Three-phase four-wire natural cooling power conversion method and system

Abstract

The present application relates to power conversion technical field, especially a kind of three-phase four-wire natural heat dissipation power conversion method and system, by collecting three-phase current voltage and based on instantaneous power theory decomposition negative sequence and zero sequence current vector, generate each phase compensation instruction distribution to multiple groups of bidirectional power conversion unit and predict power loss distribution, establish heat conduction and natural convection coupling model calculation steady temperature field identification thermal stress concentration area, accordingly, compensation task is preferentially distributed to shorter unit of heat dissipation path and reduces thermal stress area switch frequency, continuously monitor compensation effect and junction temperature and when feedback adjustment distribution weight when over threshold, realize the unbalance compensation of three-phase four-wire system and thermal stress equal distribution, improve the reliability of natural heat dissipation power supply.

Description

Technical Field

[0001] This invention relates to the field of power conversion technology, and in particular to a three-phase four-wire natural heat dissipation power conversion method and system. Background Technology

[0002] In three-phase four-wire power systems, harmonic, reactive power, and imbalance compensation are typically achieved using active power filters (APFs) or static var generators (SVGs). Current conventional methods extract negative-sequence and zero-sequence current components based on instantaneous power theory, generate compensation current commands, and then have multiple parallel power conversion units jointly undertake the compensation task. The output power of each conversion unit is evenly distributed according to its rated capacity or according to a pre-set fixed ratio. The compensation current is generated through PWM modulation, and the control objective focuses on achieving grid current balance and sinusoidal current, with little active consideration given to the loss distribution and thermal state of the conversion units themselves.

[0003] Under natural heat dissipation conditions, the spatial layout differences among multiple power conversion units significantly affect heat dissipation efficiency. Conventional fixed-ratio or equal-rated task allocation methods do not take into account the length of the heat dissipation path, the thermal resistance of devices, and local air convection characteristics, resulting in conversion units located in areas with poor heat dissipation paths experiencing power losses that are mismatched with their actual heat dissipation capacity. This unbalanced heat load can cause local junction temperature increases, accelerating the aging and failure of power semiconductor devices and reducing the long-term operational reliability of the system. If areas of concentrated thermal stress persist, they may also induce potential failures such as solder layer fatigue and package cracking, forcing the system to operate at reduced capacity or undergo unplanned shutdown.

[0004] Existing methods generally lack a closed-loop monitoring mechanism for the junction temperature of the converter unit and a dynamic adjustment mechanism for task allocation. This open-loop thermal management approach cannot fully utilize the heat dissipation margin of each converter unit. When the load fluctuates or the ambient temperature changes, it is easy to interrupt the compensation service prematurely due to local overheating, which restricts the continuous operation capability and power density improvement space of the system. Summary of the Invention

[0005] The present invention provides a three-phase four-wire natural heat dissipation power conversion method and system, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides a three-phase four-wire natural heat dissipation power conversion method, comprising: The phase line current, phase line voltage and neutral line current of a three-phase four-wire power system are collected, and the negative sequence current vector and zero sequence current vector are obtained based on instantaneous power theory. The compensation current command for each phase is generated based on the negative sequence current vector and the zero sequence current vector. The compensation task is assigned to multiple bidirectional power conversion units based on the amplitude and frequency characteristics of the compensation current command. The power loss distribution of each bidirectional power conversion unit is predicted by the loss model. Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify areas of concentrated thermal stress. Based on the thermal stress concentration area, the compensation tasks with large power loss are preferentially assigned to the bidirectional power conversion unit with a shorter heat dissipation path, and the switching frequency of the unit corresponding to the thermal stress concentration area is reduced to form a task allocation scheme. Control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit, and when the junction temperature of each unit exceeds the temperature threshold, feed the deviation back to the task allocation stage to adjust the allocation weight.

[0007] The line currents, line voltages, and neutral current of each phase in a three-phase four-wire power system are collected. Based on instantaneous power theory, the negative-sequence current vector and the zero-sequence current vector are obtained, including: The three-phase line current and neutral line current are collected synchronously using current sensors, and the three-phase line voltage is collected synchronously using voltage sensors. The three-phase line currents are filtered by a sliding window, with the window length adaptively adjusted according to the fundamental period of the three-phase line voltages, to obtain the filtered line currents. The filtered line currents and the three-phase line voltages are then transformed into a synchronous rotating coordinate system to obtain the direct-axis current component and the quadrature-axis current component. The DC component and the frequency-doubled AC component are extracted from the direct-axis current component and the quadrature-axis current component, respectively. The frequency-doubled AC component constitutes the component of the negative sequence current in the synchronous rotating coordinate system. The components of the negative sequence current in the synchronous rotating coordinate system are inversely rotated to the two-phase stationary coordinate system to obtain the orthogonal components of the negative sequence current and synthesize the negative sequence current vector. The neutral line current is used as the amplitude of the zero sequence current vector, and the phase angle of the zero sequence current vector is kept consistent with the phase of the first phase voltage in the three-phase line voltage, thus completing the decomposition of the negative sequence current vector and the zero sequence current vector.

[0008] The compensation current command for each phase is generated based on the negative-sequence current vector and the zero-sequence current vector. The compensation task is then allocated to multiple bidirectional power conversion units based on the amplitude and frequency characteristics of the compensation current command. The power loss distribution of each bidirectional power conversion unit is predicted using a loss model, including: The negative sequence current vector and the zero sequence current vector are subjected to inverse coordinate transformation to obtain the compensation current command for each phase. The compensation current command is subjected to time-frequency analysis. The component with a time proportion exceeding a preset ratio is marked as a continuous compensation component, and the component with a peak amplitude exceeding a preset multiple is marked as a transient impact component. A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component. Steady-state thermal analysis is performed by using average power loss as the node heat source of the thermal resistance network to identify units whose steady-state temperature exceeds a preset threshold, and the spatial region formed by the unit exceeding the threshold and its adjacent units is extracted as the thermal stress concentration region. The continuous compensation component of the highest temperature element in the thermal stress concentration area is transferred to the lowest temperature element outside the area. The average power loss is updated and steady-state thermal analysis is performed again. The task transfer and steady-state thermal analysis are iteratively executed until the temperature of all elements meets the standard. Transient thermal analysis is performed by superimposing the peak power loss and the duration of the transient impact component onto the steady-state temperature, and outputting a power loss distribution that includes both average power loss and peak power loss.

[0009] A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component, including: Obtain the spatial coordinates of each bidirectional power conversion unit, construct a spatial connectivity graph with the bidirectional power conversion unit as the vertex, establish thermal resistance branches between adjacent bidirectional power conversion units, and calculate the unit spacing between adjacent bidirectional power conversion units. The Nusselt number is calculated based on the unit spacing and the kinematic viscosity of the ambient fluid. The convective heat transfer coefficient is calculated by combining the fluid thermal conductivity. The convective heat transfer coefficient is correlated with the unit spacing to determine the thermal resistance value of the thermal resistance branch and to establish a thermal resistance network that includes all thermal resistance branches. The continuous compensation component is sampled in the time domain, the statistical distribution of the current amplitude in the sampling sequence is extracted, the time average value of the conduction loss is determined based on the root mean square value of the current amplitude, the time average value of the switching loss is determined based on the frequency of change of the current amplitude, and the average power loss is obtained. Identify the extreme points of the current amplitude in the transient impact component, extract the current peak value and current rise slope corresponding to the extreme points, determine the transient conduction loss at the extreme points based on the current peak value, determine the transient switching loss at the extreme points based on the current rise slope, and obtain the peak power loss.

