Power distribution cabinet heat dissipation control method and system based on phase change heat pipe

Abstract

The application discloses a power distribution cabinet heat dissipation control method and system based on a phase change heat pipe, relates to the technical field of data processing, and comprises the following steps: acquiring current and multi-dimensional temperature data of a power distribution cabinet, dividing intervals according to load rate boundaries, generating load sequences and thermal dynamic sequences of the intervals, extracting fluctuation heat integrals and thermal hysteresis gradients of the intervals respectively, calculating a dynamic current reference value in combination with external environment temperature, obtaining load dispersion and electric heat drift degree based on relative error, and then acquiring a thermal steady-state index; if steady-state verification is passed, solving an electric heat differential equation in combination with expected load, acquiring a predicted temperature rise rate and temperature rise value, finally generating a target duty cycle, and outputting a control instruction at a feedforward trigger moment after deducting a pneumatic dead zone time. The application has the advantages of heat perception, self-adaptive anti-interference and hysteresis feedforward compensation.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a heat dissipation control method and system for power distribution cabinets based on phase change heat pipes. Background Technology

[0002] With the increasingly complex temporal fluctuations in modern industrial production loads, the heating conditions inside electrical distribution cabinets exhibit strong nonlinear and transient characteristics. Currently, phase change heat pipes, with their highly efficient latent heat transfer capabilities, have been widely introduced into electrical distribution cabinets for passive heat conduction. However, the heat dissipation control mechanism based on the coupling of phase change heat pipes and active fans still faces challenges in its underlying thermodynamic logic and control closed-loop.

[0003] First, most existing monitoring methods rely directly on real-time temperature feedback or simple linear current differences to assess load status, ignoring the nonlinear physical nature of electrical heating, which strictly follows Joule's law and is proportional to the square of the current. This makes it difficult for the system to accurately quantify the fluctuating heat energy injected into the cabinet by high-frequency micro-load surges, and easily underestimates the severe thermal shock caused by heavy-load transient jumps. Second, existing technologies generally use pre-set fixed static reference thresholds for data quality verification or abnormal state judgment, without considering the physical property that the resistivity of metal materials such as copper conductors in the main circuit increases significantly with increasing ambient temperature. This lack of an environmental temperature adaptive compensation static reference leads to the system misinterpreting the natural drift of the material's background impedance as abnormal fluctuations during high summer temperatures or poor heat dissipation in the computer room, causing the control logic to deviate significantly from the actual electrothermal coupling evolution trajectory. Finally, in the control closed loop of the heat dissipation execution stage, existing feedforward or feedback strategies ideally assume that the electrical regulation of the fan and the establishment of the cold flow field are instantaneously synchronized, completely ignoring the aerodynamic dead zone time in fluid dynamics from the time the fan receives the duty cycle drive command to the time it overcomes the internal airflow resistance and finally establishes a globally stable circulating cooling flow field (especially the heat exchange flow field between the fins on the condenser side of the phase change heat pipe). This drawback of severe disconnect between electromechanical control and fluid physics causes the system to often start increasing the fan speed only when the expected high load actually breaks out. At this time, the slowly rising cold air flow cannot absorb the large amount of nonlinear heat energy released instantaneously in time, causing the transient temperature rise peak of the heat source node to easily break through the safe operating limit and cause irreversible thermal stress damage to the components. Summary of the Invention

[0004] In response to the technical problems described in the background art, the present invention provides a heat dissipation control method and system for power distribution cabinets based on phase change heat pipes. Through zoning processing, nonlinear heat integration, environmental adaptive reference correction and feedforward control, the heat dissipation of the power distribution cabinet can be predicted and intervened in advance.

[0005] A heat dissipation control method for a power distribution cabinet based on a phase change heat pipe includes: acquiring the effective current value, internal node temperature value, and external ambient temperature value of the power distribution cabinet within a reference period; classifying the effective current value and the internal node temperature value into a first load interval, a second load interval, or a third load interval according to a preset load rate boundary; generating a load sequence and a thermal dynamic sequence for each interval by combining them in the order of sampling time; acquiring a first fluctuating heat integral, a second fluctuating heat integral, and a third fluctuating heat integral based on the load sequence for each interval; acquiring a first thermal hysteresis gradient, a second thermal hysteresis gradient, and a third thermal hysteresis gradient based on the thermal dynamic sequence for each interval; calculating a dynamic current reference value based on a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value; and calculating the dynamic current reference value based on the fluctuating heat integral and the corresponding dynamic current for each interval. The load dispersion is obtained by the relative error of the reference value. The electrothermal drift is obtained by the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value. The thermal steady-state index is obtained by the load dispersion and the electrothermal drift. If the thermal steady-state index is less than the preset steady-state threshold, the expected load current of the target prediction period is obtained. The preset electrothermal differential equation is solved by the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, and the predicted temperature rise rate and the predicted temperature rise value are obtained. The target duty cycle is obtained by the predicted temperature rise rate, the predicted temperature rise value and the preset base duty cycle. The feedforward trigger time is obtained by subtracting the preset aerodynamic dead zone from the start time of the target prediction period. At the feedforward trigger time, a control command containing the target duty cycle is output to drive the cooling fan on the condenser side of the corresponding phase change heat pipe.

[0006] Optionally, based on preset load factor boundaries, the effective current value and the internal node temperature value are respectively assigned to a first load interval, a second load interval, or a third load interval, and combined in chronological order of sampling time to generate a load sequence and a thermal dynamic sequence for each interval, including: obtaining the rated current of the distribution cabinet; dividing the obtained effective current value by the rated current to obtain the real-time load factor; assigning the real-time load factor to the corresponding first load interval, second load interval, or third load interval according to three non-overlapping preset load factor percentage ranges; extracting all effective current values ​​within the same load interval and arranging them in chronological order of sampling time to form the load sequence; extracting all internal node temperature values ​​within the same load interval and arranging them in chronological order of sampling time to form the thermal dynamic sequence.

[0007] Optionally, obtaining the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral according to the load sequence of each interval includes: within the load sequence of each interval, obtaining the effective current value of two adjacent sampling time points, and calculating the absolute value of the square difference of the effective current value of the two adjacent sampling time points; multiplying the absolute value of the square difference, a preset equivalent resistance parameter, and a preset sampling time interval to obtain the fluctuating heat energy increment; and summing all the fluctuating heat energy increments within the same load sequence to obtain the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral, respectively.

[0008] Optionally, obtaining the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient according to the thermal dynamic sequence of each interval includes: calculating the temperature difference between two adjacent sampling time points within the thermal dynamic sequence of each interval, and extracting the temperature rise difference with a value greater than zero; obtaining a time decay weight with a preset decay base and an exponent of the number of time steps between the current sampling time point and the latest sampling time point; multiplying the temperature rise difference with the corresponding time decay weight and summing the results to obtain the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient, respectively.

[0009] Optionally, a dynamic current reference value is calculated based on a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value, including: obtaining a preset standard test temperature; subtracting the standard test temperature from the external ambient temperature to obtain a temperature deviation; multiplying the temperature deviation by the temperature coefficient of resistance to obtain a drift compensation ratio; and performing a positive multiplication correction by combining the initial reference value with the drift compensation ratio to obtain the dynamic current reference value.

[0010] Optionally, the load dispersion is obtained based on the relative error between the fluctuating heat integral of each interval and the corresponding dynamic current reference value; the electrothermal drift is obtained based on the relative error between the thermal hysteresis gradient of each interval and a preset dynamic thermistor reference value; and the thermal steady-state index is obtained based on the load dispersion and the electrothermal drift. This includes: calculating the absolute difference between the fluctuating heat integral of each interval and the corresponding dynamic current reference value; dividing the absolute difference by the dynamic current reference value to obtain a first error ratio for each interval; and extracting the maximum value among all first error ratios as the load dispersion; calculating the absolute difference between the thermal hysteresis gradient of each interval and the corresponding dynamic thermistor reference value; dividing the absolute difference by the dynamic thermistor reference value to obtain a second error ratio for each interval; and extracting the maximum value among all second error ratios as the electrothermal drift; multiplying the load dispersion by a preset first weight; multiplying the electrothermal drift by a preset second weight; and adding the products of the two to obtain the thermal steady-state index.

[0011] Optionally, based on the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, a preset electrothermal differential equation is solved to obtain the predicted temperature rise rate and the predicted temperature rise value, including: superimposing the expected load current with the fluctuating heat integral corresponding to the load interval to reconstruct and obtain the predicted load sequence; obtaining the electrothermal differential equation with state coefficients including a preset heating component mass and a preset specific heat capacity parameter, and the state variable being the derivative of the temperature rise with respect to time; performing a quadratic operation on the predicted load sequence and multiplying it by a preset resistance to obtain the predicted heating power; and converting the target thermal hysteresis gradient... The parameter is transformed into (this parameter integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions of the phase change heat pipe), and the convective heat dissipation power term in the electrothermal differential equation is corrected according to (this parameter integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions). The predicted heating power and the corrected convective heat dissipation power term are substituted into the electrothermal differential equation for discretization and solution. The maximum value of the time derivative of the temperature rise during the target prediction period is extracted as the predicted temperature rise rate, and the highest temperature during the solution process is extracted as the predicted temperature rise value.

[0012] Optionally, obtaining the target duty cycle based on the predicted temperature rise rate, the predicted temperature rise value, and a preset base duty cycle includes: obtaining a preset safe temperature upper limit, a preset proportional control gain, and a preset differential feedforward gain; subtracting the safe temperature upper limit from the predicted temperature rise value to obtain the over-limit temperature difference; multiplying the over-limit temperature difference by the proportional control gain to obtain a first control component; multiplying the predicted temperature rise rate by the differential feedforward gain to obtain a second control component; and adding the base duty cycle, the first control component, and the second control component to obtain the target duty cycle.

