An active method for suppressing condensation in outdoor ring main units
By constructing augmented temperature and humidity fields, active condensation areas are identified and targeted interventions are carried out, solving the problem of misjudgment of condensation phenomena inside the ring main unit, improving the accuracy and robustness of condensation prevention, and reducing energy consumption.
Patent Information
- Application Number
- CN202611402115.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-10
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-10
AI Technical Summary
Existing technologies cannot accurately identify the spatial starting point of condensation inside outdoor ring main units, leading to frequent false starts or blind intervention of dehumidification equipment, and failing to effectively overcome the uneven temperature distribution and misjudgment of humidity distribution caused by thermal bridging effect.
By acquiring multi-point temperature and humidity data and external wall temperature data, and combining the thermal bridge structure properties, an augmented temperature field and humidity field are constructed to identify active condensation areas. Based on the phase change mass transfer intensity, reverse compensation is performed, and directional dehumidification control commands are output to achieve non-uniform intervention.
Accurately identifying condensation risks reduces energy loss caused by blind global heating, improving the sensing accuracy of humidity distribution inside the ring main unit and the robustness of condensation prevention.
Smart Images

Figure CN122902759B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power equipment operation safety protection technology, and in particular to an active method for suppressing condensation in the microenvironment of outdoor ring main units. Background Technology
[0002] Outdoor ring main units, as critical node equipment in power distribution networks, are exposed to variable weather conditions for extended periods. The interior of these units is highly susceptible to condensation due to drastic temperature changes, seriously threatening the insulation safety of internal high-voltage electrical components. Traditional anti-condensation technologies for controlling humidity within the unit primarily rely on installing heaters and dehumidifiers at the bottom of the unit, along with single or multiple temperature and humidity sensors for localized monitoring. In practical applications, the control system reads the sensor signals from these dispersed physical measurement points, compares them to preset fixed thresholds, and then drives the heating or dehumidifying equipment in a fully open or fully closed control mode to provide global energy heating or dehumidification intervention within the unit.
[0003] However, in threshold control based on locally distributed measurement points, the complex geometry inside the ring main unit and high thermal conductivity components such as metal fasteners and through-holes can cause significant thermal bridging effects, resulting in a large-gradient, non-uniform temperature distribution on the inner wall and key surfaces of the unit. Existing single-point or local sensing solutions cannot capture the morphology of these hidden cold spots within the hardware monitoring blind zone, making it difficult to accurately identify the spatial initiation location of condensation. Simultaneously, as moisture condenses at the solid-liquid interface, localized free water vapor is subjected to severe suction, resulting in mass loss and causing a unidirectional negative distortion in the measurement values of nearby physical sensors. Existing technologies directly use these distorted measurements when reconstructing the environmental state, leading to misjudgments of the overall humidity distribution trend and resulting in frequent false starts or inefficient, ineffective interventions by dehumidification equipment.
[0004] Therefore, accurately characterizing the distribution morphology of large-gradient multiphysics fields inside ring main units (RNBs) in complex and ever-changing outdoor micro-environments, overcoming the interference of local phase transition disturbances on environmental perception, and achieving highly robust spatiotemporal directional intervention are common technical problems that urgently need to be solved in the current power distribution operation and maintenance field. It is necessary to study an active suppression method that can improve the accuracy of micro-environment perception and the robustness of condensation prevention in RNBs. Summary of the Invention
[0005] The purpose of this invention is to provide an active method for suppressing condensation in outdoor ring main units, in order to solve the aforementioned problems in the prior art.
[0006] Technical solution: A method for actively suppressing condensation in outdoor ring main units, comprising: Acquire multi-point temperature and humidity observation data inside the cabinet, external wall temperature data, and external environmental temperature and humidity data, and extract pre-stored thermal bridge structure property parameters; Based on the external wall temperature data, thermal bridge structural property parameters, and local air temperature reflected by multi-point temperature and humidity observation data inside the cabinet, the transient temperature of the inner end face of each thermal bridge component is estimated, and the transient temperature of the inner end face of the thermal bridge is merged with the multi-point temperature and humidity observation data inside the cabinet to reconstruct the augmented temperature field. Based on the augmented temperature field and multi-point temperature and humidity observation data inside the cabinet, an initial humidity field and a saturated water vapor density field were constructed respectively, and the two were combined to identify active condensation areas. The spatial humidity deficit caused by phase change is calculated based on the phase change mass transfer intensity of the active condensation region. The spatial humidity deficit is used to perform reverse compensation on the temperature and humidity observation data of multiple points in the cabinet to obtain the desinking compensation observation data. The corrected humidity field is obtained by iterative reconstruction based on the desinking compensation observation data. The condensation extension vector field and condensation risk score are determined based on the modified humidity field and saturated water vapor density field, and the corresponding directional dehumidification control command is output.
[0007] Beneficial effects include the ability to deduce transient temperatures at cold points by constructing a two-port thermal network model based on structural geometric priors, and the introduction of a local anisotropic covariance kernel in the spatial field reconstruction. This helps to preserve the extreme values of local temperature pits within the microenvironment, allowing the granularity of spatial temperature perception to extend from the global macroscopic level to the microscopic morphology level of key metal components.
[0008] A steady-state molecular diffusion model was established to inversely derive the spatial cumulative humidity deficit field caused by condensation sinking and absorption. By performing reverse addition correction and iterative reconstruction on the original sensor values, it is possible to suppress global distribution misjudgments caused by low-humidity distortion readings and approximately restore the humidity distribution of the gas phase microenvironment before liquefaction disturbance.
[0009] By utilizing the spatial supersaturation field to solve for the outward normal gradient at the interface, the geometric propagation path of the water film phase transition front was quantified and predicted. Furthermore, by combining multi-dimensional state parameters such as spatial distance and phase transition peak value, directional thermal barrier coverage and ultimate insulation melting control were implemented using a three-dimensional control unit. This shifted the anti-condensation method from global uniform heating to spatially directional non-uniform intervention, reducing ineffective energy loss caused by blind global heating and helping to suppress the physical link that induces surface discharge by moisture propagation to insulating components. Attached Figure Description
[0010] Figure 1 This is a data processing flowchart of the present invention.
[0011] Figure 2 This is a flowchart illustrating the steps involved in estimating the transient temperature of the inner end face of each thermal bridge component according to the present invention.
[0012] Figure 3 The flowchart illustrates the steps involved in constructing the initial humidity field and the saturated water vapor density field for this invention.
[0013] Figure 4 This is a flowchart illustrating the steps of calculating the spatial humidity loss caused by phase change in this invention. Detailed Implementation
[0014] Example 1: This example provides an active method for suppressing condensation in outdoor ring main units. The method includes the following steps: like Figure 1 As shown, in step 101, multi-point temperature and humidity observation data inside the cabinet, external wall temperature data, and external environmental temperature and humidity data are obtained, and pre-stored thermal bridge structure property parameters are extracted.
[0015] Furthermore, the acquisition of microenvironmental physical state data is achieved by deploying a distributed sensor array inside the outdoor ring main unit and setting up detection probes on the outer wall and in the external environment. The multi-point temperature and humidity observation data inside the unit consists of a set of temperature and relative humidity observation sequences that change with the sampling time. The outer wall temperature data and the external environmental temperature and humidity data reflect the macroscopic thermodynamic state at the ring main unit boundary. The pre-stored thermal bridge structural property parameters are extracted from a pre-built offline digital twin model of the unit and stored in the controller's memory. Specifically, these parameters include multi-dimensional feature matrices such as the spatial three-dimensional coordinates, cross-sectional area, length, thermal conductivity, density, and specific heat capacity of each metal fastener and through-hole component. Acquiring the above multi-source sensing data and structural prior information provides boundary driving conditions and basic physical model support for subsequent transient thermal deduction.
[0016] In some alternative implementations, conventional data cleaning and denoising processes can be performed on the acquired temperature and humidity observation sequences. For example, abnormal pulse interference points that exceed the sensor's measurement range can be removed by using median filtering or moving average algorithms to ensure the stability of the input data link.