[0010] Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify regions of concentrated thermal stress, including: The internal heat source intensity of each bidirectional power conversion unit is determined based on the power loss distribution. The heat exchange path between units is determined based on the spatial layout location, and the spatial topology relationship of the path is extracted. The internal heat source intensity and the spatial topology relationship are associated to form the heat conduction control equation. The contact boundary area between the surface of each bidirectional power conversion unit and the ambient fluid is obtained according to the spatial layout location. The density difference of the ambient fluid is calculated based on the temperature difference of the contact boundary. The fluid velocity field is determined based on the buoyancy force generated in the gravitational field by the density difference. The convective heat transfer intensity of the unit surface is determined based on the fluid velocity field. The temperature field of the heat conduction control equation is used as the driving source of the fluid velocity field, and the convective heat transfer intensity is used as the heat dissipation boundary of the heat conduction control equation. The temperature field and the fluid velocity field are iteratively updated until the temperature change rate is lower than the convergence criterion. The converged temperature field is then obtained as the steady-state temperature field. The spatial gradient distribution of the steady-state temperature field within each bidirectional power conversion unit is calculated. The thermal strain field is obtained based on the spatial gradient distribution and the thermal expansion coefficient of the material. The thermal stress field is obtained based on the thermal strain field and the elastic modulus of the material. Regions in the thermal stress field where the spatial derivative exceeds the stress gradient limit are identified as thermal stress concentration regions.

[0011] Based on the thermal stress concentration area, compensation tasks with higher power losses are preferentially assigned to bidirectional power conversion units with shorter heat dissipation paths, and the switching frequency of the units corresponding to the thermal stress concentration area is reduced, forming a task allocation scheme, including: Identify the bidirectional power conversion unit corresponding to the thermal stress concentration area, traverse the physical connection path from each unit to the heat sink, extract the material type and thickness parameters of the series heat conduction layer in the path, determine the thermal resistance of each heat conduction layer according to the material thermal conductivity and thickness parameters, and accumulate the thermal resistance of each heat conduction layer to form the total heat dissipation thermal resistance of the unit. Obtain the power time-series curve of the compensation task to be executed, perform time-domain integration on the power time-series curve to obtain the cumulative energy within the cycle, use the cumulative energy as the thermal load characterization quantity of the compensation task, and establish the correspondence between the compensation task and the thermal load characterization quantity. The thermal load characterization of the compensation task is cross-matched with the total heat dissipation thermal resistance of each unit. Among all possible matching combinations, the combination in which the product of the thermal load characterization and the total heat dissipation thermal resistance is minimized globally is found. Based on this combination, the execution allocation relationship between the compensation task and the bidirectional power conversion unit is determined. For the units corresponding to the thermal stress concentration area, extract the stress tensor components, perform Fourier transform on the stress tensor components to obtain the stress spectrum distribution, extract the dominant frequency components from the stress spectrum distribution, set the switching frequency of each unit to a frequency value that avoids the dominant frequency components, and form a task allocation scheme based on the execution allocation relationship and the frequency value.

[0012] A second aspect of the present invention provides a three-phase four-wire natural heat dissipation power conversion system, comprising: The negative-sequence and zero-sequence decomposition unit is used to collect the phase line current, phase line voltage and neutral line current of a three-phase four-wire power system, and decompose them into negative-sequence current vector and zero-sequence current vector based on instantaneous power theory. The compensation allocation prediction unit is used to generate compensation current commands for each phase based on the negative sequence current vector and the zero sequence current vector, and to allocate the compensation task to multiple bidirectional power conversion units according to the amplitude and frequency characteristics of the compensation current commands, and to predict the power loss distribution of each bidirectional power conversion unit through the loss model. The thermal stress identification unit is used to establish a coupling model of heat conduction and natural convection based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, calculate the steady-state temperature field, and identify areas of concentrated thermal stress. The thermal stress task allocation unit is used to prioritize the allocation of compensation tasks with large power losses to bidirectional power conversion units with shorter heat dissipation paths based on the thermal stress concentration area, and to reduce the switching frequency of the units corresponding to the thermal stress concentration area, thereby forming a task allocation scheme. The compensation monitoring feedback unit is used to control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit. When the junction temperature of each unit exceeds the temperature threshold, the deviation is fed back to the task allocation stage to adjust the allocation weight.

[0013] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0015] This method decomposes negative-sequence and zero-sequence currents using instantaneous power theory, significantly improving the compensation accuracy for imbalance and zero-sequence harmonics in three-phase four-wire systems. The generation of compensation commands is directly based on the dynamic characteristics of the current vector and neutral current, avoiding phase delays in traditional methods and achieving millisecond-level response. Independent control of the compensation current in each phase enables the system to maintain three-phase current balance even under severe imbalance conditions, effectively reducing neutral current and minimizing harmonic pollution and power loss in the public power grid.

[0016] Based on the thermal coupling analysis of loss model and spatial layout, the steady-state temperature field distribution is accurately predicted and the thermal stress concentration area is located. High-loss compensation tasks are preferentially assigned to units with shorter heat dissipation paths, prompting heat to be rapidly conducted to the natural heat dissipation interface along low thermal resistance paths. At the same time, the switching frequency in the thermal stress concentration area is reduced to avoid thermal failure caused by local hot spots, eliminating the need for additional forced cooling devices.

[0017] Continuous monitoring of the compensated three-phase current balance and junction temperature of each unit forms a closed-loop feedback control. When the junction temperature exceeds the limit, the allocation weights are adjusted in real time to dynamically balance the power loss and heat dissipation capacity of each unit. This adaptive mechanism effectively controls the fluctuation range of junction temperature in each unit and maintains system reliability during sudden load changes or increases in ambient temperature. The entire method has a high degree of integration and low implementation cost, making it particularly suitable for efficient power quality management in industrial power distribution and new energy grid-connected scenarios. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the three-phase four-wire natural heat dissipation power conversion method according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the method for predicting the power loss distribution of each bidirectional power conversion unit according to an embodiment of the present invention. Figure 3 Waveform diagram of compensation current command; Figure 4 This is a diagram showing the distribution of convective heat transfer coefficients for each unit. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0021] Figure 1 This is a schematic flowchart of a three-phase four-wire naturally cooled power supply conversion method according to an embodiment of the present invention. The present invention provides a three-phase four-wire naturally cooled power supply conversion method comprising: The phase line current, phase line voltage and neutral line current of a three-phase four-wire power system are collected, and the negative sequence current vector and zero sequence current vector are obtained based on instantaneous power theory. The compensation current command for each phase is generated based on the negative sequence current vector and the zero sequence current vector. The compensation task is assigned to multiple bidirectional power conversion units based on the amplitude and frequency characteristics of the compensation current command. The power loss distribution of each bidirectional power conversion unit is predicted by the loss model. Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify areas of concentrated thermal stress. Based on the thermal stress concentration area, the compensation tasks with large power loss are preferentially assigned to the bidirectional power conversion unit with a shorter heat dissipation path, and the switching frequency of the unit corresponding to the thermal stress concentration area is reduced to form a task allocation scheme. Control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit, and when the junction temperature of each unit exceeds the temperature threshold, feed the deviation back to the task allocation stage to adjust the allocation weight.

[0022] The line currents, line voltages, and neutral current of each phase in a three-phase four-wire power system are collected. Based on instantaneous power theory, the negative-sequence current vector and the zero-sequence current vector are obtained, including: The three-phase line current and neutral line current are collected synchronously using current sensors, and the three-phase line voltage is collected synchronously using voltage sensors. The three-phase line currents are filtered by a sliding window, with the window length adaptively adjusted according to the fundamental period of the three-phase line voltages, to obtain the filtered line currents. The filtered phase line current and three-phase phase line voltage are transformed into a synchronous rotating coordinate system to obtain the direct-axis current component and the quadrature-axis current component. The DC component and the frequency-doubled AC component are extracted from the direct-axis current component and the quadrature-axis current component, respectively. The frequency-doubled AC component constitutes the component of the negative sequence current in the synchronous rotating coordinate system. The components of the negative sequence current in the synchronous rotating coordinate system are reverse-rotated to the two-phase stationary coordinate system to obtain the orthogonal components of the negative sequence current and synthesize the negative sequence current vector. The neutral line current is used as the magnitude of the zero-sequence current vector. The phase angle of the zero-sequence current vector is kept consistent with the phase of the first phase voltage in the three-phase line voltage, thus completing the decomposition of the negative-sequence current vector and the zero-sequence current vector.