[0013] A heat dissipation control system for a power distribution cabinet based on a phase change heat pipe is also provided. The system includes: a data acquisition module, used to acquire the effective current value, internal node temperature value, and external ambient temperature value of the power distribution cabinet within a reference period; according to a preset load rate boundary, the effective current value and the internal node temperature value are respectively assigned to a first load interval, a second load interval, or a third load interval, and combined in the order of sampling time to generate a load sequence and a thermal dynamic sequence for each interval; a first processing module, used to acquire a first fluctuating heat integral, a second fluctuating heat integral, and a third fluctuating heat integral according to the load sequence of each interval; and to acquire a first thermal hysteresis gradient, a second thermal hysteresis gradient, and a third thermal hysteresis gradient according to the thermal dynamic sequence of each interval; a second processing module, used to calculate and acquire a dynamic current reference value according to a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value; and to calculate and acquire a dynamic current reference value according to the fluctuating heat... The load dispersion is obtained by integrating the relative error between the integral and the corresponding dynamic current reference value; the electrothermal drift is obtained by the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value; the thermal steady-state index is obtained by the load dispersion and the electrothermal drift; the predictive control module is used to obtain the expected load current of the target prediction period when the thermal steady-state index is less than the preset steady-state threshold; the preset electrothermal differential equation is solved according to the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, to obtain the predicted temperature rise rate and the predicted temperature rise value; the target duty cycle is obtained according to the predicted temperature rise rate, the predicted temperature rise value and the preset base duty cycle; the feedforward trigger time is obtained by subtracting the preset aerodynamic dead zone from the start time of the target prediction period, and a control command containing the target duty cycle is output at the feedforward trigger time to drive the cooling fan on the condenser side of the corresponding phase change heat pipe.

[0014] Optionally, the data acquisition module is further configured to: acquire the rated current of the distribution cabinet; divide the acquired effective value of the current by the rated current to obtain the real-time load rate; classify the real-time load rate into the corresponding first load interval, second load interval, or third load interval according to three non-overlapping preset load rate percentage ranges; extract all effective values ​​of the current within the same load interval and arrange them in chronological order of sampling time to form the load sequence; extract all internal node temperature values ​​within the same load interval and arrange them in chronological order of sampling time to form the thermal dynamic sequence.

[0015] The beneficial effects of this invention are reflected in: In the overall heat dissipation control method for distribution cabinets based on phase change heat pipes, firstly, physical intervals are divided according to the real-time load rate, effectively isolating time-series data under different heating conditions, avoiding the averaging interference of low-frequency fluctuations on the heavy-load thermal model, and providing a low-level data source with clear physical boundaries. Secondly, by employing nonlinear Joule energy flow integrals and thermal hysteresis gradients, based on the physical law of current square-proportional heating, the quantification accuracy of micro-load sudden change energy and the hysteresis effect of heat accumulation in the cabinet are improved, thus enhancing the quantification accuracy of the destructive properties of micro-load sudden changes. Furthermore, an ambient temperature adaptive reference drift equation is introduced to dynamically compensate for the increase in natural impedance of copper conductors due to rising room temperature, filtering out false fault signals caused by seasonal changes or environmental variations, and improving the accuracy of judgment indicators. Finally, the aerodynamic dead time feedforward compensation is integrated into the closed loop execution, enabling the central controller to increase the fan duty cycle in advance before the expected high load arrives, use the time difference to pre-emptively vent static hot air and establish a negative pressure cooling field at the peak flow rate, thereby timely absorbing the transient temperature rise peak when Joule heat is concentrated, achieving a control mode conversion from passive following cooling to active forward pre-cooling. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 This is a schematic diagram illustrating the steps of the heat dissipation control method for a power distribution cabinet based on a phase change heat pipe according to the present invention. Figure 2 This is a schematic diagram of a portion of step S1 in the heat dissipation control method for a power distribution cabinet based on a phase change heat pipe according to the present invention. Figure 3 This is a schematic diagram of part of step S2 in the heat dissipation control method for power distribution cabinet based on phase change heat pipe of the present invention; Figure 4 This is a schematic diagram of part of step S3 in the heat dissipation control method for power distribution cabinet based on phase change heat pipe of the present invention; Figure 5 This is a schematic diagram of part of step S4 in the heat dissipation control method for power distribution cabinet based on phase change heat pipe of the present invention. Detailed Implementation

[0018] 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0020] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] This invention provides a heat dissipation control method for electrical distribution cabinets based on phase change heat pipes, such as... Figure 1 As shown, in one embodiment, the method includes the following steps and sub-steps: S1. Divide the first load interval, the second load interval, and the third load interval into the order of increasing load intensity within the operating cycle of the distribution cabinet, and obtain the load characteristic sequence and thermal dynamic sequence of the first load interval, the second load interval, and the third load interval within the reference cycle preceding the current heat dissipation control cycle.

[0022] S2. Obtain the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral according to the load characteristic sequences of the first load interval, the second load interval, and the third load interval, respectively. Obtain the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient according to the thermal dynamic sequences of the first load interval, the second load interval, and the third load interval, respectively.

[0023] S3. Obtain the load dispersion based on the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral, and obtain the electrothermal drift based on the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient, and obtain the thermal steady-state index based on the load dispersion and the electrothermal drift.

[0024] S4. If the thermal steady-state index is less than the preset stability threshold, the expected load level for the target prediction period is obtained, the predicted temperature rise data of the area to be cooled is output, and advance cooling control is executed.

[0025] In this embodiment, it should be noted that in S1, the nonlinear heating condition inside the electrical equipment is decoupled by dividing the load intensity into intervals, and underlying basic data with spatial and temporal correspondence is acquired simultaneously. The main heating elements inside the distribution cabinet are conductive media such as copper busbars and circuit breaker contacts. In this embodiment, a phase change heat pipe assembly is configured inside the distribution cabinet, with its evaporation end attached to the aforementioned heating element and its condensation end placed in the heat dissipation duct where forced convection heat transfer is performed by an adjustable-speed cooling fan. The Joule heat generated by the heating element does not increase linearly with the current, but is proportional to the square of the effective value of the current. This requires that the full load condition cannot be regarded as a single homogeneous state for monitoring.

[0026] In practice, the first step is to read the rated operating current parameters of the distribution cabinet, such as setting the rated current of the main circuit of the distribution cabinet. A. The central controller calculates the real-time load rate based on the real-time collected RMS current value. The operating state is divided into three independent intervals: when the sampling current... At time A, the load factor ,fall into In the first load range, the conductor's heating power is low, and heat dissipation is mainly through metal conduction; when the sampling current rises to... At time A, the load factor ,Enter In the second load zone, the internal air expands due to heat, generating natural convection; when the sampling current reaches... At time A, the load factor Belonging In the third load range, the skin effect increases, and the heat accumulation rate exceeds the natural heat dissipation capacity. Based on the above physical range boundaries, the data acquisition module operates continuously at a high-frequency sampling rate of 1Hz within a 24-hour reference period. It strictly assigns the current values ​​acquired at different times, as well as the temperature values ​​acquired by the PT100 temperature sensors attached to the busbar nodes, into the corresponding storage matrix according to the calculated real-time load rate.

[0027] This computational logic, which rigidly divides the load characteristic sequences into three parts based on load rate, overcomes the problems of mixed data across all time periods and neglect of thermodynamic state differences under different loads in existing technologies. Its beneficial effect lies in providing a set of input sources with clear physical boundaries for subsequent data processing, enabling targeted physical mapping of time-series data under different thermal stress states. This avoids averaging interference caused by light-load temperature rise data on heavy-load thermal models and improves the accuracy of data tracing.

[0028] In S2, the true thermodynamic evolution characteristics are extracted from the discrete time series. Taking the calculation of the first wave heat integral as an example, its formula includes the product of the absolute value of the squared difference of the effective current values ​​at two adjacent sampling time points, the equivalent resistance, and the time interval. Assume the equivalent contact resistance of the main circuit... Sampling time interval s. Under light load conditions, the current fluctuates from 200A to 250A, and the current difference is... A, the increase in fluctuating thermal energy produced by substituting into the formula is: J; Under heavy load conditions, the current fluctuates from 800A to 850A, and the current difference is... A, but the fluctuation in thermal energy increment is J. This calculation logic objectively reveals that the destructive energy generated by the same current jump under different load bases varies by multiples, overcoming the deficiency of existing technologies that only calculate the current difference, leading to an underestimation of the thermal shock under heavy loads. Simultaneously, the thermal hysteresis gradient is calculated according to the formula. This logic assigns a time-decreasing weighting factor to the positive temperature rise difference between adjacent temperature sampling points. For example, if three consecutive temperature nodes of 36.0℃, 36.5℃, and 36.8℃ are obtained, the 0.5℃ temperature difference between the first two is relatively far from the current time, thus assigning an exponential decay base. The weighting is 0.45℃; the 0.3℃ temperature difference between the latter two occurs at present, assigning... A weight of 1 yields a temperature of 0.3℃, and the sum of these two values ​​results in a hysteresis gradient of 0.75℃. This computational logic, which incorporates memory-forgetting characteristics, reflects the cumulative hysteresis effect of heat caused by the specific heat capacity of the internal air and metal. The beneficial effect of S2 is that it transforms abstract monitoring data into physical indicators that quantify actual destructive energy and thermal hysteresis states, increasing the accuracy of extracting microscopic load mutation characteristics and providing solid parameter support for subsequently establishing highly realistic thermodynamic differential equations.

[0029] In S3, the reliability of the current electromechanical data link is verified, and false alarms caused by seasonal changes or changes in the data center environment are prevented. The resistivity of metallic conductors naturally increases with the rise in ambient temperature. If a static threshold is used, the normal increase in impedance due to high summer temperatures will be misjudged as equipment aging or abnormal load.