[0017] Step 102: Based on the external wall temperature data, thermal bridge structural property parameters, and local air temperature reflected by multi-point temperature and humidity observation data inside the cabinet, estimate the transient temperature of the inner end face of each thermal bridge component, and merge and reconstruct the augmented temperature field by combining the transient temperature of the inner end face of the thermal bridge with the multi-point temperature and humidity observation data inside the cabinet.
[0018] To address the issue of hidden cold spots being missed due to existing technologies treating the interior of the cabinet as a uniform temperature node, a structural feature model is introduced for cold spot deduction. An equivalent thermal network is constructed based on the thermal bridge structural property parameters. The outer wall temperature data is used as the external environmental conduction boundary, and the local air temperature as the internal air convection boundary, to deduce the transient temperatures of the inner end faces of each thermal bridge, which were originally within the blind zone of the sensor hardware. These deduced transient temperatures of each thermal bridge inner end face are then treated as measurements from virtual sensors and incorporated into the original set of real spatial coordinate observations contained in the multi-point temperature and humidity observation data within the cabinet. Spatial interpolation is performed based on the merged three-dimensional spatial coordinate set and the corresponding temperature values to generate an augmented temperature field. This augmented temperature field maintains a continuous shape globally within the cabinet while preserving local low-temperature extreme values at the physical coordinate locations corresponding to the thermal bridge components, thus avoiding the defect of traditional smooth interpolation algorithms that mask the characteristics of small, high-risk cold spots.
[0019] Step 103: Based on the multi-point temperature and humidity observation data inside the cabinet, construct an initial humidity field, and construct a saturated water vapor density field based on the augmented temperature field.
[0020] After completing the spatial reconstruction of the temperature field, a conventional spatial grid interpolation algorithm was used to globally extend the humidity variables in the multi-point temperature and humidity observation data inside the cabinet, obtaining the initial humidity field under the assumption of a gaseous phase without phase change liquefaction disturbance. Accordingly, the temperature values corresponding to each spatial grid node in the augmented temperature field were directly converted into the critical values of water vapor density under saturation state using the physical state mapping equation, forming a saturated water vapor density field containing spatial coordinate elements. The initial humidity field and the saturated water vapor density field were compared node by node. When the initial humidity field value of a certain surface grid node is greater than or equal to the corresponding saturated water vapor density field value, it is determined that a water vapor condensation phase change has occurred at that physical location, and the corresponding node is marked and included in the set of active condensation regions.
[0021] Step 104: Calculate the phase change mass transfer intensity based on the active condensation area and the difference between the initial humidity field and the saturated water vapor density field. Calculate the spatial humidity deficit caused by the phase change based on the phase change mass transfer intensity. Use the spatial humidity deficit to perform reverse compensation on the multi-point temperature and humidity observation data inside the cabinet to obtain desinking compensation observation data. Iteratively reconstruct the corrected humidity field based on the desinking compensation observation data.
[0022] When condensation occurs, the condensation surface continuously absorbs moisture from the adjacent air medium, forming a physical sink in the local space that continuously consumes water vapor. Specifically, the phase change mass transfer intensity of this water vapor sink is quantified using a local convection mass transfer model, and the influence domain of water vapor density decaying outwards in three-dimensional space is solved based on steady-state molecular diffusion laws. Then, the influence of multiple micro-sinks within the entire domain is spatially superimposed to obtain a global spatial humidity deficit. This spatial humidity deficit causes the sensor's measured values to be lower than the true macroscopic environmental humidity when condensation does not occur. The original measured values of temperature and humidity observations at multiple points inside the cabinet are added back to the absolute perturbation of the spatial humidity deficit at the corresponding sensor physical coordinates, thereby reconstructing the sink compensation observation data. Subsequently, using the sink compensation observation data as input, spatial interpolation and deficit inversion calculations are performed again, iteratively evaluating the iterative residual of the phase change mass transfer intensity until the continuous difference is less than the set iterative convergence condition. The output corrected humidity field suppresses the sensing distortion caused by condensation water absorption and approximately restores the overall picture of the high-gradient microenvironment humidity distribution.
[0023] Step 105: Determine the condensation propagation vector field and condensation risk score based on the corrected humidity field and saturated water vapor density field, and output the corresponding directional dehumidification control instructions, including: To achieve targeted intervention in the anti-condensation strategy, spatial supersaturation distribution characteristic parameters of the corrected humidity field and saturated water vapor density field are extracted. The maximum outward normal gradient is calculated at the interface between the active condensation area and the non-condensation area, and the results are summarized to form a condensation expansion vector field indicating the future direction of phase change water film propagation. A multi-dimensional weighted evaluation calculation is performed, combining the steady-state condensation area physical scale, the extreme value of phase change mass transfer intensity, and the Euclidean distance from the propagation direction to the high-voltage critical insulation hardware inside the cabinet, to obtain a condensation risk score representing the current equipment crisis level. According to pre-set risk control rules, the condensation risk score is mapped to a directional dehumidification control command matching a specific physical spatial coordinate region and the corresponding minimum heating power value. This command is sent to the underlying execution hardware mechanism, enabling non-uniform spatial directional dehumidification in high-risk condensation areas, replacing the high-energy-consuming mode of traditional global blind heating control.
[0024] In some optional implementations, if the system logic determines that the danger level corresponding to the current condensation risk score exceeds the preset threshold and approaches the critical distance for insulation safety, the system will generate an abnormal fuse signal and report it to the superior power grid dispatch center.
[0025] Example 2: Based on Example 1 above, this example further details the acquisition and conversion of multi-source sensing data, as well as the recursive evolution of transient temperature on the inner end face of the component based on the prior of the digital twin thermal bridge structure.
[0026] In one possible implementation, the following steps are included: Step 201: Calculate the saturated water vapor pressure at the corresponding temperature using empirical formulas, and combine the ideal gas law to convert the relative humidity sequence in the multi-point temperature and humidity observation data inside the cabinet and the external environment temperature and humidity data into the corresponding water vapor density observation values. Use the water vapor density observation values as the basis for constructing the initial humidity field.
[0027] As a non-core preprocessing step, to ensure a unified mass concentration dimension for subsequent mass-thermal coupling spatial reconstruction, the relative humidity percentage signal collected in real-time by the sensing devices needs to be converted into an absolute humidity physical quantity. Specifically, this is achieved by loading a gas physics parameter calculation module pre-installed in the controller to perform formula conversion on the relative humidity sequence collected by the distributed temperature and humidity sensor array inside the cabinet and the external relative humidity sequence collected by the external environment probe. During the calculation, the Tetens empirical formula is invoked, using the temperature observation value collected at the current step size as the input variable to map the saturated water vapor pressure under the corresponding temperature conditions. The Tetens empirical formula is a well-known empirical formula in meteorology and thermodynamics, and those skilled in the art can directly consult relevant engineering manuals to obtain its standard expression. The saturated water vapor pressure, the measured relative humidity observation value, the gas constant, and the thermodynamic temperature are substituted into the ideal gas law for multiplication and division operations to eliminate the influence of pressure and temperature fluctuations, outputting a sequence of water vapor density observation values in standard units. This sequence serves as the basic data source for humidity field reconstruction.
[0028] like Figure 2 As shown, in step 202, each thermal bridge component is equivalent to a one-dimensional transient heat conduction unit.
[0029] To quantify the dynamic impact of thermal bridging on the microenvironment temperature distribution, a heat transfer topological abstraction is needed for the mechanical fasteners, wall bushings, and penetrating metal components on the ring main unit shell. In constructing the physical model, based on the geometric characteristic that the axial length of each thermal bridge component is much larger than its radial dimension, lateral heat conduction in both the circumferential and radial directions is ignored. Each independent metal component is abstracted as a one-dimensional transient heat-conducting unit that experiences energy flow along its axial direction. This one-dimensional transient heat-conducting unit penetrates the steel shell of the ring main unit, with one end exposed to the external air or in contact with the outer wall, and the other end extending into the internal microspace of the cabinet, forming discrete heat transfer channels.
