A method for temperature control of a heat sink in a thermal vacuum environment simulation by nitrogen gas heating and liquid nitrogen refrigeration

Abstract

This invention relates to the field of refrigeration technology, specifically to a method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a simulated heat sink under thermal vacuum conditions. The method includes the following steps: acquiring heat sink temperature signals and recording valve opening times; denoising the temperature sequence; extracting the response start times of gaseous and liquid nitrogen and determining the hysteresis levels; generating phase adjustment indicators; generating basic control commands based on duty cycle and energy competition status; and then correcting the frequency and timing according to temperature extremes to generate target switching level commands. In this invention, by denoising the temperature sequence and extracting response start information based on valve opening times, the hysteresis characteristics of gaseous and liquid nitrogen are quantified. Phase adjustment is based on hysteresis differences, enabling coordinated matching of heating and cooling in the time dimension. Furthermore, the method combines duty cycle and energy competition status to identify and dynamically allocate valve rhythms, suppressing energy alternation interference. Simultaneously, the control cycle frequency is corrected based on temperature extremes, making the adjustment process smooth and stable and improving temperature control accuracy and environmental consistency.

Description

Technical Field

[0001] This invention relates to the field of refrigeration technology, and in particular to a method for temperature control of gaseous nitrogen heating and liquid nitrogen refrigeration in a thermal vacuum environment simulating a heat sink. Background Technology

[0002] The field of refrigeration technology mainly involves engineering technology systems that achieve heat transfer and temperature control through artificial means. It covers various implementation paths such as compression refrigeration, absorption refrigeration, throttling expansion refrigeration, and cryogenic working fluid refrigeration. Its core aspects include the phase change heat transfer process of the refrigeration medium, heat exchanger structural design, heat transfer path control, and temperature regulation strategies. In scenarios such as aerospace environment simulation, electronic device testing, and material performance testing, it is often necessary to construct a stable and adjustable thermal environment under closed or vacuum conditions to meet the test requirements of different temperature ranges.

[0003] One traditional method for simulating a heat sink in a thermal vacuum environment involves using gaseous nitrogen heating and liquid nitrogen cooling for temperature control. This method involves setting up a heat sink structure inside a vacuum container, introducing gaseous nitrogen as a heating medium and transporting it through pipelines to the heating channel for convective heat exchange, while simultaneously using liquid nitrogen to enter the cooling pipeline through a throttling device and vaporize and absorb heat on the heat exchange surface. Combined with temperature sensors collecting heat sink surface temperature data, the opening of the gaseous nitrogen flow valve and the liquid nitrogen supply are adjusted according to the set temperature range to achieve alternating temperature rise and fall control of the heat sink.

[0004] Existing gas nitrogen heating and liquid nitrogen cooling temperature control methods rely on valve opening and flow rate regulation to achieve temperature rise and fall. Temperature feedback is affected by sensor noise and environmental disturbances, resulting in fluctuations. The response process lacks quantitative identification of hysteresis characteristics, leading to time misalignment in the interaction between gas nitrogen and liquid nitrogen. The temperature change of the heat sink surface shows an oscillating trend. It is difficult to accurately match the rhythm of heat input and absorption during the adjustment process. Long-term operation is prone to energy alternation conflict, resulting in decreased temperature control stability, frequent temperature overshoot or slow recovery, and thus affecting the consistency and repeatability of the test environment. Summary of the Invention

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink, comprising the following steps: S1: Collect the temperature sensor voltage signal of the simulated heat sink surface in the thermal vacuum environment to obtain the temperature sequence, record the opening time of the gas nitrogen solenoid valve and the liquid nitrogen injection valve to obtain the opening time of the gas nitrogen valve and the liquid nitrogen valve, and denoise the temperature sequence to generate a denoised temperature sequence. S2: Based on the opening time of the gas nitrogen valve and the liquid nitrogen valve, retrieve the denoised temperature sequence to extract the start time of the gas nitrogen and liquid nitrogen response, calculate the gas nitrogen response lag time in combination with the gas nitrogen valve opening time, calculate the difference between the liquid nitrogen valve opening time and generate the liquid nitrogen response lag time. S3: Obtain the lag judgment interval sequence based on the preset time interval and calculate the gas nitrogen lag level with the gas nitrogen response lag duration; calculate the liquid nitrogen lag level with the liquid nitrogen response lag duration and the lag judgment interval sequence and compare it with the gas nitrogen lag level to generate a phase adjustment identifier. S4: Extract preset control parameters to obtain the control cycle duration and calculate the pulse duty cycle through a dual-layer PID control model. Classify the pulse duty cycle to calculate the energy competition state identifier and analyze the timing offset with the phase adjustment identifier. Combine feedforward compensation to allocate the valve action duration and generate basic valve control commands. S5: Retrieve the time of the extreme temperature point on the surface of the simulated heat sink in the thermal vacuum environment from the denoised temperature sequence, calculate the pulse frequency correction parameter based on the extreme temperature point time, and perform frequency and timing adjustment calculation on the basic valve control command through the pulse frequency correction parameter to generate the target switching level command.