[0023] In the actual operation of a three-phase four-wire power supply system, high-precision synchronous acquisition of current and voltage signals is fundamental for subsequent decomposition calculations. Current sensors are used to monitor the three-phase line currents. , , and neutral current Synchronous data acquisition is performed, and the sampling timing alignment error must be controlled within the microsecond range to avoid introducing false unbalanced components due to sampling timing deviations. Simultaneously, voltage sensors synchronously acquire the three-phase line voltages. , , The voltage and current sampling channels share the same trigger clock, ensuring strict time-coherence between voltage and current data. The analog signal output by the sensor is filtered by an anti-aliasing filter before entering the analog-to-digital converter, with a conversion accuracy of no less than 16 bits to ensure effective resolution of minute imbalance components.

[0024] The raw phase currents inevitably contain high-frequency switching noise and harmonic interference, and directly using them for coordinate transformation will lead to significant deviations in the decomposition results. Therefore, a sliding window filter is applied to the three-phase phase currents. The sliding window length is... Based on the fundamental period of the three-phase line voltage Adaptive adjustment, specifically, based on the fundamental frequency detected in real time. Calculate the fundamental period Then The corresponding number of sampling points is used as the window length. ,Right now ,in This refers to the sampling frequency. When the power grid frequency fluctuates around the mains frequency, The filter is dynamically updated to ensure it maintains precise suppression or preservation characteristics for the fundamental frequency and its integer multiples of harmonics. The filtered phase current is obtained after sliding window filtering. , , High-frequency noise components are effectively suppressed, while the fundamental and low-order harmonic components are preserved.

[0025] The filtered phase line current and the three-phase phase line voltage are fed together into the rotating coordinate transformation stage. The phase angle of the fundamental positive sequence voltage is extracted using a phase-locked loop based on the fundamental positive sequence component of the three-phase phase line voltage. ,by To adjust the rotation angle, a two-step transformation is performed on the filtered phase line currents: from a three-phase stationary coordinate system to a two-phase stationary coordinate system (Clarke transformation) and then to a synchronously rotating coordinate system (Park transformation). The Clarke transformation yields the current components in the two-phase stationary coordinate system. and Then, the direct-axis current component in the synchronous rotating coordinate system is obtained through Park transformation. With cross-axis current component Under ideal three-phase equilibrium conditions, and All are DC flow rates; when the system has three-phase imbalance... and The superimposed AC component, which is dominated by twice the fundamental frequency, is precisely the mapping of the negative sequence current in the synchronous rotating coordinate system.

[0026] from and Extracting the DC component and the harmonic AC component separately is a key step in decomposing the negative sequence current. A low-pass filter with a cutoff frequency lower than the fundamental frequency is used for this purpose. and Perform filtering to extract the DC component and These two represent the positive-sequence fundamental active and reactive current components. The original... , Subtracting the corresponding DC components from each yields the harmonic AC components. and These two harmonic AC components constitute the components of the negative sequence current in a synchronously rotating coordinate system, and their main frequency components are... This reflects the negative sequence component resulting from the asymmetry in the amplitude or phase of the three-phase current. The cutoff frequency of the low-pass filter... Must meet At the same time, its transition band needs to be steep enough to avoid the leakage of the harmonic component into the DC extraction result.

[0027] For the components of negative sequence current in a synchronous rotating coordinate system and Perform the inverse rotation transformation, that is, with Performing an inverse Park transformation on the rotation angle transforms it from a synchronous rotating coordinate system back to a two-phase stationary coordinate system, yielding the orthogonal components of the negative sequence current. and These two orthogonal components constitute a complete negative-sequence current vector description in a two-phase stationary coordinate system, and their instantaneous amplitudes... With instantaneous phase They can be represented as follows: and .Will and The synthesis yields the negative sequence current vector, which fully describes the magnitude and direction of the negative sequence component in the three-phase system, providing a direct basis for the generation of subsequent compensation current commands.

[0028] The extraction method for the zero-sequence current vector differs from that for the negative-sequence current vector. In a three-phase four-wire system, the neutral line current... It directly reflects the zero-sequence component of the three-phase current, because the sum of the three-phase zero-sequence currents equals the neutral current. The instantaneous value is directly used as the amplitude of the zero-sequence current vector. The phase angle of the zero-sequence current vector is then related to the first phase voltage in the three-phase line voltage. The phases remain consistent, that is, the reference phase angle of the zero-sequence current vector is taken as... Fundamental phase angle The phase angle is determined by the target The phase-locked loop can operate independently or be extracted directly from the output of the three-phase phase-locked loop. The physical basis of this approach is that the zero-sequence current in a three-phase four-wire system is mainly caused by the asymmetry of the loads in each phase. Its fundamental component has a definite phase correlation with the fundamental voltage of each phase. Using the phase of the first phase voltage as a reference can unify the phase reference of the zero-sequence compensation command, which facilitates the synthesis and distribution of subsequent compensation current commands for each phase.

[0029] Thus, through the complete acquisition, filtering, coordinate transformation, component extraction, and vector synthesis process described above, accurate decomposition of the negative-sequence current vector and the zero-sequence current vector in a three-phase four-wire power supply system was achieved. The entire decomposition process is executed in real time within each control cycle, and the decomposition results are dynamically updated according to changes in the system's operating state. This accurately tracks the dynamic changes in the three-phase imbalance state, providing real-time and accurate input data for the subsequent compensation task allocation of multiple sets of bidirectional power conversion units.

[0030] Figure 2 This is a flowchart illustrating the method for predicting the power loss distribution of each bidirectional power conversion unit according to an embodiment of the present invention. The method generates compensation current commands for each phase based on the negative-sequence current vector and the zero-sequence current vector, and allocates the compensation task to multiple groups of bidirectional power conversion units according to the amplitude and frequency characteristics of the compensation current commands. The method then predicts the power loss distribution of each bidirectional power conversion unit using a loss model, including: The negative sequence current vector and the zero sequence current vector are subjected to inverse coordinate transformation to obtain the compensation current command for each phase. The compensation current command is subjected to time-frequency analysis. The component with a time proportion exceeding a preset ratio is marked as a continuous compensation component, and the component with a peak amplitude exceeding a preset multiple is marked as a transient impact component. A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component. Steady-state thermal analysis is performed by using average power loss as the node heat source of the thermal resistance network to identify units whose steady-state temperature exceeds a preset threshold, and the spatial region formed by the unit exceeding the threshold and its adjacent units is extracted as the thermal stress concentration region. The continuous compensation component of the highest temperature element in the thermal stress concentration area is transferred to the lowest temperature element outside the area. The average power loss is updated and steady-state thermal analysis is performed again. The task transfer and steady-state thermal analysis are iteratively executed until the temperature of all elements meets the standard. Transient thermal analysis is performed by superimposing the peak power loss and the duration of the transient impact component onto the steady-state temperature, and outputting a power loss distribution that includes both average power loss and peak power loss.