[0030] Step 203: Based on the geometric dimensions and material properties in the thermal bridge structure properties parameters, calculate the lumped thermal resistance and lumped heat capacity of each thermal bridge component to obtain the thermal parameter set of the thermal bridge component.
[0031] During the offline construction or initial configuration phase, the controller loads the thermal bridge structure parameter table. For the j-th thermal bridge component registered in the parameter table, its spatial axial length, cross-sectional area, and thermal conductivity from the material properties are extracted. The lumped thermal resistance of the component is calculated by dividing the product of thermal conductivity and cross-sectional area as the denominator and axial length as the numerator. Correspondingly, the density, specific heat capacity, cross-sectional area, and axial length of the component are extracted, and these four physical quantities are multiplied together to calculate the lumped heat capacity of the component. The calculated lumped thermal resistance is multiplied by the lumped heat capacity, and the physical product is extracted as the thermal time constant of the component. All components in the thermal bridge structure parameter table are traversed, and a multidimensional matrix containing various thermal resistances, heat capacities, and thermal time constants is output as the thermal parameter set of the thermal bridge component.
[0032] Step 204: Calculate the transient temperature of the inner end face of the thermal bridge based on the thermal parameter set of the thermal bridge component, the outer wall temperature data, and the local air temperature reflected by the multi-point temperature and humidity observation data inside the cabinet.
[0033] After obtaining the equivalent thermal control parameters of each component, it is necessary to track the temperature fluctuations of its inner end face exposed to the air inside the cabinet in real time. This step solves the temperature dynamic response of the one-dimensional transient heat conduction unit by introducing a bidirectional boundary driving mechanism, which overcomes the time lag bias of the traditional method that assumes the inner wall temperature is passively equivalent to the outer wall temperature, and provides a basis for accurately capturing the dew point triggering time.
[0034] Step 205: The outer wall temperature data is used as the outer end conduction boundary condition of the one-dimensional transient heat conduction unit, and the convective coupling between the inner end face of the thermal bridge component and the local air temperature reflected by the multi-point temperature and humidity observation data in the cabinet is used as the second boundary condition. Combined with the lumped thermal resistance and lumped heat capacity of the thermal parameters of the thermal bridge component, the dynamic equation of the two-port thermal network is constructed.
[0035] Specifically, the one-dimensional transient heat-conducting unit has two control ports in its spatial topology. Its outer end face is in close contact with the outer shell of the cabinet, directly receiving the outer wall temperature data as a type of Dirichlet temperature boundary condition. Its inner end face exchanges heat with the local air inside the cabinet. Here, by applying a natural convection correlation, the fluid convection heat transfer coefficient of the inner end face is calculated based on the component's protrusion height and installation orientation parameters. The calculation of the fluid convection heat transfer coefficient of the inner end face adopts a well-known empirical correlation for natural convection heat transfer. Those skilled in the art can select the corresponding Nusselt number correlation (such as the Churchill-Chu correlation) according to the geometry of the thermal bridge component (such as a vertical plate, a horizontal cylinder, etc.) and its installation orientation, and calculate the Grashof number and Prandtl number by combining the current local temperature difference, and then solve for the Nusselt number and convert it to obtain the convection heat transfer coefficient. The above calculation process is a conventional technique in engineering heat transfer. At the same time, the readings of several temperature sensors closest to the spatial coordinates of the thermal bridge component are searched and weighted by distance to calculate the corresponding local air temperature. Using the convective heat transfer coefficient of the inner end face, the exposed area of the inner end face, and the difference between the local air temperature and the current temperature of the inner end face as inputs, a second boundary condition characterizing the fluid-mass coupling of convective heat transfer at the inner end face is constructed. It is important to note that the second boundary condition refers to either the boundary condition for convective heat transfer at the inner end face or the convection coupling boundary condition at the other end, and is not a Neumann boundary condition. Combining the energy conservation relationship between the control heat flux at both ends and the total lumped heat capacity of the components, a first-order dynamic equation for the two-port thermal network is constructed, expressed as follows: C _j *(dT _bj / dt)=(T _ext -T _bj ) / R _j -h _cj *A _endj *(T _bj -T _localj ); Among them, C _j Let dT be the lumped heat capacity of the j-th thermal bridge component. _bj / dt is the differential term of the temperature at the inner end face of the thermal bridge component with respect to time, T _ext To collect real-time external wall temperature data, T _bj R is the current transient temperature of the inner end face of the thermal bridge component. _j h is the lumped thermal resistance of the thermal bridge component. _cj A is the convective heat transfer coefficient of the fluid at the inner end face. _endj T represents the exposed area of the inner end face. _localj This represents the local air temperature in the vicinity of the thermal bridge component.
[0036] Step 206: The dynamic equation of the two-port thermal network is recursively updated within the set sampling step size using a discretization format, and the transient temperature of the inner end face of the thermal bridge is output.
[0037] To solve the above differential equations in real time within the digital controller, they need to be transformed into an algebraic recursive form. In the actual execution process, a first-order implicit Euler scheme is used to discretize the dynamic equations of the two-port thermal network in time. The outer wall temperature data and the local air temperature at the current sampling time t are substituted into the discrete equations as known quantities. Through algebraic rearrangement and coefficient merging, the equations are then processed at each sampling step δ. _t The solution is updated internally and the transient temperature sequence of all thermal bridge inner end faces at the current moment is calculated and output.
[0038] In some alternative implementations, as an alternative to the first-order implicit Euler discretization scheme, when the controller hardware is equipped with a high-performance floating-point arithmetic unit, a fourth-order Runge-Kutta scheme can also be used to discretize and recursively solve the dynamic equations of the two-port thermal network in order to further capture the higher-order thermal flux disturbances during drastic temperature changes on the outer wall.
[0039] Step 207: Calculate the thermal time constant corresponding to each thermal bridge component based on the lumped thermal resistance and lumped thermal capacity of the thermal parameters of the thermal bridge components.
[0040] After calculating the temperatures of each cold spot, the product of the thermal resistance and heat capacity calculated in step 203 is further extracted to determine the unique thermal time constant of each component. This parameter characterizes the temperature response speed of the metal component to the cooling impact of the external environment.
[0041] Step 208: Calculate the current ambient dew point temperature based on multi-point temperature and humidity observation data inside the cabinet; compare the rate at which the transient temperature of each thermal bridge inner end face reaches the current ambient dew point temperature, and establish a condensation timing prediction sequence for all thermal bridge components in combination with the thermal time constant.
[0042] The global ambient dew point temperature is calculated using multi-point temperature and humidity observation data collected at the current step size within the cabinet. By evaluating the critical temperature difference between the transient temperature of each thermal bridge's inner end face and the ambient dew point temperature at the current moment, and combining this with the thermal time constant of each component, the rate at which each component's temperature approaches the dew point temperature is calculated. These rates are then sorted in descending order to generate a one-dimensional structured index linked list, serving as the condensation timing prediction sequence. Components with higher rates are ranked higher, indicating that they reach the condensation trigger condition earlier.
[0043] Step 209: The positions of the thermal bridge components ranked first in the condensation time-series prediction sequence are taken as high-frequency priority monitoring nodes, and higher interpolation weights are assigned to the virtual sensors corresponding to the high-frequency priority monitoring nodes in the subsequent augmented temperature field reconstruction.
[0044] A specific number or a predetermined proportion of component numbers at the top of the condensation time-series prediction sequence are extracted, and their spatial coordinates are defined as high-frequency priority monitoring nodes. In subsequent control loops, the controller will increase the sampling frequency of sensors in the neighborhood of these nodes, or assign higher weight coefficients to these virtual measuring points in the spatial field interpolation algorithm.
