates the common timing start point for all parallel countdown timers. Indicates the first Hardware clock tick reload value for each physical pressure relief channel. This indicates the high-frequency clock frequency of the programmable logic device. This indicates the waiting time for the corresponding countdown timer to go from the start of counting to zero. This indicates the physical pressure relief channel number. Subsequently, when each countdown timer generates a zeroing interrupt, the corresponding output port is switched to a high level and returns to a low level after the preset pulse holding time expires, thus forming the square wave control signal required by the solid-state power switch. The solid-state power switch is turned on during the high-level duration, connecting the DC drive power supply to the electromagnetic tripping mechanism of the corresponding pressure relief channel, and disconnecting the drive circuit after the high level ends. The corresponding square wave power pulse is represented as:

[0078] in, Indicates the first Each physical pressure relief channel at time The obtained square wave driving voltage, This indicates the voltage value of the DC drive power supply. Represents the unit step function. Indicates the first The pulse start time of each physical pressure relief channel This refers to the pulse holding time, determined based on the energizing time required for the electromagnetic tripping mechanism to complete the release action. Used to switch the drive voltage to a high level at the start of the pulse. This is used to restore the driving voltage to a low level when the pulse holding time expires. The difference between the two unit step functions defines the effective duration of the square wave pulse. Finally, under the action of the corresponding square wave power pulse, each electromagnetic tripping mechanism releases the mechanical lock of the corresponding pressure relief channel, allowing the physical pressure relief channels with different fixed aperture areas to open in the order determined by the multi-level channel array combination coding sequence. The stepped equivalent discharge cross-section is formed by the superposition of the action areas of each channel in the time dimension.

[0079] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A control method for a transformer explosion-proof protection device integrating multi-stage pressure relief channels, characterized in that, include: S1, perform transient feature extraction on the received explosion-proof sensor raw data to obtain the arc energy injection rate characterization vector and the oil tank pressure time gradient sequence; S2, calculate the vaporization expansion volume of insulating oil by performing vaporization expansion volume calculation on the arc energy injection rate characterization vector and the oil tank pressure time gradient sequence. S3, based on the explosion-proof voltage reduction tracking constraint function, performs inverse fluid dynamics solution on the insulating oil vaporization expansion volume demand curve to obtain the target equivalent discharge section time series curve; S4. Based on the fixed aperture area and mechanical inertial displacement lag time of each of the multi-stage pressure relief channels, the time series curve of the target equivalent discharge section is subjected to discrete step approximation optimization to obtain the multi-stage channel array combination coding sequence. S5 uses a hardware timer array to count down in parallel, performing frame parsing and delayed triggering of the multi-level channel array combined encoding sequence to drive each physical pressure relief channel to open in a stepwise manner according to the encoding sequence.

2. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, The raw data from the explosion-proof sensor includes high-frequency fault current sampling signals from the transformer and dynamic pressure wave sampling signals from inside the oil tank.

3. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, Step S1 includes: The high-frequency fault current sampling signal of the transformer and the dynamic pressure wave sampling signal inside the oil tank are respectively subjected to power frequency component removal and vibration noise suppression to obtain transient high-frequency current sequence and denoised dynamic pressure sequence. The instantaneous rate of change of current is calculated for the amplitude at adjacent sampling times in the transient high-frequency current sequence to obtain the arc energy injection rate characterization vector; The rate of change of pressure values ​​at adjacent sampling times in the denoised dynamic pressure sequence is differentiated to obtain the time gradient sequence of tank pressure.

4. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, Step S2 includes: Based on discrete numerical integration within a sliding time window, the thermal energy accumulation of the arc energy injection rate characterization vector is weighted by thermal conversion coefficient to obtain the accumulated thermal energy equivalent sequence. Based on the tank pressure time gradient sequence, dynamic pressure coupling correction is performed on the basic latent heat of vaporization at atmospheric pressure to obtain a dynamic latent heat of vaporization sequence. Based on the gaseous specific volume constant, phase change mass and volume mapping is performed on the cumulative thermal energy equivalent sequence and the dynamic gasification latent heat sequence to obtain the gasification expansion volume demand curve of insulating oil.

5. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 4, characterized in that, Based on discrete numerical integration within a sliding time window, the thermal energy accumulation of the arc energy injection rate characterization vector is weighted by thermal conversion coefficients to obtain a cumulative thermal energy equivalent sequence, including: Based on the electric arc thermal power equivalent conversion coefficient, the electric arc energy injection rate characterization vector is weighted and discrete numerical integration is performed according to the sampling period in the sliding time window to obtain the cumulative thermal energy equivalent. The cumulative thermal energy equivalents at each sampling time are arranged in chronological order to obtain the cumulative thermal energy equivalent sequence.

6. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, Step S3 includes: The growth rate of the volume values ​​at adjacent moments in the volume demand curve for vaporization expansion of insulating oil is differentiated to obtain the vaporization expansion volume flow rate demand sequence. Based on the acoustic velocity congestion boundary of gas-liquid two-phase flow, the theoretical discharge velocity under the upper limit pressure difference condition of the oil tank is truncated by a minimum value constraint to obtain the discharge velocity sequence of the confined two-phase flow. The flow cross-sectional area is inverted from the gasification expansion volume flow rate demand sequence and the confined two-phase flow discharge velocity sequence to obtain the time series curve of the target equivalent discharge cross-section.

7. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, Step S4 includes: Based on the fixed aperture area of ​​each channel and the mechanical inertial displacement lag time, a single-channel action area model is derived for each candidate trigger time scale combination corresponding to the time series curve of the target equivalent discharge section to obtain the candidate multi-channel action area tensor. The channel summation results of the candidate multi-channel action area tensor are fitted with the time series curve of the target equivalent discharge section, and residual cost quantization is performed to obtain the candidate combined residual mapping set; Global minima optimization and discrete control protocol encoding are performed on the candidate combination residual mapping set to obtain a multi-level channel array combination encoding sequence.

8. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 7, characterized in that, The channel summation results of the candidate multi-channel action area tensor are fitted with the time series curve of the target equivalent discharge section to obtain the candidate combination residual mapping set by quantizing the residual cost. This includes: for each candidate trigger time-stamp combination, the multi-channel action area at each sampling time is summed, the difference between the channel summation result and the target equivalent discharge section area at the corresponding time is squared and accumulated, and the obtained fitting residual cost is correlated with the candidate trigger time-stamp combination to obtain the candidate combination residual mapping set.

9. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 1, characterized in that, Step S5 includes: By parsing the communication frame protocol, the channel address and relative delay timestamp are stripped from the multi-level channel array combined encoding sequence, and the initial zero point time of the arc fault is superimposed to obtain the absolute trigger timestamp set of each channel. Based on the high frequency of the programmable logic device, the difference between the absolute trigger timestamp set of each channel and the current system time is discretized and rounded down to obtain the hardware clock tick reload value. By using a parallel countdown timer to zero interrupt and a solid-state power switch to drive the hardware clock tick reload value, square wave pulses are synthesized and output to drive each physical pressure relief channel to open in a stepped manner according to the encoded sequence.

10. The control method for the transformer explosion-proof protection device integrating multi-stage pressure relief channels according to claim 9, characterized in that, Based on the high-frequency clock frequency of the programmable logic device, the difference between the absolute trigger timestamp set of each channel and the current system time is discretized and rounded down to obtain the hardware clock tick reload value, including: Based on the time series curve of the target equivalent discharge section, the time sensitivity quantification of the absolute trigger timestamp set of each channel is extracted to obtain the time sensitivity weight vector of each channel. Based on the preset sensitivity threshold, the timing sensitivity weight vector of each channel is used to determine the high and low sensitivity channels, and the difference between the absolute trigger timestamp set of each channel and the current system time is used to perform heterogeneous adaptive rounding discrete mapping to obtain the original hardware clock tick count value. Based on the high and low determination results of each element in the timing sensitivity weight vector of each channel, the original hardware clock tick count value is corrected by negative overflow and zeroing of high-sensitivity channels and bit width saturation truncation verification to obtain the hardware clock tick reload value.

Images