[0031] When performing inverse coordinate transformation on the negative-sequence current vector and the zero-sequence current vector, the negative-sequence and zero-sequence components in the two-phase stationary coordinate system are respectively subjected to inverse Clarke transformation and superposition of the zero-sequence components to restore the compensation current commands for each phase in the three-phase natural coordinate system. Specifically, the compensation current commands for phases A, B, and C are obtained by adding the inverse transformation result of the negative-sequence component to the corresponding zero-sequence current compensation amount, forming a complete three-phase compensation current reference waveform. After obtaining the compensation current commands for each phase, time-frequency analysis is performed on them. Short-time Fourier transform or wavelet transform methods are used, and the analysis window slides on the time axis to extract the energy distribution of each frequency component over time. Components that account for more than a preset proportion (e.g., 60%) of the analysis period are marked as continuous compensation components. These components represent the long-term unbalanced load characteristics of the system and need to be eliminated by the continuous output of compensation current from the bidirectional power conversion unit. Components whose peak amplitude exceeds a preset multiple (e.g., 3 times) of the steady-state amplitude are marked as transient impact components. These components are usually caused by sudden changes in nonlinear loads or the start-up and shutdown of high-power equipment. They have a short duration but a significant thermal impact on power devices.

[0032] like Figure 3As shown in the figure, this diagram illustrates the time-domain waveform of the three-phase compensation current command generated after the inverse transformation of the negative-sequence and zero-sequence current vector coordinates. The horizontal axis represents time (ms), and the vertical axis represents the compensation current (A). The blue, green, and yellow curves in the figure correspond to the compensation commands for phases A, B, and C, respectively, with labeled amplitudes of 8.40 A for phase A, 12.70 A for phase B, and 6.30 A for phase C. From the phase relationship, the command for phase A exhibits a leading characteristic, phase B exhibits a lagging characteristic, while phase C is in a relatively balanced compensation state. This intuitively reflects the system's differentiated suppression strategy for three-phase unbalanced loads. The leading and lagging components mainly target the amplitude asymmetry caused by the negative-sequence current, while the balanced phase undertakes the compensation for the zero-sequence and residual harmonic components. This waveform not only serves as the control reference for generating PWM modulation waves by each bidirectional power conversion unit, but its amplitude envelope and frequency components also serve as the core input for time-frequency analysis: through short-time Fourier transform, the system marks components with a time proportion exceeding 60% as continuous compensation components (used for long-term load balancing), and components with peak amplitudes exceeding three times the steady-state value as transient impact components (used to cope with sudden load disturbances), thereby providing accurate instruction-level data support for the calculation of average power loss and peak power loss in the subsequent loss model.

[0033] When establishing the thermal resistance network between each bidirectional power conversion unit, each bidirectional power conversion unit is considered as a hot node. The thermal resistance between adjacent units is determined by both the unit spacing and the natural convection heat transfer coefficient of the air. Let the spacing between two adjacent units be... The convective heat transfer coefficient is Equivalent convective thermal resistance It can be represented as ,in This represents the equivalent heat transfer area between two units. For units located inside a sealed enclosure, the parallel contribution of radiation thermal resistance must also be considered. Radiation thermal resistance is determined by the fourth square difference between the unit surface emissivity and the absolute temperature. In the initial temperature estimation stage, radiation thermal resistance can be approximated as a fixed value to simplify the calculation. After the thermal resistance network is established, the average power loss of each unit under stable operating conditions is calculated based on the continuous compensation component. The average power loss includes two parts: switching loss and conduction loss. Switching loss is proportional to the switching frequency and the square of the current amplitude, while conduction loss is related to the product of the effective current value and the device's on-resistance. The peak power loss of each unit is calculated based on the peak amplitude and duration of the transient impact component. The peak power loss reflects the maximum thermal stress that the device withstands in a short period of time.

[0034] The average power loss of each element is used as the heat source for the corresponding thermal node in the thermal resistance network. Steady-state thermal analysis is performed with ambient temperature as the reference boundary condition. The steady-state thermal analysis obtains the steady-state temperature of each element by solving the thermal node equations. Let the th element be... The steady-state temperature of each unit is It satisfies the nodal thermal balance equation: the heat power flowing into the node is equal to the sum of the heat power dissipated to adjacent nodes and the environment through the thermal resistance of each branch. After solving this set of equations, the steady-state temperature of each unit is compared with a preset threshold. Compare and identify steady-state temperatures exceeding The spatial region formed by these over-threshold cells and their spatially adjacent cells is defined as a thermal stress concentration region. Identifying thermal stress concentration regions helps to target and reduce the thermal load of cells within this region during subsequent task allocation, avoiding localized overheating that could lead to a decrease in device reliability.

[0035] After identifying the thermal stress concentration region, an iterative task transfer process is executed. In each iteration, the element with the highest steady-state temperature within the thermal stress concentration region is identified, and the heaviest portion of the continuous compensation component borne by that element is transferred to the element with the lowest steady-state temperature outside the region. The principle for determining the amount of task transfer is: after the transfer, the steady-state temperature of the element with the highest temperature within the thermal stress concentration region shall not exceed [a certain value]. Meanwhile, the steady-state temperature of the external unit receiving the transfer mission does not exceed To avoid creating new hot spots while eliminating existing thermal stress concentration areas, after completing one task transfer, the average power loss of each element is updated, the thermal resistance network node equations are re-established, and the new steady-state temperature distribution is solved. If elements still exceed the threshold, the next round of task transfer and steady-state thermal analysis is performed until the steady-state temperature of all elements does not exceed the threshold. Until then. The criteria for judging the convergence of the iteration are that the steady-state temperature of all units reaches the standard, or the maximum temperature decreases below the set convergence threshold after several consecutive iterations (at which point it is considered that the optimal allocation state has been reached under the current hardware configuration, and the frequency reduction protection mechanism needs to be triggered).

[0036] After iterative optimization of the steady-state thermal analysis, the transient thermal response is further analyzed. The peak power loss of each element is compared with the duration of the transient impact component. Superimposed to the steady-state temperature, a transient thermal analysis is performed using an RC thermal network model of thermal capacity and thermal resistance. In the transient thermal analysis, the junction temperature transient response of the device is determined by the thermal impedance. describe, Typically fitted by a multi-order RC network, it reflects the thermal capacitive effects of each layer of the device, from the chip junction to the case, and from the case to the heat sink. (Peak power loss...) Under the action, the first Transient junction temperature increment per unit satisfy ,in Duration The corresponding thermal resistance value. The transient junction temperature increment is superimposed onto the steady-state temperature. The peak junction temperature estimates for each unit under the worst operating conditions are obtained to determine whether the device is at risk of exceeding the absolute maximum junction temperature. The final output power loss distribution contains two levels of information: the average power loss describes the thermal load level of each unit under continuous operation, and the peak power loss describes the thermal stress extreme value of each unit under transient impact conditions. Together, they constitute a complete power loss distribution, providing a quantitative basis for the subsequent establishment of a coupled heat conduction and natural convection model and the accurate identification of thermal stress concentration areas.

[0037] A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component, including: Obtain the spatial coordinates of each bidirectional power conversion unit, construct a spatial connectivity graph with the bidirectional power conversion unit as the vertex, establish thermal resistance branches between adjacent bidirectional power conversion units, and calculate the unit spacing between adjacent bidirectional power conversion units. The Nusselt number is calculated based on the unit spacing and the kinematic viscosity of the ambient fluid. The convective heat transfer coefficient is calculated by combining the fluid thermal conductivity. The convective heat transfer coefficient is correlated with the unit spacing to determine the thermal resistance value of the thermal resistance branch and to establish a thermal resistance network that includes all thermal resistance branches. The continuous compensation component is sampled in the time domain, the statistical distribution of the current amplitude in the sampling sequence is extracted, the time average value of the conduction loss is determined based on the root mean square value of the current amplitude, the time average value of the switching loss is determined based on the frequency of change of the current amplitude, and the average power loss is obtained. Identify the extreme points of the current amplitude in the transient impact component, extract the current peak value and current rise slope corresponding to the extreme points, determine the transient conduction loss at the extreme points based on the current peak value, determine the transient switching loss at the extreme points based on the current rise slope, and obtain the peak power loss.