[0030] Assuming the initial standard reference value calibrated in the non-volatile memory under the standard test condition of 25°C J, the temperature coefficient of resistance of the copper conductor in the main circuit is set to... The current temperature sensor reading in the computer room is... At ℃, the parameters are substituted into the drift equation to calculate the dynamic current reference value. J.

[0031] Subsequently, the fluctuating heat integral at the current moment was calculated as 439.2 J, and compared with the corrected dynamic current reference value of 415.72 J to obtain the deviation. The maximum deviation in each interval is extracted as the load dispersion. Similarly, the electrothermal drift, which characterizes the dispersion of the temperature response, is obtained. and the two are arranged according to and The thermal steady-state exponent is output by weighted summation of the weight constants. When the calculated synergy index of 0.07176 is lower than the set stability threshold of 0.35, the current feedback signal is confirmed to be valid. This calculation logic overcomes the problems of rigid thresholds and lack of environmental compensation in existing control mechanisms, which lead to misjudgments under different climatic conditions. Its beneficial effects are reflected in giving the control the ability to adapt to external environmental temperature interference, ensuring that the captured data can objectively reflect the electrothermal coupling state of the equipment itself, filtering out false fault signals at the logical level, and providing a stable basis for subsequent predictive control actions.

[0032] In S4, the thermal dead time caused by aerodynamic delay is overcome. When an external dispatch signal is received indicating that the expected load current will jump to [a certain value] at 14:30... Following A's instructions, construct the electrothermal energy balance differential equation and substitute the equivalent mass... kg, comprehensive specific heat capacity Parameters such as J / (kg·℃) are used, and the thermal hysteresis gradient obtained from S2 is incorporated as a dynamic thermal resistance correction term. Through numerical discretization, the expected thermal resistance during the target window is calculated. The predicted temperature rise rate in °C / min, and The predicted temperature rise in °C. Based on the control law formula, in... Based on the base duty cycle, a proportional gain and differential feedforward compensation are superimposed to calculate the target duty cycle value that needs to be increased to... At this point, to address the issue of asynchronous establishment of electrical signals and cold flow fields in existing technologies, a pre-stored [data / mechanism] was introduced. The aerodynamic dead time parameter is minimized. The central controller will not wait for the load to arrive at 14:30 before increasing the speed, but will instead move the execution time forward, issuing a 60.57% duty cycle command to the DC brushless variable speed fan at 14:27, a feedforward advance.

[0033] This proactive calculation and execution logic overcomes the problem of cold air flow lagging behind heat generation speed caused by physical processes such as overcoming duct resistance and impeller acceleration in cooling fans. Its beneficial effect lies in utilizing the time difference to expel the static hot air accumulated in the condenser-side duct of the phase change heat pipe in advance. This ensures that when the nonlinear Joule heat bursts at 14:30, a stable cold air circulation field at peak flow velocity has already been formed inside. The initially established negative pressure cooling field absorbs the transient temperature rise peak of the heat source node, realizing a shift from passively following cooling to actively anticipating pre-cooling control, thus ensuring the stable operation of electrical components.

[0034] In summary, the heat dissipation control method for the distribution cabinet based on phase change heat pipes firstly divides the physical intervals according to the real-time load rate, effectively isolating time-series data under different heating conditions, avoiding the averaging interference of low-frequency fluctuations on the heavy-load thermal model, and providing a low-level data source with clear physical boundaries. Secondly, by employing nonlinear Joule energy flow integrals and thermal hysteresis gradients, based on the physical law of current square-proportional heating, the quantification accuracy of micro-load sudden change energy and the hysteresis effect of heat accumulation in the cabinet are improved, thus enhancing the quantification accuracy of the destructive properties of micro-load sudden changes. Furthermore, an adaptive ambient temperature reference drift equation is introduced to dynamically compensate for the increased natural impedance of copper conductors caused by rising room temperature, filtering out false fault signals caused by seasonal changes or environmental variations, and improving the accuracy of judgment indicators. Finally, the aerodynamic dead time feedforward compensation is integrated into the closed loop execution, enabling the central controller to increase the fan duty cycle in advance before the expected high load arrives, use the time difference to pre-emptively vent static hot air and establish a negative pressure cooling field at the peak flow rate, thereby timely absorbing the transient temperature rise peak when Joule heat is concentrated, achieving a control mode conversion from passive following cooling to active forward pre-cooling.

[0035] like Figure 2 As shown, in one embodiment, S1 includes: S11, establishing the rated operating current of the distribution cabinet, setting a load rate threshold according to three preset non-overlapping percentage ranges, and dividing the first load range, the second load range and the third load range.

[0036] S12. The main circuit current effective value, the temperature value of key nodes in the distribution cabinet and the ambient temperature value of the computer room outside the distribution cabinet are collected synchronously through the data acquisition module within the reference period.

[0037] S13. Divide the collected raw time series data into the corresponding load intervals according to the boundaries established in S11, extract all current sampling values ​​of the corresponding intervals to form a load feature sequence in chronological order, and extract all node temperature sampling values ​​of the corresponding intervals to form a thermal dynamic sequence in chronological order.

[0038] In this embodiment, it should be noted that the heat generated inside the distribution cabinet follows Joule's law, and the heat generated is proportional to the square of the current. Since the thermal shock cannot be accurately measured using linear differences, a load-rate-based partitioning method must be used to separate the nonlinear thermodynamic states.

[0039] Furthermore, the real-time load factor of the distribution cabinet during operation is defined as the ratio of the effective value of the main circuit current collected by the current transformer at the current moment to the rated operating current of the distribution cabinet, expressed as a percentage. Based on the magnitude of the real-time load factor value within the operating cycle, the following three intervals are defined: The first load range is defined as a real-time load rate greater than or equal to zero and less than 30%. Under this condition, the conductor's heat generation power is low. The second load range is defined as a real-time load rate greater than or equal to 30% and less than 75%. Under this condition, the heat generation power increases, and the internal air is heated, forming a natural convection channel. The third load range is defined as a real-time load rate greater than or equal to 75% and less than or equal to 100%. Under this condition, the skin effect of the conductor's contact resistance increases, and the heat accumulation rate exceeds the extreme value of natural convection heat dissipation. The boundary values ​​of the three load rate ranges were determined by analyzing the correspondence between load rate and temperature rise rate in the historical operating data of the distribution cabinet. Specifically, the internal node temperature rise rate of the distribution cabinet under different load rates was continuously collected over a month, and a scatter plot of load rate-temperature rise rate was plotted. It was found that when the load rate was below 30%, the temperature rise rate was slow and below 0.5℃ / min. In the range of 30% to 75%, the temperature rise rate increased linearly with the load rate to 2℃ / min. After the load rate was above 75%, the temperature rise rate rose sharply to more than 3℃ / min. Therefore, 30% and 75% were used as the physical boundaries to distinguish between light load, medium load and heavy load conditions.

[0040] Furthermore, the central controller collects data at a fixed sampling frequency within a complete 24-hour reference cycle before the current heat dissipation control cycle, and stores the separated data into the load characteristic sequence and thermal dynamic sequence of each interval.

[0041] In this embodiment, it should also be noted that in S11, by establishing the rated operating current of the distribution cabinet and setting a load rate threshold according to a preset non-overlapping percentage range, the continuous operating time axis is discretely divided into physical states. The main heat-generating components inside the distribution cabinet are conductive media such as copper busbars and circuit breaker contacts. The Joule heat generated by these components follows the basic laws of physics, namely, the heat generated is proportional to the square of the effective value of the current. This nonlinear characteristic requires that the control cannot treat all load conditions as a single homogeneous state for monitoring.

[0042] In practice, the rated operating current parameter marked on the nameplate of the distribution cabinet is read first. For example, the rated current of the main circuit of the distribution cabinet is set to 1000A. Then, the central controller calculates the real-time load rate based on the real-time effective value of the main circuit current. The calculation formula is the real-time current divided by the rated current and multiplied by 100%. The operating status is divided into three independent and non-overlapping intervals: When the sampling current acquired by the data acquisition module is 200A, the calculated load rate is 20%, which falls within the threshold of 0% <= load rate < 30%, indicating that the system is currently in the first load interval. At this time, the heat dissipation power of the conductor is low, and the heat loss in the cabinet is mainly due to the conduction of the metal components themselves, with air convection on the verge of stagnation. When the sampling current climbs to 400A, the load rate reaches 40%, entering the second load interval of 30% <= load rate < 75%. At this time, the heat dissipation power increases, and the internal air expands due to heat, generating natural convection circulation, and the heat exchange process begins to tend towards dynamic equilibrium. When the sampling current reaches as high as 850A, the load rate jumps to 85%, belonging to the third load interval of 75% <= load rate <= 100%. Under this high load condition, the skin effect of the conductor contact resistance increases, and the heat accumulation rate exceeds the limit of natural heat dissipation capacity, posing a high risk of local overheating.

[0043] In S12, the hardware acquisition module achieves strict alignment of the effective current value and multi-dimensional temperature values ​​on the time axis under a preset high-frequency sampling period. Due to the time lag in thermodynamic conduction, if there is a phase difference between the acquisition of electrical and thermal signals, the subsequent electrothermal coupling model will lose the accuracy of causal correlation.

[0044] In practice, the system includes a high-precision current transformer, PT100 contact temperature sensors attached to the easily heated nodes of the busbar, and an ambient temperature sensor deployed in the external equipment room environment of the distribution cabinet. The central controller is set to a uniform sampling frequency, such as 1Hz, which triggers a global synchronization sampling command every 1 second. Within a set 24-hour reference period, 86,400 discrete data segments are generated. Taking 14:00:01 as an example, within the same microsecond-level timestamp, the effective value of the main circuit current is synchronously locked at 450A, the busbar node temperature inside the cabinet is 45.2℃, and the ambient temperature outside the equipment room is 32.5℃, and this set of data is packaged into an indivisible time tuple. This synchronous and continuous acquisition of full-time data calculation and execution logic solves the clock drift problem existing in previous decentralized acquisition and avoids false causal inferences caused by the misalignment of current peaks and temperature peaks in the time series.