[0006] As a further embodiment of the present invention, the denoised temperature sequence includes temperature data points, time stamps, and noise filtering indicators; the liquid nitrogen response hysteresis duration includes response start time, valve opening time difference, and hysteresis amount; the phase adjustment indicator includes gas nitrogen hysteresis level, liquid nitrogen hysteresis level, and level comparison result; the basic valve control command includes pulse duty cycle, energy competition status indicator, timing offset, and valve action duration; and the target switching level command includes frequency correction parameters, timing adjustment result, and control signal level.

[0007] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Acquire temperature sensor voltage signals from the simulated heat sink surface in a thermal vacuum environment to obtain temperature sequences and align them with time; record the opening times of the gas nitrogen solenoid valve and liquid nitrogen injection valve to obtain the opening times of the gas nitrogen valve and liquid nitrogen valve; map the voltages of multiple sampling points with a unified time identifier to obtain a time-aligned temperature and voltage sequence. S102: Based on the time-aligned temperature-voltage sequence, the voltage difference between adjacent sampling points is compared with a preset voltage mutation threshold, the sampling points corresponding to the positions exceeding the voltage mutation threshold are marked, and the neighborhood mean is replaced for the marked sampling points to obtain the abnormal suppression temperature sequence. S103: Based on the abnormal suppression temperature sequence, a sliding window weighting operation is performed on the continuous sampling points. The multiple sampling points within the window are weighted and summed according to the weight coefficient of the distance from the center point, and the summing result is used to replace the sampling value at the center of the window to generate a denoised temperature sequence.

[0008] As a further aspect of the present invention, the voltage mutation threshold is determined by acquiring the reference temperature sensor voltage signal of the heat sink when the gas nitrogen solenoid valve and liquid nitrogen injection valve are not turned on, obtaining a reference voltage sequence, calculating the standard deviation of the voltage values ​​at multiple sampling points in the reference voltage sequence, obtaining the reference voltage standard deviation, and multiplying the reference voltage standard deviation with a preset tolerance coefficient.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the opening time of the nitrogen valve and the denoised temperature sequence, perform time index matching, calculate the difference between the timestamps of multiple sampling points in the sequence and the opening time of the nitrogen valve, select the turning point where the difference sign changes from negative to positive as the response judgment point and mark the timestamp to obtain the nitrogen response start time point. S202: Construct a time difference sequence based on the start time of the gas nitrogen response and the opening time of the gas nitrogen valve, perform a subtraction operation on the corresponding time values ​​of the two and record the difference result, unify the dimensions of the difference result, and obtain the gas nitrogen response lag time. S203: Based on the liquid nitrogen valve opening time and the denoised temperature sequence, perform a synchronization time retrieval, calculate the difference between the sequence timestamp and the liquid nitrogen valve opening time point by point, extract the time of the first sampling point where the difference turns from negative to positive and subtract it from the liquid nitrogen valve opening time to obtain the liquid nitrogen response lag time.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Obtain the preset time interval parameter sequence, sort the boundary values ​​of multiple time periods in order, calculate the interval length of the difference between adjacent boundaries, map multiple intervals into a lag decision interval label set according to the time span, and then encode all interval labels to establish lag decision interval sequence data. S302: Based on the lag determination interval sequence data, call the gas nitrogen response lag time, compare it with the upper and lower limits of multiple intervals and locate the interval index to which it belongs, and at the same time perform the same interval matching on the liquid nitrogen response lag time, respectively map the matching index to the level identifier, and obtain the gas nitrogen and liquid nitrogen lag level group. S303: Based on the gas nitrogen and liquid nitrogen hysteresis level group, the difference between the gas nitrogen and liquid nitrogen level codes is determined, and the difference sign and amplitude are mapped to the identifier. The determination result is converted into the corresponding phase category code sequence to obtain the phase adjustment identifier.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Extract preset control parameters to obtain control cycle duration, retrieve gas nitrogen valve opening time and liquid nitrogen valve opening time, and calculate the ratio and compensation conversion with control cycle duration through a dual-layer PID control model. Then, serialize the proportion of gas nitrogen valve and liquid nitrogen valve opening duration to generate pulse duty cycle data. S402: Based on the pulse duty cycle data, the proportion of gas nitrogen valve and liquid nitrogen valve are input into the decision tree model. According to the internal hierarchical judgment rules of the model, the proportion is compared with the preset condition boundary value layer by layer. Based on the comparison results, the judgment path is selected and the corresponding classification attribute code is extracted to generate an energy competition state identifier. S403: Based on the energy competition status identifier, call the phase adjustment identifier, verify the trigger sequence and phase adjustment point position corresponding to the status identifier, calculate the trigger timing offset, and allocate the action duration of multiple valves according to the offset and feedforward compensation to generate basic valve control commands.