[0045] As an example, assuming that after normalization, the lumped thermal resistance of a stainless steel through-fit is 2.0 and the lumped heat capacity is 0.5, then its thermal time constant is 2.0 * 0.5 = 1.0. If another aluminum fastener has a lumped thermal resistance of 0.5 and a lumped heat capacity of 0.4, then its thermal time constant is 0.5 * 0.4 = 0.2. When the external temperature drops sharply, the controller calculates through the dynamic equations of the two-port thermal network that the aluminum fastener with a thermal time constant of 0.2 approaches the dew point faster than the stainless steel through-fit, and therefore places it first in the condensation timing prediction sequence as a priority monitoring position.
[0046] Example 3: Based on Examples 1 and 2 above, this example further details the steps of constructing an augmented temperature observation set, spatial gridding, and introducing a piecewise anisotropic covariance kernel for spatial temperature field reconstruction.
[0047] In one possible implementation, the following steps are included: Step 301: The transient temperature of the inner end face of the thermal bridge is used as the measurement value of the virtual sensor and merged with the temperature observation sequence in the multi-point temperature and humidity observation data in the cabinet to form an augmented temperature observation set.
[0048] Specifically, traditional spatial interpolation relies solely on physical temperature and humidity sensors located in open areas within the cabinet, failing to directly detect local minima caused by structural thermal bridges. To make the hidden cold spot characteristics derived in the above embodiments explicit, the transient temperature of the inner end face of each thermal bridge is defined as the virtual sensor reading located at the three-dimensional spatial coordinates of the corresponding thermal bridge component's inner end face. The spatial geometric coordinates of all physical temperature and humidity sensors and their corresponding temperature observation sequences are extracted, along with the spatial geometric coordinates of the inner end faces of all thermal bridge components and their corresponding transient temperatures. The two sets of spatial geometric coordinates are then combined to form an expanded set of spatial coordinate elements. Correspondingly, the temperature observation values from the physical sensors and the temperature values from the virtual sensors are concatenated into an array to output an augmented temperature observation set.
[0049] Step 302: Based on the pre-stored cabinet geometric boundary parameters, discretize the three-dimensional space inside the cabinet into a temperature field reconstruction mesh.
[0050] To establish a digital mapping of continuous multiphysics fields in a discrete digital system, the geometric space inside the ring main unit needs to be meshed. Based on the boundary parameters of the pre-stored digital twin model of the unit, the spatial discrete step size is set. Using a finite difference meshing rule, the three-dimensional space inside the ring main unit is uniformly divided along the horizontal, vertical, and depth directions, generating a temperature field reconstruction mesh composed of a discrete spatial node matrix. Each node in the mesh has a unique spatial physical coordinate index, represented as three-dimensional spatial coordinates (x, y, z).
[0051] Step 303: Based on the augmented temperature observation set, the spatial field is reconstructed on the temperature field reconstruction grid to obtain the augmented temperature field.
[0052] After obtaining the data base containing virtual cold spots and the three-dimensional discrete mesh, it is necessary to perform temperature numerical inversion on each node of the mesh that is not equipped with sensors. This step introduces a modified multidimensional spatial interpolation algorithm to deduce the transient temperature of unknown mesh nodes based on the temperature distribution law of known measuring points, and finally outputs an augmented temperature field covering the entire domain.
[0053] Step 304: In the process of spatial field reconstruction, a piecewise covariance structure is introduced, and a local anisotropic covariance kernel is defined with the position of each thermal bridge component as the center within a radius of 2 to 5 times its characteristic size.
[0054] Standard Kriging interpolation or radial basis function interpolation algorithms tend to maintain an overall smooth interpolation surface when reconstructing the global field, which can easily lead to the smoothing out of the abrupt low-temperature characteristics of thermal bridge cold spots. To address interpolation distortion, a piecewise covariance structure is introduced into the spatial reconstruction algorithm. Specifically, a spatial distance determination module is loaded through the controller to traverse the grid nodes in the temperature field reconstruction mesh one by one. Taking the three-dimensional spatial coordinates of the inner end face of each thermal bridge component as the center, the Euclidean spatial distance from the grid node to the center is calculated. When the Euclidean spatial distance is less than or equal to a set radius threshold, the algorithm determines that the grid node has entered a locally specific reconstruction region. The radius threshold is preferably set to 3 times the feature size of the thermal bridge component. Within the locally specific reconstruction region, the globally uniform isotropic correlation function is discontinued, and instead, a dedicated local anisotropic covariance kernel is loaded.
[0055] Among them, the characteristic dimension of the thermal bridge component is the equivalent diameter of the component's cross section; for circular cross section components, the characteristic dimension is the cross section diameter; for non-circular cross section components, the characteristic dimension is the diameter of the largest circumscribed circle of the cross section.
[0056] Step 305: Adjust the correlation length of the local anisotropic covariance kernel in the axial direction of the thermal bridge component to make it shorter than the correlation length of the conventional region. Generally, the region outside the coverage of the local anisotropic covariance kernel is considered the conventional region.
[0057] In the local anisotropic covariance kernel, the correlation length parameter determines the smooth range of the propagation of measurement data into the surrounding space. Based on the physical characteristic of high-gradient temperature conduction along the axial direction of the thermal bridge component, the spatial correlation decay rate is artificially intervened. When constructing the covariance matrix, the spatial correlation length along the axial direction of the thermal bridge component is shortened and limited, making its value smaller than the conventional correlation length in the open conventional region. By shortening the axial correlation length, the low-temperature energy conducted from the outer end of the thermal bridge propagates only to its narrow axial neighborhood on the mathematically reconstructed surface, forming a large decreasing temperature gradient in space, thus leaving the local minimum pit morphology of the corresponding physical thermal bridge on the temperature field reconstruction grid.
[0058] Step 306: Based on the local anisotropic covariance kernel and the augmented temperature observation set, calculate the conditional expectation of each node on the temperature field reconstruction grid to generate an augmented temperature field that retains local extremum features.
[0059] By combining the global isotropic covariance matrix of the conventional region and the local anisotropic covariance matrix of the locally specific reconstructed region, a complete global piecewise covariance matrix is generated. Using the augmented temperature observation set as the boundary condition input, the expected value of the conditions at each spatial node on the temperature field reconstruction grid is calculated by solving the linear Kriging equations. The final output augmented temperature field exhibits a continuous and smooth spatial morphology in the open region, while at the coordinate positions corresponding to each metal fastener and through-hole component, the concave features reflecting the cold point temperature of the solid thermal bridge are explicitly preserved, providing accurate local temperature input for subsequent phase transition criteria.
[0060] In some alternative implementations, as an alternative to the piecewise anisotropic Kriging interpolation algorithm, when the internal physical structure of the ring main unit is complex and the number of thermal bridges is large, resulting in excessive computational complexity for large matrix inversion, a block-based anisotropic radial basis function interpolation method can also be used to ensure the real-time response speed of the controller. This alternative achieves the same effect of preserving local temperature dips by introducing an asymmetric Gaussian function that decays monotonically and rapidly along the axial direction within the radius around the thermal bridge coordinates.
[0061] For example, assuming that after normalization, the mesh node is located within a locally specific reconstruction region, and its axial spatial coordinate distance is 0.1. If the correlation function of the conventional region is used, and its covariance correlation length is set to 0.5, the calculated spatial correlation is 1.0 - (0.1 / 0.5) = 0.8. However, in this scheme, the axial correlation length of the local anisotropic covariance kernel is shortened and limited to 0.12, so the spatial correlation in this direction is adjusted and calculated to 1.0 - (0.1 / 0.12) = 0.167. The spatial correlation decreases from 0.8 to 0.167, indicating that the low-temperature characteristics decay rapidly in this direction, ensuring that the reconstructed augmented temperature field can highlight the low-temperature gradient of the local thermal bridge cold point.
[0062] Example 4: Based on Examples 1 and 3 above, this example, as a field reconstruction mechanism to eliminate local sucker effect, further details the sink removal iterative reconstruction process based on steady-state phase transition sink analysis.
[0063] In one possible implementation, the following steps are included: like Figure 3 As shown, in step 401, an initial humidity field is generated based on multi-point temperature and humidity observation data inside the cabinet using a spatial interpolation method.