[0038] Before establishing the thermal resistance network, it is necessary to first obtain the installation coordinates of each bidirectional power conversion unit in three-dimensional space and record the position information of each unit. Using each bidirectional power conversion unit as a vertex, a connected graph is constructed according to spatial adjacency relationships. The edges in the connected graph correspond to thermal resistance branches between adjacent units. The criterion for determining adjacency is usually based on whether the unit spacing is less than a preset neighborhood radius. Thermal coupling effects between unit pairs exceeding this radius are negligible, and no thermal resistance branches are established. For unit pairs that satisfy the adjacency condition, the Euclidean norm of their spatial coordinate differences is calculated to obtain the unit spacing corresponding to each thermal resistance branch. subscript and These represent the unit number indices at both ends of the branch.

[0039] Obtain cell spacing Then, the Nusselt number is calculated using the characteristic length correlation criterion in fluid mechanics. Under natural convection conditions, the Nusselt number... The product of Grashof numbers and Prandtl numbers (i.e., Rayleigh numbers) There is a power-law relationship between them. Rayleigh numbers Acceleration due to gravity Coefficient of volume expansion Temperature difference Feature length (take) kinematic viscosity and thermal diffusivity jointly determined, that is Nusselt numbers are calculated according to... Calculation, where and These are empirical constants related to the flow regime, with typical values ​​taken under laminar natural convection conditions. These are combined with the fluid's thermal conductivity. convective heat transfer coefficient Depend on Confirmed. With equivalent heat transfer area Substituting into the thermal resistance calculation relationship, we obtain the thermal resistance value of this thermal resistance branch. By repeating the above calculations for all thermal resistance branches in the connected graph, a complete thermal resistance network encompassing all inter-element thermal coupling effects can be established. This network uses node voltages as analogies to element temperatures and branch currents as analogies to heat flow, allowing for the solution of the steady-state temperature field using the node admittance matrix method.

[0040] When calculating power loss for the continuous compensation component, the continuous component of the compensation current is first sampled at fixed intervals in the time domain to form a current sampling sequence of sufficient length. The current amplitude at each sampling point in this sequence is statistically analyzed to construct a probability distribution histogram of the amplitude. Since conduction loss is directly related to the root mean square (RMS) value of the current, the RMS value is calculated for the sampling sequence. subscript Indicates the first Each bidirectional power conversion unit has a time-averaged conduction loss. Depend on Sure, This represents the on-resistance of the power device in this unit. Switching losses are related to the frequency of current zero-crossing or commutation events. The commutation frequency per unit time is obtained by statistically analyzing the number of times the current amplitude crosses a preset threshold in the sampling sequence. Average time of switching loss Depend on Confirmed, among which The energy loss of this unit in a single switching operation under rated current can be obtained by interpolation of the switching energy curve in the device datasheet. Summing the conduction loss and switching loss yields the first... Average power loss of each unit .

[0041] When calculating the peak power loss of transient impulse components, it is necessary to identify the extreme points of the current amplitude in the time-domain waveform. A first-order difference is performed on the sampling sequence of the transient impulse components; when the difference result changes from positive to negative, it corresponds to the maximum point of the current amplitude, and the peak current at this maximum point is extracted. Simultaneously, the current rise slope at the leading edge of the maximum point is extracted. This slope reflects the rate at which the current jumps from a low value to a peak value, and is obtained by linear fitting of several sampling points before the extreme point. Peak current The transient conduction loss of the power device at the extreme point is determined. ,Depend on The calculated current rise slope The transient switching losses of a device during the current rise phase are determined by the slope of the rise. A steeper rise slope indicates a faster rate of current change, which in turn affects the device's ability to withstand higher current losses. The higher the stress, the higher the corresponding transient switching loss. By The switching loss-current slope characteristic curve of the device is used for interpolation. The transient conduction loss is then superimposed with the transient switching loss to obtain the [then]... Peak power loss of each unit .

[0042] After obtaining the average power loss and peak power loss, they are substituted into the thermal resistance network to calculate the temperature field. The average power loss corresponds to steady-state thermal analysis, and the steady-state temperature of each element is obtained by solving the nodal equations of the thermal resistance network. Peak power loss is determined by the duration of the transient impact component. With device thermal impedance function ,according to Calculate the transient junction temperature increment of each unit under peak power surge, where For the first Each unit in duration The corresponding thermal impedance value is obtained. By superimposing the steady-state temperature and the transient junction temperature increment, the estimated junction temperature of each unit under the worst operating conditions can be obtained, which can be used for subsequent identification of thermal stress concentration areas and dynamic adjustment of task allocation weights.

[0043] The above-mentioned thermal resistance network establishment and power loss decomposition process is executed periodically in each control cycle to ensure that the thermal management strategy can track and compensate for the redistribution of power loss caused by dynamic changes in the task, thereby achieving a balanced distribution of the heat load of each bidirectional power conversion unit under natural heat dissipation conditions.

[0044] Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify regions of concentrated thermal stress, including: The internal heat source intensity of each bidirectional power conversion unit is determined based on the power loss distribution. The heat exchange path between units is determined based on the spatial layout location, and the spatial topology relationship of the path is extracted. The internal heat source intensity and the spatial topology relationship are associated to form the heat conduction control equation. The contact boundary area between the surface of each bidirectional power conversion unit and the ambient fluid is obtained according to the spatial layout location. The density difference of the ambient fluid is calculated based on the temperature difference of the contact boundary. The fluid velocity field is determined based on the buoyancy force generated in the gravitational field by the density difference. The convective heat transfer intensity of the unit surface is determined based on the fluid velocity field. The temperature field of the heat conduction control equation is used as the driving source of the fluid velocity field, and the convective heat transfer intensity is used as the heat dissipation boundary of the heat conduction control equation. The temperature field and the fluid velocity field are iteratively updated until the temperature change rate is lower than the convergence criterion. The converged temperature field is then obtained as the steady-state temperature field. The spatial gradient distribution of the steady-state temperature field within each bidirectional power conversion unit is calculated. The thermal strain field is obtained based on the spatial gradient distribution and the thermal expansion coefficient of the material. The thermal stress field is obtained based on the thermal strain field and the elastic modulus of the material. Regions in the thermal stress field where the spatial derivative exceeds the stress gradient limit are identified as thermal stress concentration regions.

[0045] Before establishing the coupled model of heat conduction and natural convection, it is necessary to extract the internal heat source intensity of each bidirectional power conversion unit from the power loss distribution data. The average power loss and peak power loss of each unit together determine the time-averaged intensity and transient intensity of its internal heat source. The power loss of each unit is converted into volumetric heat source density, that is, the power loss value is divided by the effective heat generation volume of the power device in that unit to obtain the heat generation rate per unit volume. At the same time, based on the installation coordinates of each bidirectional power conversion unit inside the device, the spatial topological relationship between adjacent units is extracted, including the relative position between units, the geometry of contact or adjacent interfaces, and the contact thermal resistance between each unit and the external heat dissipation shell. The above volumetric heat source density and spatial topological relationship are substituted into the steady-state heat conduction control equation to form a discretized linear equation system with temperature as the unknown. The internal nodes of the heat conduction control equation satisfy energy conservation, that is, the sum of the heat flow into the node is equal to the heat generated by the node, and the boundary nodes are subject to corresponding constraints according to different heat dissipation boundary conditions.

[0046] After obtaining the governing equations for heat conduction, it is necessary to further establish the boundary conditions for natural convection heat transfer. Based on the spatial layout of each bidirectional power conversion unit, the outer surface area of ​​each unit exposed to the ambient fluid is extracted. Let the... The external surface temperature of each unit is The ambient fluid temperature is The temperature difference at the contact boundary is Because fluid density decreases with increasing temperature, the fluid density near the hot surface is lower than that farther away, thus generating buoyancy in the gravitational field. This buoyancy drives the fluid upwards along the hot surface, forming a natural convection boundary layer. The strength of the fluid velocity field is determined by the dimensionless Rayleigh number. Characterization, This comprehensively reflects the competitive relationship between buoyancy, viscous forces, and thermal diffusion. When When the temperature is high, convective heat transfer is significant; conversely, when the temperature is low, conduction is dominant. According to... The range of values ​​is selected according to the corresponding Nusselt number correlation, and the first value is calculated. Convection heat transfer coefficient of each unit surface This allows us to obtain the convective heat flux density per unit area of ​​the unit surface. ,in Indicates the first The convective heat flux density of each unit surface, Indicates the first The convective heat transfer coefficient of each unit surface. Indicates the first The temperature difference between the surface of each unit and the ambient fluid.