[0045] In S13, these unordered data tuples are transformed into characteristic sequences that can be independently accessed by each interval. The core logic lies in performing data routing and reorganization based on load factor attribution conditions. Each set of data tuples generated by the data acquisition module within the 24-hour reference cycle enters the buffer memory of the central controller. Based on the interval boundaries established in S11, the current values ​​in each data tuple are examined one by one.

[0046] Taking the previously acquired data as an example, if the current at timestamp t1 is 250A and the calculated load rate is 25%, then the 250A current value at that moment is appended to the end of the first load characteristic sequence, and the node temperature of 38.5℃ at that moment is appended to the end of the corresponding first thermal dynamic sequence. If the current at timestamp t2 suddenly increases to 800A and the calculated load rate is 80%, then the 800A current value and the corresponding temperature value of 48.1℃ are routed and appended to the third load characteristic sequence and the third thermal dynamic sequence, respectively. After 24 hours of classification and reorganization, the originally continuous total data stream is broken down into three sets of parallel characteristic sequences. The data elements within each set of sequences maintain the original chronological order, but data nodes that do not belong to the load intensity range are removed. This logic, which forcibly divides the original time-series data according to boundary conditions and forms independent sequences, solves the problem that existing algorithms are easily disturbed by extreme outliers when processing non-stationary time series. It achieves data purification at the physical state level and ensures that, for example, each temperature rise record in the third thermal dynamic sequence occurs when the equipment is indeed under heavy load and high thermal stress, thereby eliminating the interference of the equipment's temperature naturally decreasing during standby or light load cooling periods.

[0047] like Figure 3 As shown, in one embodiment, S2 includes: S21, independently extracting the absolute value of the square difference of the effective current values ​​of two adjacent sampling time points in each load interval, summing them together with the equivalent impedance of the main circuit and the time interval parameter, and calculating the fluctuating heat integral characterizing the nonlinear heating characteristics.

[0048] S22. Independently calculate the temperature difference between two adjacent sampling time points within each load interval, extract the temperature difference values ​​that are greater than zero, multiply them with the time decay weighting factor, and then sum them up to calculate the thermal hysteresis gradient that characterizes the heat accumulation effect.

[0049] In this embodiment, it should be noted that when obtaining the first wave heat integral, a nonlinear Joule energy flow integral algorithm is used instead of the conventional data difference algorithm, which can be specifically expressed as follows:

[0050] In the above expression, The first fluctuating heat integral representing the first load range, in joules; The total number of valid sampling points representing the first load characteristic sequence; Represents the sequential number of the sampling points; Represents the first load characteristic sequence of the first The effective value of the current at each sampling time point, in amperes; Represents the first load characteristic sequence of the first The effective value of the current at each sampling time point, in amperes; The estimated constant value of the equivalent contact resistance of the main circuit of the distribution cabinet, in ohms; This represents the time interval between two adjacent sampling points, in seconds. The equivalent resistance parameter is also included. The voltage drop was measured using the DC voltage drop method under power outage conditions. The specific procedure was as follows: disconnect the load of the main circuit of the distribution cabinet and inject a known DC current from the incoming line. A. Measure the voltage drop between the incoming and outgoing terminals. V, calculated according to Ohm's law Ω, to eliminate the randomness of contact resistance, the average value is taken as the final preset value after five repeated measurements. The sampling time interval is specified in the original text. Based on the thermal time constant and load fluctuation spectrum analysis of the distribution cabinet, the following analysis was conducted: First, a step response experiment showed that the time required for the cabinet temperature to rise to 63% of its steady state was approximately 300 seconds, indicating a thermal time constant of approximately 300 seconds. Second, Fourier analysis of historical load data revealed that the main fluctuation periods were concentrated between 120 and 600 seconds. Considering both the sampling theorem and the amount of data, the following parameters were selected. As a fixed interval, it can capture major load fluctuations without generating too much redundant data.

[0051] The calculation process of the above expression directly quantifies the actual fluctuating heat energy injected into the distribution cabinet by micro-load mutations within the current interval. Using the same algorithm, the second fluctuating heat integral for the second load interval and the third fluctuating heat integral for the third load interval are obtained respectively.

[0052] For the calculation of the first thermal hysteresis gradient, the central controller extracts the temperature difference between two adjacent sampling time points, filters out negative differences caused by cooling using a comparison function, and retains only the temperature rise difference representing heat accumulation. Then, the retained temperature rise difference is multiplied by a time decay weighting factor. The base of this time decay weighting factor is 0.90. Specifically, the base of the time decay weighting factor is calibrated through thermal hysteresis characteristic experiments. A step heating power is applied to the nodes inside the distribution cabinet, for example, the load current is increased from 200A to 800A, and the complete curve of the node temperature rising from the initial value to the new steady state is recorded. An exponential decay model is used. Fitting the data yields the thermal time constant. ;Will Convert to decay base in discrete time step (60s step) Thus, the attenuation weight is determined as follows. ,in The time step is the number of time steps from the current sampling point to the end of the reference period. All product results are summed. The closer the time is to the current control period, the greater the hysteresis effect of the resulting temperature rise on the subsequent state, and the larger the weight is assigned. Following this process, the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient are obtained sequentially.

[0053] In this embodiment, it should also be noted that in S21, a nonlinear Joule energy flow integral algorithm is used to extract the fluctuating heat integral. This logic is based on the core law of electrical heating, namely, Joule heat is proportional to the square of the current. The calculation formula includes the product of the absolute value of the square difference of the effective current values ​​at two adjacent sampling time points, the equivalent resistance, and the time interval. Specifically, the expression for obtaining the first fluctuating heat integral is as follows: .

[0054] Assuming the equivalent contact resistance of the main circuit of the distribution cabinet Ω is 0.0002Ω, sampling time interval For 60 seconds; in a sample containing 3 sampling points ( Within the tiny observation window, the sequence of effective current values ​​recorded in the first load interval is as follows: Time 200A, Time 250A, Time 220A.

[0055] First, calculate arrive Fluctuating thermal energy ( The current jumps from 200A to 250A. Instead of calculating the simple difference (50A), the absolute value of the squared difference is calculated and multiplied by the resistance and time: 270J. This means that an additional 270J of thermal shock energy was injected during the rising edge fluctuation of this 60s.

[0056] Then, calculate arrive Fluctuating thermal energy ( The current dropped from 250A to 220A. Continue calculating the absolute value of the difference of squares: 169.2J. Through the effect of absolute value, even if the current is decreasing, due to the existence of thermal inertia, this high-frequency sharp drop will still generate strong thermal cyclic stress on the electrical joint. The absolute value also extracts the destructive force of this reverse fluctuation as a positive thermal energy integral.

[0057] Finally, the integration results 439.2J.

[0058] In summary, if we use linear interpolation (only calculating...) according to existing conventional techniques... So, the difference between 200A and 250A is 50A; and if the device is under heavy load, fluctuating from 800A to 850A, the difference is also 50A; conventional algorithms would consider these two fluctuations to pose the same threat; but the real physical world follows Joule's law (…). Substituting into the nonlinear squared difference formula of this invention, the fluctuation thermal energy increment under heavy load is: 990J. Therefore, the same 50A current jump generates a thermal shock of 270J under light load and 990J under heavy load (a difference of nearly 4 times). This algorithm, through the accumulation of squared differences, can more accurately reflect the Joule heating law, enabling the central controller to accurately quantify and sense the real destructive force brought about by micro-load changes under different base backgrounds, and provide accurate feedforward control parameters for subsequent applications.

[0059] In S22, the thermal hysteresis gradient is obtained by extracting the positive temperature rise difference and combining it with a time decay weighting factor. The metallic conductor and the air inside the cabinet have objective specific heat capacities, and there is a time difference between heat generation and temperature rise. A time decay factor with a base of, for example, 0.90 is set in the calculation formula.

[0060] Taking three consecutive temperature sampling points of 36.0℃, 36.5℃, and 36.8℃ within a certain interval as an example, for the first two time points, the difference is calculated to be 0.5℃, which is retained because it is greater than zero. Assuming that the time span between this time point and the latest sampling point is one step, the assigned decay weight is 0.90^1 = 0.90, and the weighted temperature rise is 0.45℃. For the latter two time points, the temperature difference is 0.3℃, and the time span is 0 steps, so the assigned decay weight is 0.90^0 = 1, and the weighted temperature rise is 0.3℃. Summing up all the processed weighted values ​​in the sequence yields a hysteresis gradient of 0.75℃. The logic extracts the temperature difference greater than zero to filter out the cooling process and focus on internal heat accumulation; the time decay weight factor is introduced because the more recent the temperature rise, the greater the risk of undissipated heat energy posing a thermal shock to the present and the near future.

[0061] This computational logic with memory-forgetting characteristics solves the problem that simple calculations of average temperature or macroscopic temperature difference cannot reflect the internal real-time heat accumulation rate and thermal inertial potential energy. It transforms static temperature readings into a dynamic, predictive physical gradient index, enabling the control to accurately assess the scale of latent heat energy that has actually accumulated inside the current distribution cabinet after experiencing preceding load fluctuations and has not yet manifested as visible high temperature.