[0064] To establish a comprehensive humidity distribution data base covering the entire cabinet area, the water vapor density observation sequence obtained from the preprocessing stage is extracted, and the interpolation module is invoked to perform numerical spatial expansion on the temperature field reconstruction grid. During this process, the algorithm model pre-assumes that the environment is in a pure gas-phase mixture state without local liquefaction phase transition disturbances. In the specific calculations, radial basis functions are used to perform global continuous processing on the discrete measurement point data to generate the initial humidity field. In some optional implementations, as an alternative to radial basis functions, when the spatial density of physical sensors within the cabinet exceeds the pre-set reciprocal of the sensor physical spacing, an inverse distance weighted algorithm is used to perform nearest-neighbor interpolation on the data points to generate the initial humidity field, in order to reduce the memory stack usage of the controller.
[0065] Step 402: Using the augmented temperature field, calculate the saturated water vapor density corresponding to each spatial grid node to obtain the saturated water vapor density field.
[0066] The augmented temperature field matrix output from the preceding computational link is extracted, and the expected temperature conditions stored within each spatial coordinate node are read. These extracted expected temperature conditions are input into a gas state mapping table constructed based on the thermodynamic phase equilibrium principle. The saturated water vapor density boundary values for each physical coordinate node under gas-liquid critical equilibrium are output via table lookup or function mapping. The boundary values of all nodes within the grid space are combined and concatenated to construct the saturated water vapor density field. This data field provides a reference surface for subsequent determination of whether water is in a supersaturated free state in multidimensional space.
[0067] like Figure 4 As shown, in step 403, the initial humidity field and the saturated water vapor density field are compared node by node, and the surface nodes that meet the condition that the water vapor density is greater than or equal to the saturated water vapor density are marked as active condensation areas; where surface nodes refer to spatial grid nodes located on the solid surface inside the cabinet.
[0068] The local mass transfer coefficient is determined based on the local temperature difference and fluid properties of each node in the active condensation region. By combining the local mass transfer coefficient and the difference between the initial humidity field and the saturated water vapor density field, the instantaneous condensation mass flux of each node in the active condensation region is calculated and used as the phase change mass transfer intensity.
[0069] The controller loads the matrix condition operation module, performing a scalar subtraction operation between the initial humidity field matrix and the saturated water vapor density field matrix corresponding to the same mesh topology. Based on the microenvironment mass transfer thermodynamic criterion, the coordinate information of spatial mesh nodes whose subtraction result is greater than or equal to 0 is extracted. Since the augmented temperature field containing hidden cold point temperature pits was introduced in the previous reconstruction process, the mesh nodes located at the three-dimensional spatial coordinate position of the inner end face of the thermal bridge component have a lower local saturated water vapor density numerical benchmark, and therefore will first exceed the phase change triggering threshold in the global comparison algorithm. The coordinate indices of solid surface mesh nodes that satisfy the above inequality conditions are extracted, and the spatial envelope domain formed by their set is defined as the condensation active region. In step 404, the instantaneous condensation mass flux of each node in the condensation active region is calculated by combining the local mass transfer coefficient and the difference between the initial humidity field and the saturated water vapor density field, and is used as the phase change mass transfer intensity.
[0070] For grid nodes whose spatial coordinates fall within the active condensation region, the system loads a local surface mass transfer model under natural convection conditions. Specifically, the Grashof number of the driving fluid is calculated based on the local temperature difference between the current grid node and the surrounding gas. The Schmidt number of the current medium environment is then incorporated into the empirical correlation of fluid mechanics to obtain the Sherwood number, which is then used to derive the local mass transfer coefficient at that coordinate location. The empirical correlation is a well-known natural convection mass transfer correlation in the field. Those skilled in the art can select an appropriate Sherwood number empirical correlation based on the geometric characteristics and orientation of the component surface; this is a conventional technique in fluid mechanics and mass transfer. The absolute difference between the local mass transfer coefficient and the initial humidity field value and the saturated water vapor density field value at the current node coordinate location is multiplied by a scalar multiplication operation to calculate and output the water condensation mass parameter per unit surface area and per unit time, which is used as the instantaneous condensation mass flux characterizing the intensity of local liquefaction.
[0071] Step 405: Each node in the active condensation area is designated as a condensation sink.
[0072] In terms of microscopic fluid dynamics, the formation of a phase change water film is accompanied by a continuous unidirectional transfer of gaseous free water molecules to the solid-liquid interface. Spatial nodes within the active condensation region, leveraging local supersaturation concentration differences, continuously draw water vapor molecules from the adjacent open air medium, effectively acting as a negative source that continuously consumes water vapor mass at the topological level of the physical field. By defining and marking the coordinates of such nodes undergoing substantial phase change as condensation sinks in the system memory, the central reference point for subsequent three-dimensional mass diffusion calculations is determined.
[0073] Step 406: Based on the steady-state molecular diffusion approximation model, and combining the instantaneous condensation mass flux with the corresponding surface micro-element area determined by the spatial grid discrete parameters, calculate the local humidity deficit of each condensation sink to the surrounding space.
[0074] Conventional environmental monitoring schemes cannot detect the localized moisture absorption phenomenon caused by phase change condensation sinks, and directly using measured values can easily lead to distortion in the global distribution reconstruction. To address this error quantification issue, it is assumed that the relaxation time of the water vapor diffusion process in a finite confined space is less than the control sampling step size configured by the controller, approximating its dynamic behavior as quasi-steady-state diffusion. Simultaneously, to eliminate dimensional conflicts in unit calculations, the instantaneous condensation mass flux output by the matrix calculation is multiplied by the surface micro-area corresponding to its grid node, converting it into a flow rate parameter characterizing the absolute mass absorption rate of the nodal source. This parameter is then substituted into the three-dimensional diffusion control equations to calculate the attenuation of its influence domain radius. The specific calculation logic is shown in the following equation: δ _ρv =-(m _dotc *δ _As ) / (4*π*D_v *L _dist ); Where, δ _ρv To address the localized humidity deficit caused by condensation at the target node, m _dotc δ represents the instantaneous condensation mass flux at the current condensation sink coordinates. _As D represents the surface area of the grid node at that coordinate, where π is the constant of pi. _v L is the steady-state molecular diffusion coefficient of water vapor in the current air mixture. _dist The distance between the physical coordinates of the target spatial node to be calculated and the physical coordinates of the condensation sink center is the Euclidean distance in space.
[0075] Step 407: Superimpose the effects of all condensation sinks to obtain the cumulative humidity deficit field, which is used as the spatial humidity deficit.
[0076] Considering that multiple spatial condensation sources often exist inside outdoor ring main units due to the complex connections of multiple components, a three-dimensional spatial summation operation is performed on the local concentration disturbances generated by all grid nodes within the active condensation area, based on the principle of independent superposition of linear diffusion fields. By traversing the temperature field to reconstruct each target index node in the grid, the water vapor density reduction caused by all condensation sinks in the space is accumulated and merged to generate a cumulative humidity deficit field covering the entire effective volume of the unit. This matrix field is used to characterize the overall spatial humidity deficit situation faced by the system.
[0077] Step 408: Add the measured values of the multi-point temperature and humidity observation data inside the cabinet at each measuring point to the disturbance amount of the cumulative humidity loss field at the corresponding location to obtain the sink compensation observation data.
[0078] Due to the phase change suction effect, the sensor measurement data acquired by the physical sensor during condensation has actually undergone unidirectional negative distortion. The corresponding cumulative humidity deficit field disturbance value in the three-dimensional coordinate system of the distributed physical sensor is extracted and used as an inverse error compensation operator applied to the original hardware data link. Its calculation logic is shown in the following equation: ρ _vc =ρ _vo - _ρvs ; Where, ρ _vc The data generated after bias correction and sink compensation are ρ _vo This represents the original measured water vapor density output by the physical sensor at its coordinate grid points. _ρvs This is the value of the disturbance output at the same coordinate grid point for the cumulative humidity deficit field, and this parameter has a negative value attribute.