[0047] like Figure 4As shown in the figure, this bar chart displays the natural convection heat transfer coefficients (unit: W / m²·K) of the surfaces of eight bidirectional power conversion units, from #1 to #8. Due to the non-uniform spatial layout of each unit within the chassis, the contact boundary area and relative height between their surfaces and the ambient fluid differ. This results in varying magnitudes of the buoyancy force generated by the difference in fluid density caused by wall temperature rise in the gravitational field. Consequently, this drives the formation of natural convection boundary layers of varying intensities around each unit. As shown in the figure, units #3 and #6, being located in the main upward airflow channel within the chassis and with fully exposed heat dissipation fins, achieve higher heat transfer coefficients (close to 9.0 W / m²·K); while units #4 and #8, located in dead airflow areas or near the thickened boundary layer region of the chassis wall, have significantly lower heat transfer coefficients (approximately 6.0–6.4 W / m²·K). This non-uniformly distributed data is a key input parameter for constructing the inter-unit thermal resistance network. By correlating the convective heat transfer coefficient of each unit with the spatial spacing between them, the system can determine the thermal resistance value of each branch in the thermal resistance network. Then, these branch thermal resistances are substituted into the coupled model of heat conduction and natural convection for iterative solution of the steady-state temperature field. Accurate heat transfer coefficient distribution directly determines the accuracy of the heat dissipation boundary in the steady-state temperature field calculation, and is an important prerequisite for accurately locating areas of concentrated thermal stress and prioritizing the allocation of high-loss compensation tasks to units with shorter heat dissipation paths.

[0048] The heat conduction governing equations are coupled with natural convection boundary conditions for iterative solution. The initial condition for the iteration process is set so that the temperature of each element is equal to the ambient temperature. Starting from this point, the governing equations for heat conduction are solved to obtain the initial temperature distribution at each node. This temperature distribution is then substituted into the natural convection model to update the convective heat transfer coefficients on each element surface. With convective heat flux density Then, the updated convective heat flux density is used as the heat dissipation boundary condition for the heat conduction governing equation, and the temperature field is solved again. This process is repeated iteratively, and after each iteration, the maximum change in the temperature field between two adjacent iterations is calculated. .when Below the preset convergence criterion When the temperature field reaches a steady state, the iteration stops. The converged temperature field is considered to have reached a steady state, and the steady-state node temperature of each element is denoted as . (here) Consistent with the definitions in claims 3 and 4 above, it indicates that the first Steady-state temperature of each unit). Convergence criterion The typical value range is 0.01°C to 0.1°C, which can be adjusted appropriately according to the accuracy requirements of the project.

[0049] After obtaining the steady-state temperature field, the spatial gradient distribution of the temperature field within each bidirectional power conversion unit is calculated. For the discretized temperature node mesh, the spatial gradient is approximated using the finite difference method. Let the nodes... The temperature at that location is , its in The temperatures of adjacent nodes in the direction are respectively and The node spacing is Then at that node Temperature gradient in the direction for ,in Represents a node Place Temperature gradient in the direction, and Representing nodes respectively exist Temperature of adjacent nodes on both sides of the direction Represents a node The grid spacing at that location. Similarly, at... direction and Calculate the corresponding temperature gradient components for each direction, and synthesize a three-dimensional temperature gradient vector. Its modulus It reflects the degree of temperature change at that node.

[0050] After the temperature gradient distribution is determined, the thermal expansion coefficients of each unit material are considered. Calculate the thermal strain field. Under the assumption of isotropic materials, the nodes... thermal strain at the point It is proportional to the temperature change at that node, that is ,in Represents a node Thermal strain at the point, Indicates the first The coefficient of thermal expansion of each unit material This represents the stress-free reference temperature of the material. In actual packaging structures, different material layers (such as solder layers, substrates, and power chips) have different coefficients of thermal expansion. Thermal expansion mismatch at the interface will generate additional interfacial thermal stress. Therefore, when establishing the thermal strain field, it is necessary to assign corresponding values ​​to each material layer. value.

[0051] Based on thermal strain field and material elastic modulus Further calculations of the thermal stress field are performed. Under the assumption of linear elasticity, the nodes... Thermal stress at the location According to Hooke's Law, ,in Represents a node Thermal stress at the location, Indicates the first The elastic modulus of a single unit material Represents a node Thermal strain at the interface. In multilayer encapsulation structures, due to the different elastic moduli of the materials in each layer, there are abrupt changes in thermal stress on both sides of the interface. Interface compatibility conditions need to be introduced at the interface nodes to ensure displacement continuity. The spatial derivative of the thermal stress field is calculated to obtain the thermal stress gradient distribution. ,in Represents a node The spatial gradient of thermal stress. When a certain region... Exceeding the preset stress gradient limit At that time, the area was identified as a region of thermal stress concentration. Stress gradient limit. Based on material fatigue characteristics and packaging reliability requirements, the settings are typically determined with reference to the reliability specifications provided by power device manufacturers. Information on identified thermal stress concentration areas is then transmitted to the task allocation stage, serving as the basis for adjusting the weighting of compensation tasks and the switching frequency. This ensures effective power quality compensation while reducing the risk of thermal fatigue at critical packaging interfaces, thereby extending the overall device lifespan.

[0052] Based on the thermal stress concentration area, compensation tasks with higher power losses are preferentially assigned to bidirectional power conversion units with shorter heat dissipation paths, and the switching frequency of the units corresponding to the thermal stress concentration area is reduced, forming a task allocation scheme, including: Identify the bidirectional power conversion unit corresponding to the thermal stress concentration area, traverse the physical connection path from each unit to the heat sink, extract the material type and thickness parameters of the series heat conduction layer in the path, determine the thermal resistance of each heat conduction layer according to the material thermal conductivity and thickness parameters, and accumulate the thermal resistance of each heat conduction layer to form the total heat dissipation thermal resistance of the unit. Obtain the power time-series curve of the compensation task to be executed, perform time-domain integration on the power time-series curve to obtain the cumulative energy within the cycle, use the cumulative energy as the thermal load characterization quantity of the compensation task, and establish the correspondence between the compensation task and the thermal load characterization quantity. The thermal load characterization of the compensation task is cross-matched with the total heat dissipation thermal resistance of each unit. Among all possible matching combinations, the combination in which the product of the thermal load characterization and the total heat dissipation thermal resistance is minimized globally is found. Based on this combination, the execution allocation relationship between the compensation task and the bidirectional power conversion unit is determined. For the units corresponding to the thermal stress concentration area, extract the stress tensor components, perform Fourier transform on the stress tensor components to obtain the stress spectrum distribution, extract the dominant frequency components from the stress spectrum distribution, set the switching frequency of each unit to a frequency value that avoids the dominant frequency components, and form a task allocation scheme based on the execution allocation relationship and the frequency value.