[0012] As a further embodiment of the present invention, the decision tree model consists of a root-level condition decision node, a multi-level branch addressing rule base, and a terminal state classification label set.

[0013] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the denoised temperature sequence, extract the time series of extreme temperature points on the surface of the heat sink in the simulated thermal vacuum environment. Perform adjacent difference operations on the temperature at multiple sampling times in the sequence. Mark the time index corresponding to the difference exceeding the preset temperature change threshold and reconstruct it in time order to generate a temperature extreme value time series index set. S502: Extract the time interval sequence of adjacent extreme points based on the temperature extreme value time series index set, perform a reciprocal transformation on the time interval sequence and perform a difference operation with the preset control cycle reference frequency, map the difference result to the corresponding interval in the frequency correction coefficient table and assign coefficient values ​​to generate pulse frequency correction parameters. S503: Based on the pulse frequency correction parameter, call the original pulse frequency and timing sequence in the basic valve control instruction, perform frequency correction on the product operation of the original pulse frequency, and resample and rearrange the timing sequence according to the period length corresponding to the correction frequency to generate the target switching level instruction.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by denoising the temperature sequence and extracting the response start information at the valve opening time, the hysteresis characteristics of gaseous nitrogen and liquid nitrogen are quantitatively characterized. Based on the hysteresis difference, a phase adjustment basis is formed, enabling the heating and cooling processes to achieve coordinated matching in the time dimension. At the same time, by combining the duty cycle and energy competition state identification, the valve action rhythm is dynamically allocated to suppress the phenomenon of alternating energy interference. Furthermore, the control cycle frequency is corrected by the temperature extreme value change, making the temperature regulation process smoother and more stable, and improving the temperature control accuracy and environmental consistency of the heat sink surface. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] Please see Figure 1 This invention provides a method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a simulated heat sink under thermal vacuum conditions, comprising the following steps: S1: Collect the temperature sensor voltage signal of the simulated heat sink surface in the thermal vacuum environment to obtain the temperature sequence, record the opening time of the gas nitrogen solenoid valve and the liquid nitrogen injection valve to obtain the opening time of the gas nitrogen valve and the liquid nitrogen valve, and denoise the temperature sequence to generate a denoised temperature sequence. S2: Extract the gas nitrogen and liquid nitrogen response start times from the denoised temperature sequence based on the opening time of the gas nitrogen valve and liquid nitrogen valve, calculate the gas nitrogen response lag time in combination with the gas nitrogen valve opening time, calculate the difference between the liquid nitrogen valve opening time and generate the liquid nitrogen response lag time. S3: Based on the preset time interval, obtain the lag judgment interval sequence and calculate the gas nitrogen lag level with the gas nitrogen response lag time. Calculate the liquid nitrogen lag level with the liquid nitrogen response lag time and the lag judgment interval sequence and compare it with the gas nitrogen lag level to generate a phase adjustment identifier. S4: Extract preset control parameters to obtain control cycle duration and calculate pulse duty cycle through dual-layer PID control model. Classify pulse duty cycle to calculate energy competition status identifier and analyze timing offset with phase adjustment identifier. Combine feedforward compensation to allocate valve action duration and generate basic valve control command. S5: Retrieve the temperature extreme point time of the heat sink surface in the thermal vacuum environment simulation by retrieving the denoised temperature sequence. Based on the temperature extreme point time, calculate the pulse frequency correction parameter for the control cycle frequency correction. Calculate the frequency and timing adjustment of the basic valve control command using the pulse frequency correction parameter to generate the target switching level command.