[0079] For example, assuming that after dimensionless normalization, the original measured value of water vapor density at a certain physical sensor coordinate location is 0.85, the normalized perturbation of the cumulative humidity deficit field at that location calculated according to the diffusion model is -0.05, indicating that the physical amplitude of gaseous water molecules in this region is reduced by 0.05 due to the suction of nearby condensation sinks. Using the above reverse compensation calculation logic, the observation data obtained after sink removal compensation is 0.85 - (-0.05) = 0.90. The calculated parameter value of 0.90 represents the true background humidity data benchmark that this micro-space environment should maintain after eliminating local suction interference.
[0080] Step 409: Replace the original measured values with the sink compensation observation data and re-perform spatial interpolation, and update the instantaneous condensation mass flux and cumulative humidity deficit field.
[0081] The generated one-dimensional sequence of desinking compensation observation data is input into the controller interpolation module. The original measured dataset that has been physically distorted is cleared from the memory stack, and parameter replacement is performed. The initialization global interpolation step is performed again on the temperature field reconstruction grid of the same dimension to generate the iterative process field. Based on the replaced iterative humidity field data, it is imported again into the phase transition benchmark comparison stage to refresh the coordinate space envelope of the active condensation region. A new round of instantaneous condensation mass flux matrix and cumulative humidity deficit field matrix are updated according to the corresponding equations, thereby establishing a closed loop of numerical data iterative update in the control system.
[0082] Step 410: Determine whether the rate of change of instantaneous condensation mass flux calculated in two consecutive iterations is less than a set threshold. If it is greater than or equal to the set threshold, return to the step of adding back the disturbance based on the updated cumulative humidity deficit field to perform the next iteration. If it is less than the set threshold, determine that the iteration has converged and output the converged humidity field as the corrected humidity field.
[0083] Within the iterative update closed loop of the control system, the instantaneous condensation mass flux calculation result matrix from the previous iteration cycle and the corresponding result matrix for the current iteration cycle are read. The absolute value of the difference between the coordinate data of the same grid node positions in both cycles is calculated, and the maximum infinite norm of all absolute differences in the matrix is calculated as a rate of change detection index for evaluating the convergence evolution of the physical field. The calculated rate of change detection index is compared with a stop judgment threshold preset by the controller. The stop judgment threshold can be set by those skilled in the art based on the controller's calculation accuracy and the actual engineering requirements for humidity field reconstruction accuracy. Its specific value can be determined by observing the convergence curve of the iterative residual under typical operating conditions. Those skilled in the art can select a suitable convergence criterion without creative effort. If the judgment result shows that the index is greater than or equal to the set threshold, the system is instructed to save the intermediate state and extract new desinking compensation observation data to start the next spatial field reconstruction. If the judgment result shows that the index is less than the set threshold, it indicates that the nonlinear phase change pumping and diffusion replenishment behavior in the microenvironment space has reached a quasi-steady-state numerical equilibrium. The controller interrupts the iterative loop operation, extracts the global humidity field matrix value generated in the last cycle, and outputs it as a corrected humidity field. The output set of corrected humidity field matrices eliminates physical interference from hidden phase transition sinks through dynamic compensation, providing a three-dimensional meteorological data layer that closely approximates the real objective distribution state for subsequent intervention and control.
[0084] Example 5: Based on Examples 1 and 4 above, this example, as a closed-loop control link that transforms the reconstructed physical field into spatially oriented execution actions, further details the prediction mechanism for the expansion of the condensation front based on the spatial gradient of oversaturation and the underlying hardware execution process of spatially oriented non-uniform intervention.
[0085] In one possible implementation, the following steps are included: Step 501: The ratio of the corrected humidity field to the saturated water vapor density field is processed to generate a spatial supersaturation field.
[0086] To quantify the local spatiotemporal evolution dynamics of water vapor exceeding the critical saturation state in the air inside the cabinet, a non-uniform dimensionless field needs to be constructed. The controller loads a field operator analysis module, extracting the corrected humidity field values from the converged output of the previous steps at each grid node, as the numerator; simultaneously, it extracts the saturated water vapor density field values corresponding to the same grid coordinate index, as the denominator. By performing scalar division operations node by node, a spatial supersaturation field reflecting the local water vapor accumulation saturation level is calculated and output. Regions in this data field with values greater than 1 correspond to physical features where condensation has occurred or is about to occur at a high risk of condensation.
[0087] Step 502: Locate the boundary node of the interface between the active condensation area and the non-condensation area.
[0088] Extract the set of 3D physical coordinates of grid nodes marked as active condensation regions from system memory, and simultaneously extract the set of coordinates of grid nodes in non-condensation regions with values less than 1. Load the boundary search operator and use spatial topological neighborhood discrimination rules to compare the phase change attribute states of adjacent grid nodes. When a node belonging to an active condensation region is found to have at least one node belonging to a non-condensation region in its neighborhood, mark the boundary grid points of this set of neighborhoods as interface boundary nodes to delineate the spatial physical boundary of the phase change water film within the microenvironment at the current moment.
[0089] Step 503: Calculate the spatial gradient of each interface boundary node in the spatial supersaturation field, and spatially summarize the outward normal gradient that points to the non-condensation region and has the largest value to obtain the condensation extension vector field.
[0090] Traditional control methods rely solely on local measurement point exceedances to trigger start-stop actions, failing to predict the phase transition propagation trend caused by high environmental gradient changes. This step defines a prediction vector using field theory. For each extracted interface boundary node, its first-order partial derivative in the spatial supersaturation field is calculated, yielding a local three-dimensional gradient vector reflecting the spatial rate of change of the water concentration gradient. According to the laws of physical mass transfer and diffusion, water tends to diffuse and spread towards unsaturated regions.
[0091] The execution condition filtering removes gradient components pointing into active condensation regions, retaining only the local outward normal gradient vector pointing into non-condensation regions with the largest modulus. All calculated local outward normal gradient vectors on the temperature field reconstruction mesh boundary are then combined into a spatial tensor to output a continuous condensation expansion vector field. This vector field indicates the spatial orientation and geometric path of the condensation area to spread in the next control cycle. Its expression is shown in the following equation: NablaS(x _bnd ,t)=Nabla[ρ _vh *(x,t) / ρ _v [sat(x,t)]|x=x _bnd ; Among them, NablaS(x _bnd Let x(t) be the spatial gradient of the interface boundary node at sampling time t, and Nabla be the gradient operator. _bnd ρ represents the spatial physical coordinates of the located interface boundary nodes. _vh *(x,t) represents the corrected humidity field at coordinate x, where the superscript indicates the corrected value after convergence through the sink removal iteration, and ρ _v ,sat(x,t) is the value of the saturated water vapor density field at coordinate x.
[0092] Step 504: Extract the steady-state condensation area in the corrected humidity field, and extract the peak value and spatial mean of the instantaneous condensation mass flux.
[0093] Before calculating the final control strategy, it is necessary to extract macroscopic characteristic indicators of the current phase transition intensity within the cabinet from multiple dimensions. By accumulating the values of grid nodes satisfying the phase transition triggering inequality and multiplying them by the element volume or the solid surface area of a single grid, the current steady-state condensation area is calculated and output. Simultaneously, the numerical matrix of all nodes marked as active condensation regions is traversed to retrieve the peak value of the current instantaneous condensation mass flux, and a spatial arithmetic mean is calculated to output the corresponding spatial mean. These three parameters quantify the degree of environmental thermodynamic crisis from the dimensions of spatial scale, local maximum deterioration, and overall phase transition rate.
[0094] Step 505: Calculate the spatial distance between the condensation spread vector field along its pointing direction and the critical insulation surface set inside the cabinet.
[0095] The controller loads the three-dimensional spatial layout geometry parameters of the equipment. These parameters pre-store the spatial physical coordinate envelope of key insulating surfaces such as the high-voltage busbars, circuit breaker bushings, and supporting insulators within the cabinet. Following the spatial pointing paths of the gradient arrows in the condensation propagation vector field, the controller extends outward along straight lines to intersect and detect. It calculates the spatial straight-line distance from each interface boundary node to the physical outer edge of the key insulating surface and extracts the set of smaller distances as the spatial distance characterizing the degree of threat posed by condensation propagation to the system's electrical insulation.