[0053] After identifying the areas of concentrated thermal stress, a detailed model of the heat dissipation path for each bidirectional power conversion unit is required. This involves traversing the physical connection path from the power device node to the final heat dissipation boundary for each unit. This path typically passes through multiple series of thermally conductive layers, including a chip bonding layer, a copper substrate, an insulating ceramic layer, a bottom copper layer, a thermally conductive silicone grease layer, and heat sink fins or a housing. For each thermally conductive layer in the path, its material type, thickness, and cross-sectional area parameters are extracted. The thermal resistance of each thermally conductive layer is also determined. Determined by the following formula: ,

[0054] in, For the first The thickness of the thermally conductive layer, The thermal conductivity of this layer of material, This represents the effective thermal conductivity cross-sectional area of ​​the layer. The thermal resistances of all series-connected thermally conductive layers in the path are summed sequentially to obtain the [n]th [layer]. The total thermal resistance of each unit from the chip node to the heat dissipation boundary : ,

[0055] The value directly reflects the unit's ability to dissipate internal heat to the outside. The smaller the value, the shorter the heat dissipation path and the stronger the heat dissipation capacity. The more suitable the unit is for undertaking compensation tasks with large power losses.

[0056] After obtaining the power timing curves of the compensation task to be performed, each curve is processed within one compensation cycle. Perform time-domain integration to obtain the cumulative energy generated by the compensation task within the period. : , in, For the first The compensation task is at any time Instantaneous power, To compensate for the mission's duration, energy is accumulated. As a characterization of the thermal load of the execution unit for this compensation task, it can comprehensively reflect the thermal accumulation effect of the task on the device over time, avoiding the potential neglect of thermal accumulation problems caused by continuous low-power tasks when only peak power is used as a measure. Establish the thermal load characterization of each compensation task and its corresponding value. The mapping relationship is used to form a task hot load list for subsequent cross-matching.

[0057] During the cross-matching phase, the task thermal load characterization quantity will be... Total thermal resistance of each unit Perform a combination evaluation. For any matching combination of task and unit, define its hot matching cost as... This product, in a physical sense, approximately corresponds to the steady-state temperature rise contribution of the unit when undertaking the task; a larger product indicates higher thermal stress for the unit when performing the task. Among all possible matching combinations, the optimal task allocation scheme is sought with the goal of minimizing the sum of global thermal matching costs. Specifically, let there be a total of... Compensation tasks and One usable bidirectional power conversion unit, optimize the objective function. for: , in, Indicates the first The task was assigned to the first The mapping relationship of each unit. Solving for the mapping among all valid mappings using enumeration or a heuristic search algorithm (such as the Hungarian algorithm or a greedy sorting algorithm) results in... The minimum allocation scheme determines the execution allocation relationship between each compensation task and its corresponding execution unit. This method ensures that compensation tasks with high power loss are preferentially allocated to units with lower total thermal resistance (i.e., shorter heat dissipation paths), thereby balancing the thermal pressure distribution across all units as a whole.

[0058] After task allocation, switching frequency optimization was further performed on each unit corresponding to the thermal stress concentration region. The stress tensor components of these units under thermal stress concentration conditions were extracted, including normal stress components and shear stress components along the chip length, width, and thickness directions. Discrete Fourier transforms were performed on the time series of each stress tensor component to obtain the stress spectrum distribution. Identify dominant frequency components with amplitudes significantly higher than background noise from the spectral distribution. These dominant frequencies typically correspond to the structural resonant frequencies of the unit under thermal cycling excitation or the fatigue-sensitive frequency bands of the material. When the switching frequency or harmonic components of the bidirectional power conversion unit are... When these frequencies coincide, they can trigger mechanical resonance or accelerate material fatigue damage. Therefore, the switching frequency needs to be set to a value that effectively avoids the aforementioned dominant frequency components. .

[0059] The selection principle is: under the premise of meeting the compensation current ripple requirements, to make Its harmonics do not fall into the range of Centered on, with a width of Within the avoidance frequency band, the following conditions are met: , in, For harmonic order, For the highest harmonic order that needs to be considered, The avoidance bandwidth is set around the dominant frequency component. Among the candidate frequencies that meet the above constraints, the lower frequency value that can reduce switching losses is preferred to further reduce the thermal load on the thermal stress concentration unit.

[0060] The above execution allocation relationship is compared with the adjusted switching frequency of each unit. A comprehensive task allocation scheme is formed through integrated analysis. This scheme includes two levels of information: firstly, which bidirectional power conversion unit (PCU) each compensation task is assigned to, clarifying the compensation current command content undertaken by each unit; secondly, the switching frequency settings of each unit within the thermal stress concentration area, ensuring that these units can still output compensation current normally while reducing thermal stress risks. The task allocation scheme can be dynamically corrected after each thermal state update. When a new thermal stress concentration area or unit junction temperature exceeds the threshold is detected, the heat dissipation thermal resistance calculation, thermal load matching, and frequency avoidance calculation processes are retried, outputting an updated allocation scheme and distributing it to the controllers of each PCU, thereby achieving dynamic thermal optimization management of the compensation tasks of the three-phase four-wire power system.

[0061] A second aspect of the present invention provides a three-phase four-wire natural heat dissipation power conversion system, comprising: The negative-sequence and zero-sequence decomposition unit is used to collect the phase line current, phase line voltage and neutral line current of a three-phase four-wire power system, and decompose them into negative-sequence current vector and zero-sequence current vector based on instantaneous power theory. The compensation allocation prediction unit is used to generate compensation current commands for each phase based on the negative sequence current vector and the zero sequence current vector, and to allocate the compensation task to multiple bidirectional power conversion units according to the amplitude and frequency characteristics of the compensation current commands, and to predict the power loss distribution of each bidirectional power conversion unit through the loss model. The thermal stress identification unit is used to establish a coupling model of heat conduction and natural convection based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, calculate the steady-state temperature field, and identify areas of concentrated thermal stress. The thermal stress task allocation unit is used to prioritize the allocation of compensation tasks with large power losses to bidirectional power conversion units with shorter heat dissipation paths based on the thermal stress concentration area, and to reduce the switching frequency of the units corresponding to the thermal stress concentration area, thereby forming a task allocation scheme. The compensation monitoring feedback unit is used to control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit. When the junction temperature of each unit exceeds the temperature threshold, the deviation is fed back to the task allocation stage to adjust the allocation weight.

[0062] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0063] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0064] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A three-phase four-wire natural heat dissipation power supply conversion method, characterized in that, include: The phase line current, phase line voltage and neutral line current of a three-phase four-wire power system are collected, and the negative sequence current vector and zero sequence current vector are obtained based on instantaneous power theory. The compensation current command for each phase is generated based on the negative sequence current vector and the zero sequence current vector. The compensation task is assigned to multiple bidirectional power conversion units based on the amplitude and frequency characteristics of the compensation current command. The power loss distribution of each bidirectional power conversion unit is predicted by the loss model. Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify areas of concentrated thermal stress. Based on the thermal stress concentration area, the compensation tasks with large power loss are preferentially assigned to the bidirectional power conversion unit with a shorter heat dissipation path, and the switching frequency of the unit corresponding to the thermal stress concentration area is reduced to form a task allocation scheme. Control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit, and when the junction temperature of each unit exceeds the temperature threshold, feed the deviation back to the task allocation stage to adjust the allocation weight.

2. The method according to claim 1, characterized in that, The line currents, line voltages, and neutral current of each phase in a three-phase four-wire power system are collected. Based on instantaneous power theory, the negative-sequence current vector and the zero-sequence current vector are obtained, including: The three-phase line current and neutral line current are collected synchronously using current sensors, and the three-phase line voltage is collected synchronously using voltage sensors. The three-phase line currents are filtered by a sliding window, with the window length adaptively adjusted according to the fundamental period of the three-phase line voltages, to obtain the filtered line currents. The filtered line currents and the three-phase line voltages are then transformed into a synchronous rotating coordinate system to obtain the direct-axis current component and the quadrature-axis current component. The DC component and the frequency-doubled AC component are extracted from the direct-axis current component and the quadrature-axis current component, respectively. The frequency-doubled AC component constitutes the component of the negative sequence current in the synchronous rotating coordinate system. The components of the negative sequence current in the synchronous rotating coordinate system are inversely rotated to the two-phase stationary coordinate system to obtain the orthogonal components of the negative sequence current and synthesize the negative sequence current vector. The neutral line current is used as the amplitude of the zero sequence current vector, and the phase angle of the zero sequence current vector is kept consistent with the phase of the first phase voltage in the three-phase line voltage, thus completing the decomposition of the negative sequence current vector and the zero sequence current vector.