[0020] The denoised temperature sequence includes temperature data points, time stamps, and noise filtering indicators. The liquid nitrogen response hysteresis includes response start time, valve opening time difference, and hysteresis amount. The phase adjustment indicator includes gas nitrogen hysteresis level, liquid nitrogen hysteresis level, and level comparison result. The basic valve control command includes pulse duty cycle, energy competition status indicator, timing offset, and valve action duration. The target switching level command includes frequency correction parameters, timing adjustment result, and control signal level.

[0021] Please see Figure 2 The specific steps of S1 are as follows: S101: Acquire temperature sensor voltage signals from the simulated heat sink surface in a thermal vacuum environment to obtain temperature sequences and align them with time; record the opening times of the gas nitrogen solenoid valve and liquid nitrogen injection valve to obtain the opening times of the gas nitrogen valve and liquid nitrogen valve; map the voltages of multiple sampling points with a unified time identifier to obtain a time-aligned temperature and voltage sequence. A high-precision copper constantan thermocouple sensor network deployed on the heat sink surface of the thermal vacuum test chamber continuously reads the voltage signals from the temperature sensors on the heat sink surface. The data acquisition frequency is strictly set to 10 times per second. During the synchronization phase of reading continuous voltage signals, the high-level control command signal sent by the controller to the valve is extracted. During the extraction operation, the level transition state in the electrical circuit is continuously monitored, and the precise trigger moment of the rising edge generated by the closure of the gas nitrogen solenoid valve electrical contacts is recorded, along with the trigger moment of the same rising edge appearing at the control terminal of the liquid nitrogen injection valve. This trigger moment is converted into a standard timestamp format based on a unified slice clock source to obtain the opening time of the gas nitrogen valve and the liquid nitrogen valve. In specific execution, 10,000 continuously acquired discrete sampling point voltage data are extracted, and each voltage data is individually bound to a timestamp generated by a high-precision timer. Under the binding operation, the voltage fluctuation data segments before and after the trigger moment of the gas nitrogen solenoid valve and the liquid nitrogen injection valve are extracted, and the time stamp is aligned with the reference. Under this operation, a direct mapping relationship between the multi-sampling point voltage and the unified time stamp is established. For example, the timestamp for the opening time of the gas nitrogen valve is recorded as the 5000th millisecond of the current test cycle, and the timestamp for the opening time of the liquid nitrogen valve is recorded as the 8000th millisecond. Under the same time axis reference, the voltage value of the corresponding heat sink surface sampling point is read as 2.5 volts at the 5000th millisecond. All discretely distributed time stamp values ​​and their corresponding voltage values ​​are packaged and encapsulated to form a two-dimensional data column strictly arranged in chronological order, resulting in a time-aligned temperature-voltage sequence.

[0022] S102: Based on the time-aligned temperature-voltage sequence, the voltage difference between adjacent sampling points is compared with a preset voltage mutation threshold. The sampling points corresponding to the positions exceeding the voltage mutation threshold are marked, and the neighborhood mean is replaced for the marked sampling points to obtain the abnormal suppression temperature sequence. Based on the obtained time-aligned temperature-voltage sequence, the voltage value of the currently processed sampling point is extracted one by one, and the voltage value of the previous normal sampling point is extracted simultaneously. An absolute difference calculation is performed on the two voltage data points to obtain the absolute value of the voltage deviation between them. Then, this absolute value of the voltage deviation is directly compared with a pre-set voltage mutation threshold. The preset voltage mutation threshold is obtained based on the accumulation of data from previous no-load thermal vacuum tests. In the no-load test, the acquisition unit runs continuously and records the absolute voltage deviation data of adjacent sampling points under normal conditions for 100,000 consecutive days. A sorting operation is performed to extract the maximum deviation value, and this maximum deviation value is multiplied by a safety redundancy factor of 1.2 to generate the preset voltage mutation threshold. The measured maximum normal deviation is 0.1 volts, and substituting this into the multiplication operation directly yields a preset voltage mutation threshold of 0.12 volts. The actual calculated deviation is compared with this preset voltage mutation threshold. When the comparison result determines that the actual deviation is greater than the preset voltage mutation threshold of 0.12 volts, a unique logical label is added to the corresponding position in the time-aligned temperature-voltage sequence, marking the sampling point exceeding the voltage mutation threshold. After identifying all anomaly locations, the system extends outwards from a single anomaly, extracting the voltage values ​​of the immediately preceding and following normal sampling points in the time series (the difference between the preceding and following normal sampling points is used to determine if the following point is normal). The arithmetic mean of these two normal sampling point voltage values ​​is then calculated, and the final arithmetic mean directly overwrites and replaces the original erroneous voltage value of the anomaly. For example, if the voltage at the 100th sampling point in the time-aligned temperature-voltage sequence is 2.50 volts, and the voltage at the 101st sampling point increases dramatically to 2.80 volts due to electromagnetic interference, the absolute difference between this and the preceding normal sampling point (i.e., the 100th point) is calculated to be 0.30 volts. Since 0.30 volts is greater than the preset voltage mutation threshold of 0.12 volts, the 101st sampling point is marked as an anomaly. The 102nd sampling point was then evaluated. Its original voltage was 2.52 volts. This was compared with the 2.50 volts of the nearest normal sampling point (i.e., the 100th point). The absolute deviation was 0.02 volts, which was less than the preset threshold of 0.12 volts, and it was marked as a normal point. The arithmetic mean of the 2.50 volts from the 100th sampling point and the 2.52 volts from the 102nd normal sampling point was then calculated, yielding an arithmetic mean of 2.51 volts. This 2.51 volts was then forcibly written into the storage location of the 101st sampling point, completing the neighborhood mean replacement and obtaining the anomaly suppression temperature sequence.