[0062] like Figure 4 As shown, in one embodiment, S3 includes: S31, retrieving an initial standard reference value pre-stored in a non-volatile memory, dynamically compensating and correcting the initial standard reference value using the computer room ambient temperature value, and obtaining the dynamic current reference value at the current moment. Wherein, the initial reference value... The data was collected and statistically analyzed using healthy operating data at a standard ambient temperature of 25℃. The specific process involved placing the distribution cabinet in a 25℃ constant-temperature laboratory and continuously collecting current data under normal load fluctuations for one week. The fluctuation heat integral for each 60-second window was calculated using a formula, resulting in approximately 10,080 samples. The 95th percentile of this sample set was taken as the upper limit of normal fluctuation. This value represents the permissible normal fluctuation energy limit under light load conditions.

[0063] S32. Calculate the absolute difference between the fluctuating heat integral and the corresponding dynamic current reference value for each interval. Divide this absolute difference by the dynamic current reference value to obtain the first error ratio for each interval, and extract the maximum value among all first error ratios as the load dispersion. Similarly, calculate the absolute difference between the thermal hysteresis gradient and the corresponding dynamic thermistor reference value for each interval. Divide this absolute difference by the dynamic thermistor reference value to obtain the second error ratio for each interval, and extract the maximum value among all second error ratios as the electrothermal drift. The dynamic thermistor reference value is also obtained through long-term statistical analysis of the thermal hysteresis gradient under a standard 25℃ environment. Simultaneously with the dynamic current reference value, thermal hysteresis gradient data are collected for each 60s window within the same week, and the 95th percentile is used as the reference value. For example, if the 95th percentile of the thermal hysteresis gradient sample is 0.5℃, then the dynamic thermistor reference value is set to 0.5℃.

[0064] S33. The load dispersion and electrothermal drift are weighted and summed to calculate the thermal steady-state index, which is then compared with a preset stability threshold. The steady-state threshold is determined by analyzing the thermal steady-state index distribution of a large amount of historical healthy operating data. 200 reference cycles (each cycle 24 hours) of the distribution cabinet under normal operating conditions are selected, and the thermal steady-state index for each cycle is calculated. This yielded a set of 200 samples; the 99th percentile of this set of samples was taken as the anomaly detection threshold, i.e. A value above 0.35 is considered healthy; a value exceeding 0.35 may indicate an abnormality. In this embodiment, 0.35 is the 99th percentile.

[0065] In this embodiment, it should be noted that, considering the resistivity of metallic conductors increases with rising ambient temperature, the thermal shock baselines generated in summer and winter are completely different under the same current fluctuations. Therefore, an ambient temperature-adaptive baseline drift equation is introduced. Taking the first load range as an example, the calculation expression is as follows:

[0066] In the above expression, Represents the current dynamic current reference value for the first load range; This represents the initial standard reference value pre-set in the non-volatile memory under standard test conditions at 25 degrees Celsius; The temperature coefficient of resistance of the copper conductor in the main circuit is taken as a fixed value of 0.00393. 25 represents the real-time ambient temperature value of the computer room acquired by the data acquisition module, in degrees Celsius; 25 represents the standard ambient temperature constant.

[0067] Furthermore, after obtaining the current dynamic current reference value for each interval, the relative error ratio between the first fluctuating heat integral and the corresponding dynamic current reference value is calculated. Similarly, the relative error ratios for the second and third intervals are calculated. The maximum value among these three relative error ratios is the load dispersion.

[0068] Furthermore, using the same dynamic compensation and comparison logic, the electrothermal drift is calculated based on the first, second, and third thermal hysteresis gradients. The load dispersion and electrothermal drift characterize the severity of deviation from the normal physical evolution trajectory at the electrical energy input end and the thermal energy response end, respectively.

[0069] Subsequently, the load dispersion is multiplied by a preset constant of 0.4, and the electrothermal drift is multiplied by a preset constant of 0.6. The products of the two are then added together to calculate the thermal steady-state index. When the thermal steady-state index is less than the pre-calibrated stability threshold of 0.35, it is determined that the current electromechanical state is intact and has not been affected by sensor damage or external extreme interference, allowing the triggering of predictive control logic.

[0070] In this embodiment, it should also be noted that in S31, an adaptive ambient temperature reference drift equation is constructed. Specifically, the expression for obtaining the current dynamic current reference value of the first load range is as follows: Assuming that in non-volatile memory, the initial standard reference value is calibrated at the factory in a constant temperature laboratory at 25°C. The temperature coefficient of resistance of the copper conductor in the main circuit is 400J (representing the permissible normal fluctuation energy limit under light load). The constant is 0.00393, for the ambient temperature sensor. The real-time data transmitted back is 35°C.

[0071] First, calculate the ambient temperature deviation. The difference between the current temperature and the standard temperature is: Then, the impedance thermal drift compensation coefficient was calculated. The physical properties of copper determine that for every 1°C increase in temperature, the resistivity increases by approximately 0.393%. The compensation ratio for a deviation of 10°C is: (That is, the base resistance is increased by 3.93%). Finally, a dynamic current reference value is generated, and the drift ratio is superimposed on the base value: J.

[0072] In summary, in the existing technology's judgment logic, the baseline threshold is usually fixed (e.g., fixed at 400J). If, in a computer room at 35°C in summer, the equipment experiences the same minute current fluctuation as usual, the actual Joule heat fluctuation integral might reach 415J due to the increased basic resistance of the copper busbar at high temperatures. However, if a traditional static threshold (400J) is used, comparison reveals... This will immediately trigger a data mutation / abnormality alarm, causing the control logic to lock up or triggering an invalid full-speed exhaust. This expression introduces a flexible scale that adapts to ambient temperature. This is achieved by incorporating the temperature coefficient of resistance of the physical material. Calculations showed that under 35°C conditions, normal energy fluctuations should be relaxed to 423.58 J. Therefore, the actual monitored value of 415 J was compared with the dynamically corrected 423.58 J to identify... This indicates that the problem is a reasonable physical drift rather than a malfunction.

[0073] In S32, the most extreme operating condition deviation phenomena are captured by comparing the relative errors of multiple intervals. Taking the fluctuating heat integral as an example, the central controller compares the current actual energy integral calculated in S21, such as 439.2J in the first interval, with its corresponding dynamic current reference value of 415.72J, calculates its absolute error and then divides it by the reference value, i.e., |439.2-415.72| / 415.72, to obtain the relative deviation of this interval as 5.64%. The relative deviations of the second and third intervals are calculated in parallel, for example, 3.2% and 4.8% respectively. Then, the maximum value extraction function is executed, comparing the three values, and selecting the largest one, 5.64%, as the load dispersion representing the overall electrical fluctuation heterogeneity. The same operation logic is applied to the processing of the thermal hysteresis gradient, comparing it with the corresponding temperature reference value, and extracting the maximum value of the temperature rise deviation in each interval, for example, 8.2%, as the electrothermal drift.

[0074] Furthermore, this logic of calculating the deviation and extracting the maximum difference between multiple operating conditions solves the problem that simply relying on the global data average value can easily mask serious faults in local intervals. It can examine the consistency of data with the most stringent standards and effectively identify the hidden tearing of the physical law of electrothermal coupling under cross-load conditions.

[0075] In S33, a single thermal steady-state index is output through weighted fusion, and a safety verification threshold comparison is performed. In the operating mechanism of the distribution cabinet, abnormal thermal response typically has a higher destructive potential than short-term power fluctuations because heat accumulation directly leads to the aging of insulation materials and the burnout of components. Therefore, a first weighting constant, such as 0.4, is assigned to the load dispersion in the formula, and a higher second weighting constant, such as 0.6, is assigned to the electrothermal drift. Wherein, the weighting coefficients... and Principal component analysis based on historical failure cases was used to determine the load dispersion within a reference period prior to the failure, based on 20 distribution cabinet overheating failure events that occurred in the past year. and thermal drift Principal component analysis revealed that... Its contribution to the failure rate is approximately 60%. The contribution rate is approximately 40%, therefore it is set as follows: , .

[0076] Substituting the value obtained from S32 into the calculation: 0.4*0.0564+0.6*0.082, we get 0.02256+0.0492, finally yielding a thermal steady-state index of 0.07176. This result is compared with a stability threshold, such as 0.35, pre-calibrated using a large number of healthy samples in the non-volatile memory. Since 0.07176 < 0.35, the central controller determines that the current acquired data is logically consistent and the sensor feedback is reliable, allowing subsequent predictive computing gateways to be opened. If the index exceeds 0.35, it will determine that the data is distorted or that the equipment has a serious hidden danger, thus refusing to execute predictive control, instead outputting a hardware alarm and maintaining maximum safe heat dissipation power.

[0077] Furthermore, this logic of weighted summation and comparison with thresholds solves the problem that fluctuations in a single indicator can easily lead to frequent malfunctions, balancing sensitivity and fault tolerance. Before entering the complex and computationally intensive forward prediction model, it ensures that only high-quality data sequences reflecting healthy physical laws can drive the regulation of the cooling fan, avoiding the risk of oscillations or reverse adjustments in the cooling actuator due to inputting incorrect data.

[0078] like Figure 5 As shown, in one embodiment, S4 includes: S41, receiving the expected load current value for the target prediction period, and calculating the predicted load current sequence by combining the fluctuating heat integral of the corresponding interval.

[0079] S42. Establish the differential equation for the electrical-thermal energy balance in the distribution cabinet, substitute it with the predicted load current sequence and the thermal hysteresis gradient of the corresponding interval, and solve to obtain the predicted temperature rise rate.

[0080] S43. Calculate the aerodynamic dead zone time from startup to the establishment of a steady-state flow field for the cooling fan.

[0081] S44. Generate a target duty cycle value based on the predicted temperature rise rate, and output the target duty cycle value to the heat dissipation actuator when the expected high load occurs and the dead time is subtracted to obtain the feedforward advance.

[0082] In this embodiment, it should be noted that after the stability is determined, the central controller receives the expected load current value for the target prediction period and predicts future load fluctuations based on the autoregressive moving average algorithm.