3. The method according to claim 1, characterized in that, The compensation current command for each phase is generated based on the negative-sequence current vector and the zero-sequence current vector. The compensation task is then allocated to multiple bidirectional power conversion units based on the amplitude and frequency characteristics of the compensation current command. The power loss distribution of each bidirectional power conversion unit is predicted using a loss model, including: The negative sequence current vector and the zero sequence current vector are subjected to inverse coordinate transformation to obtain the compensation current command for each phase. The compensation current command is subjected to time-frequency analysis. The component with a time proportion exceeding a preset ratio is marked as a continuous compensation component, and the component with a peak amplitude exceeding a preset multiple is marked as a transient impact component. A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component. Steady-state thermal analysis is performed by using average power loss as the node heat source of the thermal resistance network to identify units whose steady-state temperature exceeds a preset threshold, and the spatial region formed by the unit exceeding the threshold and its adjacent units is extracted as the thermal stress concentration region. The continuous compensation component of the highest temperature element in the thermal stress concentration area is transferred to the lowest temperature element outside the area. The average power loss is updated and steady-state thermal analysis is performed again. The task transfer and steady-state thermal analysis are iteratively executed until the temperature of all elements meets the standard. Transient thermal analysis is performed by superimposing the peak power loss and the duration of the transient impact component onto the steady-state temperature, and outputting a power loss distribution that includes both average power loss and peak power loss.

4. The method according to claim 3, characterized in that, A thermal resistance network is established between each bidirectional power conversion unit. The thermal resistance value of each branch in the thermal resistance network is determined by the unit spacing and the convective heat transfer coefficient. The average power loss of each unit is calculated based on the continuous compensation component, and the peak power loss of each unit is calculated based on the transient impact component, including: Obtain the spatial coordinates of each bidirectional power conversion unit, construct a spatial connectivity graph with the bidirectional power conversion unit as the vertex, establish thermal resistance branches between adjacent bidirectional power conversion units, and calculate the unit spacing between adjacent bidirectional power conversion units. The Nusselt number is calculated based on the unit spacing and the kinematic viscosity of the ambient fluid. The convective heat transfer coefficient is calculated by combining the fluid thermal conductivity. The convective heat transfer coefficient is correlated with the unit spacing to determine the thermal resistance value of the thermal resistance branch and to establish a thermal resistance network that includes all thermal resistance branches. The continuous compensation component is sampled in the time domain, the statistical distribution of the current amplitude in the sampling sequence is extracted, the time average value of the conduction loss is determined based on the root mean square value of the current amplitude, the time average value of the switching loss is determined based on the frequency of change of the current amplitude, and the average power loss is obtained. Identify the extreme points of the current amplitude in the transient impact component, extract the current peak value and current rise slope corresponding to the extreme points, determine the transient conduction loss at the extreme points based on the current peak value, determine the transient switching loss at the extreme points based on the current rise slope, and obtain the peak power loss.

5. The method according to claim 1, characterized in that, Based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, a coupled model of heat conduction and natural convection is established to calculate the steady-state temperature field and identify regions of concentrated thermal stress, including: The internal heat source intensity of each bidirectional power conversion unit is determined based on the power loss distribution. The heat exchange path between units is determined based on the spatial layout location, and the spatial topology relationship of the path is extracted. The internal heat source intensity and the spatial topology relationship are associated to form the heat conduction control equation. The contact boundary area between the surface of each bidirectional power conversion unit and the ambient fluid is obtained according to the spatial layout location. The density difference of the ambient fluid is calculated based on the temperature difference of the contact boundary. The fluid velocity field is determined based on the buoyancy force generated in the gravitational field by the density difference. The convective heat transfer intensity of the unit surface is determined based on the fluid velocity field. The temperature field of the heat conduction control equation is used as the driving source of the fluid velocity field, and the convective heat transfer intensity is used as the heat dissipation boundary of the heat conduction control equation. The temperature field and the fluid velocity field are iteratively updated until the temperature change rate is lower than the convergence criterion. The converged temperature field is then obtained as the steady-state temperature field. The spatial gradient distribution of the steady-state temperature field within each bidirectional power conversion unit is calculated. The thermal strain field is obtained based on the spatial gradient distribution and the thermal expansion coefficient of the material. The thermal stress field is obtained based on the thermal strain field and the elastic modulus of the material. Regions in the thermal stress field where the spatial derivative exceeds the stress gradient limit are identified as thermal stress concentration regions.

6. The method according to claim 1, characterized in that, Based on the thermal stress concentration area, compensation tasks with higher power losses are preferentially assigned to bidirectional power conversion units with shorter heat dissipation paths, and the switching frequency of the units corresponding to the thermal stress concentration area is reduced, forming a task allocation scheme, including: Identify the bidirectional power conversion unit corresponding to the thermal stress concentration area, traverse the physical connection path from each unit to the heat sink, extract the material type and thickness parameters of the series heat conduction layer in the path, determine the thermal resistance of each heat conduction layer according to the material thermal conductivity and thickness parameters, and accumulate the thermal resistance of each heat conduction layer to form the total heat dissipation thermal resistance of the unit. Obtain the power time-series curve of the compensation task to be executed, perform time-domain integration on the power time-series curve to obtain the cumulative energy within the cycle, use the cumulative energy as the thermal load characterization quantity of the compensation task, and establish the correspondence between the compensation task and the thermal load characterization quantity. The thermal load characterization of the compensation task is cross-matched with the total heat dissipation thermal resistance of each unit. Among all possible matching combinations, the combination in which the product of the thermal load characterization and the total heat dissipation thermal resistance is minimized globally is found. Based on this combination, the execution allocation relationship between the compensation task and the bidirectional power conversion unit is determined. For the units corresponding to the thermal stress concentration area, extract the stress tensor components, perform Fourier transform on the stress tensor components to obtain the stress spectrum distribution, extract the dominant frequency components from the stress spectrum distribution, set the switching frequency of each unit to a frequency value that avoids the dominant frequency components, and form a task allocation scheme based on the execution allocation relationship and the frequency value.

7. A three-phase four-wire natural cooling power conversion system, used to implement the method as described in any one of claims 1-6, characterized in that, include: The negative-sequence and zero-sequence decomposition unit is used to collect the phase line current, phase line voltage and neutral line current of a three-phase four-wire power system, and decompose them into negative-sequence current vector and zero-sequence current vector based on instantaneous power theory. The compensation allocation prediction unit is used to generate compensation current commands for each phase based on the negative sequence current vector and the zero sequence current vector, and to allocate the compensation task to multiple bidirectional power conversion units according to the amplitude and frequency characteristics of the compensation current commands, and to predict the power loss distribution of each bidirectional power conversion unit through the loss model. The thermal stress identification unit is used to establish a coupling model of heat conduction and natural convection based on the power loss distribution and the spatial layout of each bidirectional power conversion unit, calculate the steady-state temperature field, and identify areas of concentrated thermal stress. The thermal stress task allocation unit is used to prioritize the allocation of compensation tasks with large power losses to bidirectional power conversion units with shorter heat dissipation paths based on the thermal stress concentration area, and to reduce the switching frequency of the units corresponding to the thermal stress concentration area, thereby forming a task allocation scheme. The compensation monitoring feedback unit is used to control the output compensation current of each bidirectional power conversion unit, continuously monitor the balance of the three-phase current after compensation and the junction temperature of each unit. When the junction temperature of each unit exceeds the temperature threshold, the deviation is fed back to the task allocation stage to adjust the allocation weight.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images