[0023] Table 1 Voltage Signal Anomaly Judgment Table

[0024] Table 1 shows the baseline comparison status and anomaly detection results of three consecutive sampling points before and after the introduction of extreme value determination.

[0025] S103: Based on the abnormal suppression temperature sequence, a sliding window weighted operation is performed on the continuous sampling points. The multiple sampling points within the window are weighted and summed according to the weight coefficient of the distance from the center point. The summation result is used to replace the sampling value at the center of the window to generate a denoised temperature sequence. The system receives the abnormal suppression temperature sequence and extracts a sliding data window of fixed length with 5 sampling points. For each isolated sampling point within the data window, its corresponding timestamp value is extracted and subtracted from the timestamp value corresponding to the center point of the window. The absolute value of the subtraction is then taken to obtain the absolute value of the time difference. Further, the natural constant calculation logic is invoked, using the product of a preset baseline attenuation coefficient and this absolute value of the time difference as a negative power parameter, which is then substituted into an exponential function to calculate the preliminary weight coefficient for each sampling point. The baseline attenuation coefficient is determined through a grid search experiment on 1000 independent thermal cycling test samples, with a test range of 0.1 to 1.0. The results show that when the value is 0.5, the signal-to-noise ratio after denoising reaches a maximum of 35 dB, therefore the baseline attenuation coefficient is forcibly set to 0.5. After calculating all the preliminary weight coefficients, the 5 preliminary weight coefficients within the window are summed to obtain the total window weight. Subsequently, the preliminary weight coefficient of each sampling point is divided by the total window weight to derive the normalized weight coefficient for each sampling point. Finally, the voltage values ​​of each sampling point within the sliding window are multiplied by their corresponding normalized weight coefficients, and all multiplications are summed to output the final weighted sum. This weighted sum is then used to directly replace the center sampling value of the window. For example, the current data window contains the 101st to 105th sampling points generated in the previous steps, with the center point being the 103rd sampling point, and the sampling time interval is 1 second. The voltage value after replacing the 101st sampling point is 2.55 volts, and the voltages of the other four points are 2.58 volts, 2.60 volts, 2.62 volts, and 2.64 volts, respectively. The absolute value of the time difference for the 101st sampling point is 2 seconds, and its initial weight coefficient is calculated as the negative 1 power of the natural constant, with a value of approximately 0.368. Similarly, the initial weight coefficients for the other four points are calculated as 0.606, 1.000, 0.606, and 0.368, respectively. The total weight of the window is approximately 2.948. Dividing the initial weight of the 101st point (0.368) by 2.948 yields a normalized weight coefficient of approximately 0.125. The normalized weights for the remaining four points are approximately 0.206, 0.339, 0.206, and 0.125, respectively. Multiplying each of the five voltage values ​​by its respective normalized weight coefficient and summing the results, the weighted sum is 2.601 volts. This value replaces the original 2.60 volts of the 103rd sampling point. By iterating through the entire sequence using a sliding window, a denoised temperature sequence is generated.