[0083] Furthermore, the Runge-Kutta numerical method is used to solve the differential equation for the electro-thermal energy balance within the distribution cabinet. The input term of the differential equation is the Joule heating power generated by the predicted load current, and the output term is the heat dissipation power carried away by the adjustable-speed fan. The state variables include the specific heat capacity of the components. The thermal hysteresis gradient for the corresponding interval is converted into a parameter that hinders heat dissipation in the equation (this parameter combines the nonlinear phase change heat transfer thermal resistance of the phase change heat pipe under different temperature difference conditions and the convective thermal resistance of the condenser side surface). The highest predicted temperature rise and the predicted temperature rise rate within the target prediction period are obtained by solving the equation.

[0084] During the control command generation phase, a lead control mechanism for the cooling dead time is introduced. In fluid dynamics, there is a physical time delay between the fan receiving the electrical duty cycle command and overcoming duct resistance to establish a stable cooling circulation field within the cabinet. The central controller retrieves the preset aerodynamic dead time. .

[0085] Further, the target duty cycle value is calculated. The base duty cycle is added to two adjustment components: the first component is the proportional feedback quantity when the predicted temperature rise exceeds the safety threshold, and the second component is the feedforward compensation quantity when the predicted temperature rise rate is multiplied by the differential feedforward gain coefficient.

[0086] Furthermore, intervention is performed on the timeline: the expected load surge time for external scheduling is set as... The central controller is not located there. Actions are triggered at all times, but only when... At this feedforward timing, the target duty cycle value is immediately output to the heat dissipation actuator. By forcibly exhausting air in advance, the static hot air accumulated inside the cabinet is evacuated, forming an initial negative pressure cooling field. When a sudden high load causes a rapid release of nonlinear heat energy, it is completely absorbed by the airflow field that has been established to its peak value, effectively suppressing the transient temperature rise peak of the heat source node.

[0087] In this embodiment, it should also be noted that in S41, the macroscopic instructions from external scheduling are combined with the microscopic high-frequency characteristics of the equipment to reconstruct a high-resolution expected load curve. In actual industrial environments, external scheduling often issues step-like average setpoints, such as notifying that the expected load current will jump to 900A at 14:30. If 900A is used as a smooth straight line input to the thermal model, the key information for generating Joule thermal spikes will be lost. Therefore, the central controller internally calls the autoregressive integral moving average algorithm, using 900A in the target prediction period as the baseline parameter, and simultaneously extracts the third fluctuation thermal integral of the corresponding third interval calculated in S21. This historically extracted high-frequency oscillation feature is injected into the baseline parameter in the form of variance superposition. After the algorithm's fusion calculation, the originally flat 900A instruction is restored into a predicted load current sequence with high-frequency fluctuations between 880A and 930A, containing real industrial harmonics and load perturbation characteristics.

[0088] In S42, the temperature evolution trajectory of the equipment is processed on the time axis based on the differential equation of the electrical-thermal energy balance of the distribution cabinet. The left side of the differential equation includes the product of the equivalent mass of the heat-generating components (e.g., 15 kg) and the comprehensive specific heat capacity (e.g., 385 J / (kg*℃), multiplied by the derivative of temperature with respect to time, i.e., the temperature rise slope); the right side of the equation is the predicted heat generation power minus the convective power loss. The expression for the differential equation of the electrical-thermal energy balance of the distribution cabinet is as follows: This is used to dynamically simulate the thermodynamic evolution trajectory of components inside a power distribution cabinet under anticipated electrical energy surges. The equivalent mass of the heating element and the evaporator end of the phase change heat pipe and To take into account specific heat capacity, Characterize the physical quality of the main heat-generating components inside the cabinet (such as the main circuit copper busbar, circuit breaker contacts, etc.) and the coupling heat pipes. The product of the two reflects the inherent physical property of this particular material that absorbs heat, leading to an increase in temperature. This constitutes the heat capacity or thermal inertia parameter of this physics. It should be noted that the equivalent mass... The mass of the main heat-generating components (main circuit copper busbar, circuit breaker contacts, etc.) in the distribution cabinet is obtained by weighing or from design drawings. In this embodiment, the total mass of the copper busbar and the evaporator end of the heat pipe is 15 kg. (The overall specific heat capacity is then considered.) The specific heat capacity of copper is taken as 385 J / (kg·℃), which is the standard value given in the material handbook.

[0089] in, To predict the time derivative of temperature rise, i.e., the temperature rise slope, this is used as the core unknown state variable in solving the differential equation, representing the rate of temperature change per unit time. This allows the control system to calculate the rate characteristics of temperature rise, rather than being limited to predicting a final static high-temperature value. Extracting this rate parameter provides the necessary dynamic basis for subsequently developing feedforward intervention strategies with time lead. To predict the internal Joule heating power, the energy input source for the thermodynamic differential equation represents the transient heat generation rate generated when the expected external load current flows through the equivalent resistance. The expected power fluctuation parameters issued by the production schedule are transformed into input variables for the thermodynamic model, achieving an equivalent mapping of the energy source in the electro-thermal cross-coupling model and ensuring that the assessment of the heat source input conforms to the Joule heating law. The convective power loss represents the rate at which heat is carried away by the cooling airflow within the distribution cabinet. This parameter incorporates the previously extracted thermal hysteresis gradient as a dynamic thermal resistance correction term in the calculation. It objectively reflects the dynamic changes in the actual heat dissipation efficiency inside the distribution cabinet under different fan speeds and different initial heat accumulation conditions, thus improving the physical accuracy of the net heat flow calculation. Furthermore, in this embodiment, Instead of using a fixed convective heat transfer coefficient, it is dynamically corrected based on the thermal hysteresis gradient. The corrected calculation formula is as follows: ,in, This represents the current internal node temperature. External ambient temperature, For thermal hysteresis gradient The converted parameter integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions. ,in The basic thermal resistance is determined by the equipment structure and calibrated at the factory. The correction coefficient is obtained by fitting dynamic response data under different loads. For example, multiple step loads are first applied, the temperature rise curves are recorded, and the least squares method is used to minimize the error between the model output and the actual temperature, thus fitting the corrected coefficient. This correction allows the convective heat dissipation power term to accurately reflect the real-time impact of the phase change heat pipe's own heat transfer hysteresis and the airflow state on the condenser side on heat dissipation efficiency under specific load histories. Based on the above calculation process, the corrected... and By substituting the differential equation of the electrical-thermal energy balance of the distribution cabinet into the numerical solution using the fourth-order Runge-Kutta method, the temperature rise trajectory within the target time period can be predicted, and the maximum temperature rise rate can be extracted. and the highest temperature appreciation .

[0090] Furthermore, the mean of the predicted load current sequence output by S41 is substituted into Joule's law formula to calculate the predicted heating power. , This is the predicted value of heat generation power. To predict the mean of the load current sequence, (The equivalent contact resistance of the main circuit of the distribution cabinet), for example, the square of the current 900A multiplied by 0.0002 ohms, yields a heat output of 162W.

[0091] Furthermore, instead of using a fixed convective heat dissipation coefficient, the thermal hysteresis gradient obtained in S22 is transformed into a parameter (which integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions) to correct the power loss in real time, for example, to 62W. Substituting into the equation, (162-62) / (15*385) calculates a temperature rise slope of 0.0173℃ / s, or a predicted temperature rise rate of 1.038℃ / min. Simultaneously, the predicted temperature rise value for this period is calculated using a discretized numerical method, for example, 72.5℃.

[0092] In S43, when the central controller changes the PWM duty cycle signal, the motor rotor of the DC brushless adjustable speed fan needs to overcome its own rotational inertia to accelerate. The fan blades push the air to overcome the damping of the inlet and outlet filters, and establish a stable convective circulation flow field that meets expectations between the fin arrays on the condenser side of the phase change heat pipe. This physical conversion process from electrical signal to steady-state airflow field has an insurmountable time difference. The dead time parameters obtained through wind tunnel experiments and field calibration are pre-stored internally. For example, at the current base speed, the time t_dead required to increase to a high speed and complete the global air replacement in the cabinet is set to 3 minutes.

[0093] In S44, multidimensional prediction parameters are fused to generate the target duty cycle, and a feedforward action with lead time is executed. A built-in duty cycle adjustment formula is included, the expression of which is: This is a feedforward-feedback joint control law with predictive properties, used to convert the solved multidimensional thermodynamic parameters into pulse-width modulated electrical signals that can be directly recognized by the adjustable-speed fan. The target pulse width modulation duty cycle, serving as the output command terminal of the entire algorithm's control equations, directly determines the average input voltage of the fan motor, thereby controlling the cold air flow. This completes the closed-loop transformation from software-level thermodynamic data prediction to hardware-level electromechanical parameter driving. Among these, To maintain the basic duty cycle, a pre-set basic operating speed parameter corresponding to the current load range (light load, medium load, or heavy load) of the distribution cabinet is used. This ensures the basic cooling convection requirements of the distribution cabinet under steady-state conditions, maintains the initial cooling flow field inside the cabinet, and avoids mechanical static friction wear and additional aerodynamic start-up delay caused by starting the fan from zero speed. It should also be noted that the basic duty cycle corresponds to the minimum fan speed required for the distribution cabinet to maintain basic convection cooling under light load conditions. Under conditions where the load rate is below 30% and there is no external heat source interference, the fan duty cycle is gradually reduced until the internal temperature begins to rise slowly, and the duty cycle at this point is recorded as the basic value. Testing showed that when the duty cycle is 30%, the internal node temperature stabilizes at ambient temperature +5℃, meeting the light load heat dissipation requirements; therefore, this value is set accordingly. .