[0096] Step 506: A weighted score is calculated based on the steady-state condensation area, peak value, spatial mean, and spatial distance to obtain the condensation risk score.
[0097] The calculated steady-state condensation area, peak instantaneous condensation mass flux, and spatial mean instantaneous condensation mass flux are multiplied by the corresponding preset positive weighting coefficients of the controller using scalar multiplication. The calculated spatial distance is input into a nonlinear negative correlation penalty function. The output score of this penalty function is negatively correlated with the spatial distance and increases nonlinearly and monotonically as the spatial distance decreases, increasing the penalty score as the spatial distance approaches a critical value. In an optional implementation, the positive weighting coefficients can be determined by those skilled in the art based on the electrical insulation level and operational safety margin requirements of the ring main unit through engineering calibration or experiments under verification conditions. Those skilled in the art can determine a suitable configuration without creative effort. The nonlinear negative correlation penalty function can be a negative exponential function, and its attenuation constant can be adaptively configured according to the cabinet's internal geometry and insulation creepage distance specifications. The above four sub-scores are summed to finally quantify and output a scalar condensation risk score.
[0098] The nonlinear negative correlation penalty function is specifically f(d) = exp(- d / d crit ),in, d The spatial distance , d crit This is the critical insulation distance.
[0099] Step 507: Map the condensation risk score to a graded intervention instruction. The graded intervention instruction includes the spatial area that needs to be prioritized for heating, determined by spatial distance, and the lower limit of heating power, determined by peak value. Output the graded intervention instruction as a directional dehumidification control instruction.
[0100] The controller has a pre-stored control rule mapping table. When the condensation risk score falls within a specific numerical range, a corresponding graded intervention command is triggered. To achieve spatially targeted regulation of energy and fluid, the command directly designates a specific local spatial area, determined by spatial distance, that is about to be invaded by the condensation front as the priority heating area. In terms of energy intensity, to ensure that the provided heat can cover the latent heat required for water vaporization, the peak value of the instantaneous condensation mass flux is converted into mass flow rate, multiplied by the latent heat of vaporization parameter of water, and the lower limit of heat demand is calculated and defined as the lower limit of heating power. Combining the spatial coordinate parameters of the priority heating area with the electrical energy control parameters of the lower limit of heating power, a directional dehumidification control command is output.
[0101] Step 508: The spatial region is resolved into a control unit matrix in the three-dimensional spatial coordinate system within the ring main unit.
[0102] After receiving the directional dehumidification control command, the underlying hardware driver module needs to convert the spatial region in the command into a hardware topology index in order to direct the physical hardware to act. By calling the spatial mapping operator, the physical coordinate geometric parameters of the space region to be heated first are converted into a control unit matrix that the controller can recognize. The row and column indices of this control unit matrix directly correspond to the physical addresses of the hardware actuators installed at different spatial levels and orientations inside the ring main unit.
[0103] Step 509: According to the lower limit of heating power, independently drive the specific area heater array located in the control unit matrix, and adjust the angle of the guide fan linked with the specific area heater array so that the dehumidifying heat flow covers the space in the opposite direction of the condensation expansion vector field.
[0104] Specifically, the ring main unit is equipped with a distributed array of heaters for specific areas, consisting of multiple independently controllable electric heating plates. Each heating plate is equipped with a guide fan with a stepper motor on its side. The underlying hardware driver module outputs a duty cycle control signal to the physical address corresponding to the control unit matrix, independently driving the corresponding heater array for specific areas to output heat flow energy matching the lower limit of the heating power. At the same time, pulse signals are sent to the stepper motors to adjust the deflection angle of the guide fans, so that the generated dehumidifying heat flow is directionally sprayed and covered along the opposite direction of the condensation expansion vector field. This forms a high-temperature, low-humidity thermal barrier on the non-condensing insulation transition zone, intercepting and evaporating the moisture spreading from the condensation expansion front.
[0105] In some alternative implementations, as an alternative to the hardware combination of a heater array in a specific area and a flow guide fan, the semiconductor cooling dehumidifier can be driven independently in a specific area corresponding to the control unit matrix. By controlling the speed of the local cold end semiconductor suction fan, the moisture in the air is forcibly diverted and stripped along the lateral direction of the condensation expansion vector field, thus achieving the same technical effect of directional suppression of spread.
[0106] Compared to traditional global heating control schemes based on single-point temperature and humidity threshold triggering, this method, by introducing thermal bridge transient temperature extrapolation and sink removal iterative compensation mechanisms, can identify phase change precursors in hidden cold spot areas at the initial stage of condensation, reducing the number of false starts of dehumidifiers caused by sensor distortion and lowering the cumulative runtime of the heaters inside the cabinet. Those skilled in the art will understand that the specific performance improvement may vary depending on the ring main unit model, installation environment, and climatic conditions.
[0107] Example 6 further explains the safety fuse and abnormal state intervention branch mechanism when the evolution of environmental moisture accumulation causes an insulation crisis.
[0108] In one possible implementation, the following steps are included: Step 601: Determine whether the condensation risk score exceeds the pre-configured safety alarm threshold.
[0109] Following the multidimensional quantitative evaluation phase, the controller loads a logic state discrimination operator to intercept and compare the condensation risk score in scalar numerical form of the quantized output. The pre-configured safety alarm threshold in the device's memory reflects the limit safety margin parameter that the insulation medium of this type of ring main unit can withstand under specific ambient temperatures due to liquid water film coverage. The calculated condensation risk score is used as the input variable, and a numerical comparison arithmetic operation is performed with the pre-configured safety alarm threshold. Based on the Boolean logic result output by the comparison operator, the control flow is driven into a parallel branch intervention path.
[0110] In the first execution path, if the condensation risk score is determined to be less than or equal to the pre-configured safety alarm threshold, specifically, the system maintains the normal microenvironment adjustment control logic, generates the corresponding directional dehumidification control command according to the control mapping rules described in the previous embodiment, and independently drives the heater array in a specific area to output heat energy through the underlying hardware drive control loop to cope with the physical disturbance of moisture condensation that has not exceeded the limit.
[0111] In the second execution path, if the pre-configured safety alarm threshold is exceeded and the spatial distance is less than the set critical insulation distance, the insulation fault alarm of the upper-level scheduling system will be triggered and the corresponding local high-voltage electrical circuit will be cut off while the directional dehumidification control command is output; if the pre-configured safety alarm threshold is exceeded but the spatial distance is greater than or equal to the set critical insulation distance, the directional dehumidification control command will be maintained without triggering the insulation fault alarm.
[0112] Specifically, when the logic state discrimination operator outputs a risk score exceeding the limit, the controller further reads the Euclidean space distance parameter between the front end of the condensation extension vector field and the critical insulation surface, and performs a secondary condition judgment with the preset critical insulation distance. The critical insulation distance is set based on the minimum creepage distance constant in the electrical engineering anti-creep distance specification. When the calculated Euclidean space distance value is detected to be less than the critical insulation distance setting value, it indicates that the leading edge of the phase change condensation film has approached the high-voltage busbar or supporting insulator, and is in a high-probability fault window period that may induce high-voltage discharge. Under this condition, in addition to continuing to drive the heater array to perform microenvironmental heat flow evaporation, the controller's underlying protection module synchronously generates a high-priority interrupt level signal. An alarm data packet, including the device's physical network card address, the current humidity field distribution matrix, and the over-limit status code, is sent to the upper-level power grid dispatch center node through the industrial communication Ethernet interface. Upon receiving an interruption signal, the serially connected hardware relay protection circuit closes, providing drive current to the trip coil of the high-voltage circuit breaker. This forces the circuit breaker to open mechanically, physically isolating the power supply branch in the humidity hazard zone. This cuts off the physical medium channel where arc discharge occurs, preventing the accumulation of water vapor phase change from causing unexpected phase-to-phase short circuits and electrical damage.