[0026] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the opening time of the nitrogen valve and the denoised temperature sequence, time index matching is performed. The difference between the timestamps of multiple sampling points in the sequence and the opening time of the nitrogen valve is calculated. The turning point where the difference sign changes from negative to positive is selected as the response judgment point and the timestamp is marked to obtain the nitrogen response start time point. Based on the acquired nitrogen valve opening time and the denoised temperature sequence, a time retrieval window containing 100 data slices is established on the continuous time axis of the denoised temperature sequence. The starting edge of this time retrieval window is aligned with the initial zero point of the denoised temperature sequence, and time index matching is performed by sliding forward at a step size of 10 slices per second (i.e., a fixed time span interval of 100 milliseconds). During the sliding process, the timestamp data corresponding to each independent sampling point in the denoised temperature sequence is extracted sequentially. The extracted timestamps of multiple sampling points are subtracted from the pre-stored nitrogen valve opening time to calculate the time deviation value between the two. The generated time deviation values ​​are read one by one, and the positive or negative sign bit attribute of the value is extracted. The change state of the sign bit attribute is continuously monitored. When the time deviation values ​​of two consecutive sampling points are detected to change from negative to positive, the specific position where a positive value is generated is locked. The position where a positive value is generated is established as the response determination point, and a binary flag bit 1 is appended to the end of the corresponding data line to complete the timestamp marking operation. The absolute time value of the marked timestamp is extracted, and the nitrogen response start time point is output. For example, the baseline value for the nitrogen valve opening time recorded in the previous sequence is 4950 milliseconds. Five consecutive timestamps are extracted from the denoised temperature sequence: 4800 milliseconds, 4900 milliseconds, 5000 milliseconds, 5100 milliseconds, and 5200 milliseconds. These timestamps are subtracted from the 4950 millisecond opening time of the nitrogen valve, resulting in time deviations of -150 milliseconds, -50 milliseconds, +50 milliseconds, +150 milliseconds, and +250 milliseconds, respectively. In this sequence, the time deviation at 5000 milliseconds is +50 milliseconds, and its sign changes from negative to positive for the first time. This 5000th millisecond is established as the response determination point, marked with a 1 in its storage area, and the value at 5000 milliseconds is extracted to directly obtain the nitrogen response start time point.

[0027] S202: Construct a time difference sequence based on the start time of the gas nitrogen response and the opening time of the gas nitrogen valve, perform subtraction on the corresponding time values ​​of the two and record the difference result, unify the dimensions of the difference result and obtain the gas nitrogen response lag time; The system retrieves the output data for the start time of the nitrogen response and the opening time of the nitrogen valve. It allocates contiguous storage space in memory, using the opening time of the nitrogen valve as the starting coordinate and the start time of the nitrogen response as the ending coordinate, constructing a time difference sequence containing the characteristics of both time points. The system extracts the time value corresponding to the starting time of the nitrogen response from the ending coordinate, and simultaneously extracts the time value corresponding to the opening time of the nitrogen valve from the starting coordinate. It subtracts the time value corresponding to the opening time of the nitrogen valve from the starting time of the nitrogen response, performing a direct subtraction operation to generate the absolute time difference between the two time points. The system reads the current time unit attribute of this absolute time difference result and determines whether the current unit is the standard international unit, seconds. If the current unit is milliseconds or microseconds, the absolute time difference result is multiplied by the corresponding conversion factor to convert it to a standard real number in seconds, completing the unitization operation. Finally, the unitized time value is directly output as the nitrogen response lag duration. For example, substituting the nitrogen response start time value of 5000 milliseconds obtained in the previous steps, and substituting the nitrogen valve opening time reference value of 4950 milliseconds, subtracting 4950 milliseconds from 5000 milliseconds yields an absolute time difference of 50 milliseconds. Since the current unit is milliseconds, a conversion factor of 0.001 is used. Multiplying 50 milliseconds by 0.001 yields 0.050 seconds, thus unifying the units from milliseconds to seconds. This 0.050 seconds is directly output as the nitrogen response lag time. This 0.050 seconds represents the physical delay from when the nitrogen solenoid valve receives the command to when the actual gas reaches the temperature measurement node and triggers effective heat exchange; its value falls within the reasonable range of 0.040 seconds to 0.080 seconds for a typical gas path delay.