[0094] in, For proportional feedback adjustment, To predict the highest temperature, The preset safe temperature upper limit, The proportional control gain constant is used to calculate the basic steady-state cooling compensation for the portion of the predicted temperature exceeding the safety threshold. This provides targeted airflow intervention for cases where the absolute temperature exceeds the limit, ensuring the final physical temperature is kept within the safety limit and providing basic static temperature protection for equipment operation. It should be noted that the upper limit of the safe temperature is set based on the maximum allowable operating temperature of key components in the distribution cabinet (such as circuit breaker contacts and insulation materials). Consulting the component specifications, for example, if the maximum allowable temperature for circuit breaker contacts is 70℃, a 5℃ safety margin is considered, and this is set accordingly. This value can also be verified through accelerated aging tests to ensure that the lifespan of components is not significantly affected when operating below this temperature for a long period of time.

[0095] in, For differential feedforward compensation term, The predicted temperature rise rate is obtained by solving the differential equation. The differential feedforward gain constant is used to calculate and increase the cooling compensation amount in advance based on the rate parameter of temperature change. This differential feedforward compensation term is the core adjustment term for achieving proactive thermal management. Even if the current actual temperature or the predicted final temperature does not exceed the safety limit, as long as the temperature is predicted to rise at a high slope, this term will independently calculate the duty cycle increment. This allows the heat dissipation to respond in advance according to the rate of thermal shock evolution, effectively suppressing the transient temperature rise peak caused by sudden high loads. It should also be noted that the control gain... and The method was determined by combining simulation tuning and on-site fine-tuning. First, a thermal model of the distribution cabinet and a fan control model were built in MATLAB / Simulink. Typical load disturbances (such as a step change from 300A to 900A) were input. With temperature overshoot and settling time as optimization objectives, a particle swarm optimization algorithm was used to find the optimal values, and preliminary results were obtained. , Subsequently, on-site testing was conducted on the actual distribution cabinet, with fine-tuning to ensure the overshoot was less than 2°C and the adjustment time was minimized, ultimately determining the optimal solution. , .

[0096] For example, the current base duty cycle is 30%, the safe upper temperature limit is 65℃, the proportional control gain is 2%, and the differential feedforward gain is 15%. Substituting the predicted temperature rise of 72.5℃ and the ramp rate of 1.038℃ / min obtained from S42 into the formula: 30%+2%*(72.5-65)+15%*1.038, the target duty cycle of 60.57% that needs to be output to the fan is calculated.

[0097] In terms of execution timing, based on the 3-minute aerodynamic dead zone time calibrated by S43, this 3-minute advance is subtracted from the expected high load time of 14:30. Therefore, at the feedforward time of 14:27, the central controller instantly increases the duty cycle from 30% to 60.57%.

[0098] This logic, which combines the predicted slope to generate the control quantity and subtracts the physical dead time to execute the advance action, solves the problem that passive cooling always waits for the temperature to exceed the limit before starting to turn on the fan, causing components to inevitably suffer thermal damage during the temperature rise ramp-up period. It uses the time difference to vent the static hot air in the cabinet before the thermal shock arrives, thus completing the preparation of the cold environment.

[0099] A heat dissipation control system for a distribution cabinet based on a phase change heat pipe is also provided, which is used to implement a heat dissipation control method for a distribution cabinet based on a phase change heat pipe. The system includes: The data acquisition module is used to acquire the effective current value, internal node temperature value and external ambient temperature value of the distribution cabinet within the reference period; according to the preset load rate boundary, the effective current value and the internal node temperature value are respectively assigned to the first load interval, the second load interval or the third load interval, and the load sequence and thermal dynamic sequence of each interval are generated by combining them in the order of sampling time. The first processing module is used to obtain the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral respectively according to the load sequence of each interval; and to obtain the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient respectively according to the thermal dynamic sequence of each interval. The second processing module is used to calculate and obtain a dynamic current reference value based on a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value; obtain the load dispersion based on the relative error between the fluctuating heat integral of each interval and the corresponding dynamic current reference value; obtain the electrothermal drift based on the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value; and obtain the thermal steady-state index based on the load dispersion and the electrothermal drift. The predictive control module is used to: obtain the expected load current for the target prediction period when the thermal steady-state index is less than a preset steady-state threshold; solve a preset electrothermal differential equation based on the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, to obtain the predicted temperature rise rate and the predicted temperature rise value; obtain the target duty cycle based on the predicted temperature rise rate, the predicted temperature rise value, and a preset base duty cycle; subtract a preset aerodynamic dead zone from the start time of the target prediction period to obtain the feedforward trigger time, and output a control command containing the target duty cycle at the feedforward trigger time to drive the cooling fan on the condenser side of the corresponding phase change heat pipe.

[0100] In one embodiment, the data acquisition module is further configured to: acquire the rated current of the distribution cabinet; divide the acquired effective current value by the rated current to obtain the real-time load rate; classify the real-time load rate into the corresponding first load interval, second load interval, or third load interval according to three non-overlapping preset load rate percentage ranges; extract all effective current values ​​within the same load interval and arrange them in chronological order of sampling time to form the load sequence; extract all internal node temperature values ​​within the same load interval and arrange them in chronological order of sampling time to form the thermal dynamic sequence. To enable those skilled in the art to fully understand and implement the technical solutions described in this specification, the following section, using a specific application scenario, provides a detailed deduction and data analysis of the entire implementation principle of this phase change heat pipe-based power distribution cabinet heat dissipation control system. Assume that a power distribution cabinet with rated current parameters is operating in an industrial plant. For low-voltage distribution cabinet A, the estimated constant value of the equivalent contact resistance of the main heating nodes (main circuit copper busbars and circuit breaker contacts) is as follows: Equivalent mass of heating element kg, overall specific heat capacity J / (kg·℃). Currently operating under high summer temperatures, the data acquisition module obtains the real-time ambient temperature value of the computer room. ℃. During the monitoring time window from 14:00 to 14:03, the system uses s represents the time interval during which 3 sampling points were continuously collected (i.e., The system first executes the interval division logic, collecting the current value for the first minute. A, Substitute into the real-time load factor formula Calculation Since 20% falls within the threshold of 0% to 30%, the central controller accurately determines that the current system is in the first load range (light load state). The effective current values ​​collected in the 2nd and 3rd minutes are as follows: A and A, all belong to the first load characteristic sequence.

[0101] After obtaining the above basic sequence, the system calls the nonlinear Joule energy flow integral algorithm to extract the power fluctuation characteristics inside the distribution cabinet. Based on the formula... The central controller performs discrete calculations to determine the energy difference between adjacent time points. For the first minute to the second minute (… The injected fluctuating thermal energy increment is J. For the 2nd to 3rd minute ( Although the current decreased, it still caused a thermal shock to the contact surface, with a fluctuating increase in thermal energy of [value missing]. J. Summing the integrals of these two fluctuations yields the first fluctuation heat integral within the monitoring window. J. This value strips away the simple linear difference of current in its physical essence, truly reflecting the severity of nonlinear Joule thermal shock.

[0102] Simultaneously, the data acquisition module obtained the corresponding temperature sample values ​​from the three nodes: ℃, ℃, ℃. The system calculates the first thermal hysteresis gradient. The calculation logic is to extract the positive temperature rise and accumulate it by combining it with a time decay weighting factor (setting the decay base to 0.90): .for The temperature difference is ℃, distance from the latest sampling point ( The time step span is Therefore, its weighted value is ℃. For The temperature difference is ℃, time step span is Its weighted value is ℃. Summing these values ​​yields the first thermal hysteresis gradient as: ℃. This value precisely quantifies the hysteretic thermal stress accumulated in the air and metal components within the cabinet due to recent heating behavior.

[0103] Next, the system executes a rigorous benchmark correction and stability assessment process. The initial standard benchmark value set in the non-volatile memory under the standard test condition of 25°C is known. J, Temperature coefficient of resistance of the copper conductor in the main circuit To eliminate the error caused by the increased material background impedance due to high summer temperatures, the system introduces a reference drift equation that is adaptive to ambient temperature: Substitute the parameters to obtain the current dynamic current reference value. J. Subsequently, the system calculates the fluctuating heat integral for the first load interval. J and its corresponding dynamic current reference value The absolute difference of J, divided by the dynamic current reference value, yields the first error ratio for that interval. Similarly, the first error ratios for the second and third load intervals are calculated in parallel, for example, 0.032 and 0.048 respectively. The maximum value of these three first error ratios, 0.0564, is extracted as the load dispersion. Simultaneously, based on the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value, the same operation is performed to calculate the second error ratio of each interval, and the maximum value is extracted as the electrothermal drift degree, thus obtaining... .

[0104] After extracting the drift degree in each dimension, the central controller calculates the thermal steady-state index and substitutes it into the formula. The first weighting constant Set to 0.4, the second weighting constant. Set it to 0.6. The calculation yields... Due to the calculation The value is much lower than the preset steady-state threshold of 0.35. The central controller determines that the current electrical system has not suffered from abnormal electromagnetic interference and that the sensor thermodynamic feedback link is healthy. It allows the system to proceed and starts the forward-looking heat dissipation differential prediction core engine.

[0105] In the forecasting phase, the external production scheduling system issues instructions, predicting the target time. The distribution cabinet will be connected to a high-power motor, and the expected load current will jump to [a certain value]. A. At this point, the system constructs an electro-thermal energy balance differential equation. Perform a virtual simulation. Predicted heat generation power. W. Assuming the convective power loss at the current wind speed, corrected by the just calculated thermal hysteresis gradient. W. Substituting into the equation, we get W. Solve for the slope of the temperature rise. ℃ / s, converted to minute-level predicted temperature rise rate ℃ / min. Simultaneously, the predicted temperature rise during the impact window period is obtained through discrete solution. ℃.