[0113] It should be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present invention will not describe the various possible combinations separately.
Claims
1. An outdoor ring main unit condensation active suppression method, characterized in that, include: Acquire multi-point temperature and humidity observation data inside the cabinet, external wall temperature data, and external environmental temperature and humidity data, and extract pre-stored thermal bridge structure property parameters; Based on the external wall temperature data, thermal bridge structural property parameters, and local air temperature reflected by multi-point temperature and humidity observation data inside the cabinet, the transient temperature of the inner end face of each thermal bridge component is estimated, and the transient temperature of the inner end face of the thermal bridge is merged with the multi-point temperature and humidity observation data inside the cabinet to reconstruct the augmented temperature field. The augmented temperature field is obtained by merging and reconstructing the transient temperature at the inner end face of the thermal bridge with the temperature and humidity observation data from multiple points inside the cabinet, including: The transient temperature of the inner end face of the thermal bridge is used as the measurement value of the virtual sensor and merged with the temperature observation sequence in the multi-point temperature and humidity observation data in the cabinet to form an augmented temperature observation set. Discretize the three-dimensional space inside the cabinet into a temperature field and reconstruct the mesh; The augmented temperature field is obtained by reconstructing the spatial field on the temperature field reconstruction grid based on the augmented temperature observation set. Based on multi-point temperature and humidity observation data inside the cabinet and the augmented temperature field, an initial humidity field and a saturated water vapor density field were constructed respectively, and the active condensation areas were identified by combining the two. Based on the active condensation area and the difference between the initial humidity field and the saturated water vapor density field, the phase change mass transfer intensity is calculated, and the spatial humidity deficit caused by the phase change is calculated based on the phase change mass transfer intensity. The spatial humidity deficit is used to perform reverse compensation on the temperature and humidity observation data of multiple points in the cabinet to obtain the desinking compensation observation data, and the corrected humidity field is obtained by iterative reconstruction based on the desinking compensation observation data. Identify active condensation regions and determine the phase change mass transfer intensity within these regions, including: The initial humidity field is compared with the saturated water vapor density field node by node, and the surface nodes that meet the condition that the water vapor density is greater than or equal to the saturated water vapor density are marked as active condensation regions; surface nodes refer to spatial grid nodes located on the solid surface inside the cabinet. The local mass transfer coefficient is determined based on the local temperature difference and fluid properties of each node in the active condensation region. The instantaneous condensation mass flux of each node in the active condensation region is calculated by combining the local mass transfer coefficient and the difference between the initial humidity field and the saturated water vapor density field, and is used as the phase change mass transfer intensity. The spatial humidity deficit caused by phase change is calculated based on the phase change mass transfer intensity in the active condensation region, specifically as follows: Each node within the active condensation area is considered a condensation sink; Based on the steady-state molecular diffusion approximation model, and combining the instantaneous condensation mass flux with its corresponding surface micro-area, the local humidity deficit of each condensation sink to the surrounding space is calculated. The effects of all condensation sinks are superimposed to obtain the cumulative humidity deficit field, which is used as the spatial humidity deficit. Based on the corrected humidity field and saturated water vapor density field, the condensation extension vector field and condensation risk score are determined, and corresponding directional dehumidification control commands are output, specifically including: The spatial supersaturation field is generated by comparing the corrected humidity field with the saturated water vapor density field. Locate the boundary node at the interface between the active condensation area and the non-condensation area; Calculate the spatial gradient of each interface boundary node in the spatial supersaturation field, and spatially summarize the outward normal gradient that points to the non-condensation region and has the largest value to obtain the condensation extension vector field. Extract the steady-state condensation area in the corrected humidity field, and extract the peak value and spatial mean of the instantaneous condensation mass flux; Calculate the spatial distance between the condensation propagation vector field along its pointing direction and the critical insulating surface set inside the cabinet; A condensation risk score is obtained by weighting the steady-state condensation area, peak value, spatial mean, and spatial distance. The condensation risk score is mapped to a graded intervention instruction, which includes the spatial area that needs to be prioritized for heating, determined by spatial distance, and the lower limit of heating power, determined by peak value. The graded intervention instruction is output as a directional dehumidification control instruction.
2. The method of claim 1, wherein, After acquiring multi-point temperature and humidity observation data inside the cabinet and external environmental temperature and humidity data, the following is also included: The saturated water vapor pressure at the corresponding temperature is calculated using empirical formulas. Combined with the ideal gas law, the relative humidity sequence in the multi-point temperature and humidity observation data inside the cabinet and the external environment temperature and humidity data is converted into the corresponding water vapor density observation values. The water vapor density observation values are used as the basis data for constructing the initial humidity field.
3. The method of claim 1, wherein, Estimate the transient temperature of the inner end face of each thermal bridge component, including: Each thermal bridge component is equivalent to a one-dimensional transient heat conduction unit; Based on the geometric dimensions and material properties in the thermal bridge structure, the lumped thermal resistance and lumped heat capacity of each thermal bridge component are calculated to obtain the thermal parameter set of the thermal bridge component. The transient temperature of the inner end face of the thermal bridge is calculated based on the thermal parameter set of the thermal bridge component, the external wall temperature data, and the local air temperature reflected by the multi-point temperature and humidity observation data inside the cabinet.
4. The method of claim 3, wherein, The transient temperature of the inner end face of the thermal bridge is calculated based on the thermal parameter set of the thermal bridge component, specifically as follows: The outer wall temperature data is used as the outer end conduction boundary condition of the one-dimensional transient heat conduction unit, and the convective coupling between the inner end face of the thermal bridge component and the local air temperature reflected by the multi-point temperature and humidity observation data in the cabinet is used as the second boundary condition. Combined with the lumped thermal resistance and lumped heat capacity of the thermal bridge component, the dynamic equation of the two-port thermal network is constructed. The second boundary condition refers to either the boundary condition for convective heat transfer at the inner end or the convective coupling boundary condition at the other end. The dynamic equations of the two-port thermal network are recursively updated within a set sampling step using a discretization scheme, and the transient temperature of the inner end face of the thermal bridge is output.
5. The method according to claim 1, characterized in that, Based on the augmented temperature observation set, spatial field reconstruction is performed on the temperature field reconstruction grid to obtain the augmented temperature field, specifically: In the process of spatial field reconstruction, a piecewise covariance structure is introduced, and a local anisotropic covariance kernel is defined within a radius that is a multiple of the characteristic size of each thermal bridge component, with the position of each thermal bridge component as the center. Adjust the correlation length of the local anisotropic covariance kernel in the axial direction of the thermal bridge component to make it shorter than the correlation length of the normal region; treat the area outside the coverage of the local anisotropic covariance kernel as the normal region; Based on the local anisotropic covariance kernel, the conditional expectation of each node on the temperature field reconstruction grid is calculated to generate an augmented temperature field that retains local extremum features.
6. The method according to claim 1, characterized in that, The initial humidity field and the saturated water vapor density field are constructed separately, including: An initial humidity field is generated based on multi-point temperature and humidity observation data inside the cabinet using a spatial interpolation method. By using the augmented temperature field, the saturated water vapor density corresponding to each spatial grid node is calculated, and the saturated water vapor density field is obtained.
7. The method according to claim 1, characterized in that, Before acquiring multi-point temperature and humidity observation data inside the cabinet, the process also includes pre-constructing the physical property parameters of the thermal bridge structure, specifically: Extract the spatial coordinates and cross-sectional geometric dimensions of each metal fastener and through-piece from the pre-stored digital twin model of the cabinet; Match the thermal conductivity and specific heat capacity of each metal fastener and through-part, combine them to generate a thermal bridge structure parameter table and store it in the controller as pre-stored thermal bridge structure property parameters.
Citation Information
Patent Citations
High-voltage switch cabinet anti-condensation method based on simulation optimization and storage medium
CN120372930A
Electrical cabinet condensation defense method and system based on environmental parameter monitoring
CN120848658A