[0028] S203: Based on the liquid nitrogen valve opening time and the denoised temperature sequence, perform synchronization time retrieval, calculate the difference between the sequence timestamp and the liquid nitrogen valve opening time point by point, extract the time of the first sampling point where the difference turns from negative to positive and subtract it from the liquid nitrogen valve opening time to obtain the liquid nitrogen response lag time; The process retrieves the denoised temperature sequence data and the previously acquired liquid nitrogen valve opening time data. Using the liquid nitrogen valve opening time data as the retrieval anchor, a synchronous time retrieval operation is initiated in the global time index table of the denoised temperature sequence to extract the timestamp of each sampling point contained in the denoised temperature sequence. The liquid nitrogen valve opening time data is subtracted from each extracted sampling timestamp, and a point-by-point difference calculation is performed to generate a dynamic time deviation sequence containing positive and negative signs. The sign bits of each element in this dynamic time deviation sequence are traversed, and the specific position where the sign bit first shows a positive value is identified. The absolute time value in the original denoised temperature sequence corresponding to this position is extracted and determined as the first sampling time. The determined first sampling time is directly subtracted from the previously retrieved liquid nitrogen valve opening time, and the absolute time difference is obtained by performing a subtraction operation. This absolute time difference is then converted into a standard format in seconds, and the liquid nitrogen response lag time is output. For example, the previously recorded liquid nitrogen valve opening time is retrieved as 7920 milliseconds. Under synchronous retrieval, the timestamps of consecutive sampling points at 7800 ms, 7900 ms, 8000 ms, and 8100 ms in the denoised temperature sequence are extracted. These timestamps are subtracted by 7920 ms respectively, generating time deviations of -120 ms, -20 ms, +80 ms, and +180 ms. The first positive value is detected at +80 ms, and the corresponding original sampling point time, i.e., 8000 ms, is extracted. Subtracting 7920 ms from 8000 ms yields an absolute time difference of 80 ms. Multiplying 80 ms by a unit conversion factor of 0.001 converts it to 0.080 seconds, which is output as the liquid nitrogen response lag time. This 0.080-second value reflects the fluid inertial delay of the liquid working medium filling and spraying in the pipeline, conforming to the calibration range of 0.050 seconds to 0.100 seconds for the actual physical response of cryogenic pipelines.

[0029] Please see Figure 4 The specific steps of S3 are as follows: S301: Obtain the preset time interval parameter sequence, sort the boundary values ​​of multiple time periods in order, calculate the interval length of the difference between adjacent boundaries, map multiple intervals into a lag decision interval label set according to the time span, and then encode all interval labels to establish lag decision interval sequence data. The system reads a preset time interval parameter sequence stored in the internal storage medium. The raw data in this sequence is based on statistics of the response time from fluid valve actuation to temperature drop at the temperature measurement node during the past 50 no-load cycle tests in the thermal vacuum chamber. Pre-defined multi-time-segment boundary values ​​are extracted from the sequence, specifically 0 seconds, 0.060 seconds, 0.100 seconds, and 0.150 seconds. These multi-time-segment boundary values ​​are then sorted in ascending order based on their numerical values, generating a strictly ordered set of boundary values ​​from low to high. Adjacent boundary values ​​in this ordered set are then extracted sequentially, and a subtraction operation is performed: the preceding time boundary value is subtracted from the following time boundary value, and the absolute difference between the two is used as the interval length for the corresponding segment. For example, extracting the first boundary value of 0 seconds and the second boundary value of 0.060 seconds, subtracting 0 seconds from 0.060 seconds yields the first interval length of 0.060 seconds. Similarly, extracting 0.060 seconds and subtracting 0.100 seconds yields the second interval length of 0.040 seconds, and extracting 0.100 seconds and subtracting 0.150 seconds yields the third interval length of 0.050 seconds. Based on the time span distribution range corresponding to each independent interval, the calculated multiple numerical intervals are bound one-to-one with fixed state description terms, directly generating a set of lag judgment interval labels. The operation rule is to forcibly map the interval from 0 seconds to 0.060 seconds to the ultra-fast response interval label, the interval from 0.060 seconds to 0.100 seconds to the standard response interval label, and the interval from 0.100 seconds to 0.150 seconds to the slow response interval label. For all the character-type interval labels generated above, perform integer value assignment operations to convert them into easily manageable digital states. Directly assign the high-speed response interval label code to the integer 1, the standard response interval label code to the integer 2, and the slow response interval label code to the integer 3. Integrate all the codes and interval correspondences to establish a complete and discrete hysteresis decision interval sequence data output.