[0106] Finally, the system generates and executes electromechanical actions with aerodynamic feedforward compensation. This is based on the adjustment formula. Given the current base duty cycle Upper limit of safe temperature ℃, proportional control gain Differential feedforward gain The target duty cycle is calculated by substituting the parameters. To eliminate hydrodynamic hysteresis, the system retrieves a preset aerodynamic dead zone. Minutes. The central controller did not wait for the high load to actually occur at 14:30, but instead triggered the feedforward at that moment, i.e. At 14:27, the PWM duty cycle of the adjustable speed fan was instantly increased from 30% to 60.57%. This 3-minute advance replacement significantly reduced the static hot air in the distribution cabinet. When the 162W Joule heat actually erupted at 14:30, it was completely offset by the cold air circulation at its maximum flow rate peak, keeping the actual physical temperature rise within a safe threshold and closing the complete technical link from load fluctuation identification to dynamic feedforward cooling.

[0107] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.

[0108] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0109] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.

[0110] 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. Such 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, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A heat dissipation control method for a power distribution cabinet based on a phase change heat pipe, characterized in that, The methods include: The effective current value, internal node temperature value and external ambient temperature value of the distribution cabinet within the reference period are obtained. According to the preset load rate boundary, the effective current value and the internal node temperature value are respectively assigned to the first load interval, the second load interval or the third load interval. The load sequence and thermal dynamic sequence of each interval are generated by combining them in the order of sampling time. The first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral are obtained according to the load sequence of each interval, and the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient are obtained according to the thermal dynamic sequence of each interval. The dynamic current reference value is calculated and obtained based on the preset initial reference value, the preset temperature coefficient of resistance, and the external ambient temperature value. The load dispersion is obtained based on the relative error between the fluctuating heat integral of each interval and the corresponding dynamic current reference value. The electrothermal drift is obtained based on the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value. The thermal steady-state index is obtained based on the load dispersion and the electrothermal drift. If the thermal steady-state index is less than the preset steady-state threshold, the expected load current for the target prediction period is obtained. Based on the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, the preset electrothermal differential equation is solved, and the predicted temperature rise rate and predicted temperature rise value are obtained. The target duty cycle is obtained based on the predicted temperature rise rate, the predicted temperature rise value, and the preset base duty cycle. The feedforward trigger time is obtained by subtracting the preset aerodynamic dead zone from the start time of the target prediction period. At the feedforward trigger time, a control command containing the target duty cycle is output to drive the cooling fan on the condenser side of the corresponding phase change heat pipe.

2. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The process involves classifying the effective current value and the internal node temperature value into a first load interval, a second load interval, or a third load interval based on a preset load rate boundary, and generating a load sequence and thermal dynamic sequence for each interval according to the sampling time sequence, including: Obtain the rated current of the distribution cabinet; Divide the obtained effective value of the current by the rated current to obtain the real-time load rate; Based on three non-overlapping preset load rate percentage ranges, the real-time load rate is respectively assigned to the corresponding first load interval, second load interval, or third load interval; Extract all effective current values ​​within the same load range and arrange them in chronological order of sampling time to form the load sequence; The temperature values ​​of all internal nodes within the same load range are extracted and arranged in chronological order of sampling time to form the thermal dynamic sequence.

3. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The step of obtaining the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral based on the load sequence of each interval includes: Within the load sequence of each interval, the effective current values ​​of two adjacent sampling time points are obtained, and the absolute value of the squared difference of the effective current values ​​of the two adjacent sampling time points is calculated. The absolute value of the square difference, the preset equivalent resistance parameter, and the preset sampling time interval are multiplied to obtain the fluctuation thermal energy increment. The first, second, and third fluctuating heat integrals are obtained by summing up all the fluctuating heat energy increments within the same load sequence.

4. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The step of obtaining the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient based on the thermal dynamic sequence of each interval includes: Within the thermal dynamic sequence of each interval, the temperature difference between two adjacent sampling time points is calculated, and the temperature difference with a value greater than zero is extracted. Obtain the time decay weight with a preset decay base and an exponent of the number of time steps between the current sampling time point and the latest sampling time point; The temperature difference is multiplied by the corresponding time decay weight and then summed to obtain the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient.

5. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The step of calculating and obtaining a dynamic current reference value based on a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value includes: Obtain the preset standard test temperature; The temperature deviation is obtained by subtracting the standard test temperature from the external ambient temperature. The drift compensation ratio is obtained by multiplying the temperature deviation by the temperature coefficient of resistance. The initial reference value is combined with the drift compensation ratio and positively multiplied to obtain the dynamic current reference value.

6. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The process of obtaining load dispersion based on the relative error between the fluctuating heat integral of each interval and the corresponding dynamic current reference value, obtaining electrothermal drift based on the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value, and obtaining thermal steady-state index based on the load dispersion and the electrothermal drift includes: Calculate the absolute difference between the fluctuating heat integral and the corresponding dynamic current reference value for each interval, divide the absolute difference by the dynamic current reference value to obtain the first error ratio for each interval, and extract the maximum value among all the first error ratios as the load dispersion. Calculate the absolute difference between the thermal hysteresis gradient and the corresponding dynamic thermistor reference value for each interval, divide the absolute difference by the dynamic thermistor reference value and obtain the second error ratio for each interval, and extract the maximum value among all the second error ratios as the electrothermal drift. The load dispersion is multiplied by a preset first weight, the electrothermal drift is multiplied by a preset second weight, and the products of the two are added together to obtain the thermal steady-state index.

7. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The step of solving a preset electrothermal differential equation based on the expected load current and the target thermal hysteresis gradient corresponding to the load range to which the expected load current belongs, and obtaining the predicted temperature rise rate and predicted temperature rise value, includes: The expected load current is superimposed with the fluctuating heat integral corresponding to the load range to reconstruct and obtain the predicted load sequence; The state coefficients are obtained from the electrothermal differential equation, which includes the preset mass of the heating element and the preset specific heat capacity parameter, and the state variable is the derivative of the temperature rise with respect to time. The predicted load sequence is squared and multiplied by a preset resistance to obtain the predicted heating power. The target thermal hysteresis gradient is transformed into (this parameter integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions of the phase change heat pipe), and the convective heat dissipation power term in the electrothermal differential equation is corrected according to (this parameter integrates the nonlinear phase change heat transfer thermal resistance and the convective thermal resistance of the condenser side surface under different temperature difference conditions). The predicted heating power and the corrected convective heat dissipation power are substituted into the electrothermal differential equation for discretization and solution. The maximum value of the time derivative of the temperature rise during the target prediction period is extracted as the predicted temperature rise rate, and the highest temperature during the solution process is extracted as the predicted temperature rise value.

8. The heat dissipation control method for a distribution cabinet based on a phase change heat pipe according to claim 1, characterized in that, The step of obtaining the target duty cycle based on the predicted temperature rise rate, the predicted temperature rise value, and the preset base duty cycle includes: Obtain the preset safe temperature limit, preset proportional control gain, and preset differential feedforward gain; Subtract the safe temperature limit from the predicted temperature rise value to obtain the over-limit temperature difference, multiply the over-limit temperature difference by the proportional control gain to obtain the first control component; Multiply the predicted temperature rise rate by the differential feedforward gain and obtain the second control component; The target duty cycle is obtained by adding the base duty cycle, the first control component, and the second control component.

9. A heat dissipation control system for a distribution cabinet based on a phase change heat pipe, characterized in that, The system is used to implement the heat dissipation control method for power distribution cabinets based on phase change heat pipes as described in any one of claims 1 to 8, and the system includes: The data acquisition module is used to acquire the effective current value, internal node temperature value and external ambient temperature value of the distribution cabinet within the reference period; according to the preset load rate boundary, the effective current value and the internal node temperature value are respectively assigned to the first load interval, the second load interval or the third load interval, and the load sequence and thermal dynamic sequence of each interval are generated by combining them in the order of sampling time. The first processing module is used to obtain the first fluctuating heat integral, the second fluctuating heat integral, and the third fluctuating heat integral respectively according to the load sequence of each interval; and to obtain the first thermal hysteresis gradient, the second thermal hysteresis gradient, and the third thermal hysteresis gradient respectively according to the thermal dynamic sequence of each interval. The second processing module is used to calculate and obtain a dynamic current reference value based on a preset initial reference value, a preset temperature coefficient of resistance, and the external ambient temperature value; obtain the load dispersion based on the relative error between the fluctuating heat integral of each interval and the corresponding dynamic current reference value; obtain the electrothermal drift based on the relative error between the thermal hysteresis gradient of each interval and the preset dynamic thermistor reference value; and obtain the thermal steady-state index based on the load dispersion and the electrothermal drift. The predictive control module is used to: obtain the expected load current for the target prediction period when the thermal steady-state index is less than a preset steady-state threshold; solve a preset electrothermal differential equation based on the expected load current and the target thermal hysteresis gradient corresponding to the load interval to which the expected load current belongs, to obtain the predicted temperature rise rate and the predicted temperature rise value; obtain the target duty cycle based on the predicted temperature rise rate, the predicted temperature rise value, and a preset base duty cycle; subtract a preset aerodynamic dead zone from the start time of the target prediction period and obtain the feedforward trigger time; and output a control command containing the target duty cycle at the feedforward trigger time to drive the cooling fan on the condenser side of the corresponding phase change heat pipe.

10. The heat dissipation control system for a distribution cabinet based on a phase change heat pipe according to claim 9, characterized in that, The data acquisition module is also used for: Obtain the rated current of the distribution cabinet; Divide the obtained effective value of the current by the rated current to obtain the real-time load rate; Based on three non-overlapping preset load rate percentage ranges, the real-time load rate is respectively assigned to the corresponding first load interval, second load interval, or third load interval; Extract all effective current values ​​within the same load range and arrange them in chronological order of sampling time to form the load sequence; The temperature values ​​of all internal nodes within the same load range are extracted and arranged in chronological order of sampling time to form the thermal dynamic sequence.

Images