[0030] S302: Based on the lag judgment interval sequence data, call the gas nitrogen response lag time, compare it with the upper and lower limits of multiple intervals and locate the interval index. At the same time, perform the same interval matching for the liquid nitrogen response lag time, map the matching index to the level identifier respectively, and obtain the gas nitrogen and liquid nitrogen lag level group. The system receives the established lag determination interval sequence data and simultaneously retrieves the gaseous nitrogen response lag duration calculated in the previous stage. It extracts the lower and upper time limits for each independent interval contained in the lag determination interval sequence data. The retrieved gaseous nitrogen response lag duration is then compared with the extracted lower and upper limits for each interval to determine which numerical range it falls within. In the case of a gaseous nitrogen response lag duration of 0.050 seconds, this value is substituted into the comparison logic. It is determined that 0.050 seconds is greater than the lower limit of 0 seconds and less than the upper limit of 0.060 seconds. Based on this comparison result, it is determined that 0.050 seconds perfectly matches the distribution characteristics of the first interval, and its interval index is directly set to 1. Next, the liquid nitrogen response lag duration extracted in the previous stage is retrieved, and the same interval matching and comparison operation is performed. Substituting the previously obtained liquid nitrogen response lag time value of 0.080 seconds, we compare 0.080 seconds with the preset interval boundaries. Since 0.080 seconds is greater than the lower limit of the second interval (0.060 seconds) and less than the upper limit (0.100 seconds), the liquid nitrogen response lag time is determined to fall within the second interval range, and its matching index number is directly located as 2. After obtaining the interval indices for the gas and liquid phases, we extract the located gas nitrogen matching index number 1 and liquid nitrogen matching index number 2, and map these two independent index numbers to the corresponding level status identifier parameters. Specifically, we assign the gas nitrogen index 1 directly to the level identifier parameter of the gas nitrogen channel, recording it as gas nitrogen level identifier 1, and assign the liquid nitrogen index 2 to the level identifier parameter of the liquid phase channel, recording it as liquid nitrogen level identifier 2. We then concatenate and encapsulate all the generated independent level identifier parameters according to the time sequence and channel category to construct a set array containing the time sequence status of the gas and liquid dual channels, thus obtaining the gas nitrogen and liquid nitrogen lag level groups.

[0031] Table 2 Matching Table of Hysteresis Intervals in Gas-Liquid Channels

[0032] Table 2 shows the actual values ​​of the hysteresis time of the gas-liquid dual-channel fluid response and the specific level labels generated after numerical comparison.

[0033] S303: Based on the hysteresis level group of gaseous nitrogen and liquid nitrogen, the difference between the gaseous nitrogen and liquid nitrogen level codes is determined, and the difference sign and amplitude are mapped to the identifier. The determination result is converted into the corresponding phase category code sequence to obtain the phase adjustment identifier. The generated dataset of hysteresis levels for gaseous and liquid nitrogen is retrieved. For each element encapsulated within this dataset, the level codes for gaseous and liquid nitrogen are extracted separately. A direct algebraic difference operation is performed on these two discrete level codes. Specifically, the gaseous nitrogen level code is forcibly subtracted from the liquid nitrogen level code, and the difference is calculated. The absolute value and sign of this difference are then extracted. Substituting the previously obtained parameter values ​​(gase nitrogen level code 1, liquid nitrogen level code 2), the subtraction operation (1 minus 2) directly yields a difference result of -1. This difference result is analyzed, and its sign attribute (negative) and absolute amplitude value (1) are extracted. Based on the internal phase identifier mapping rule base, the extracted sign attribute and absolute amplitude value are rigorously compared with the rule base conditions. The rule base is set such that if the difference is negative and the amplitude is 1, a conditional branch for gas phase leading by 1 order is triggered; if the difference is positive and the amplitude is 1, a conditional branch for liquid phase leading by 1 order is triggered; and if the difference is 0, a conditional branch for complete gas-liquid synchronization is triggered. Since the currently acquired sign is negative and the amplitude is 1, the first conditional branch instruction is directly triggered and executed, mapping and outputting it as a thermodynamic phase identifier representing the early arrival of the gas phase fluid. After generating the initial text identifier, the protocol conversion program of the underlying communication interface is called to forcibly convert the text state judgment result into a binary code string that can be directly read and parsed by the lower-level machine. For example, the state is translated into the corresponding binary phase category encoding sequence 1001. After translating and outputting this binary sequence, the phase adjustment identifier for direct use by the solenoid valve group is finally obtained. This sequence value directly reflects the effective time difference of heat exchange between gaseous nitrogen and liquid nitrogen at the heat sink interface in the current pipeline, guiding subsequent control loops to actively compensate for the opening timing of the gas-liquid injection valves.

[0034] Please see Figure 5 The specific steps of S4 are as follows: S401: Extract preset control parameters to obtain control cycle duration, retrieve gas nitrogen valve opening time and liquid nitrogen valve opening time, and calculate the ratio and compensation conversion with control cycle duration through a dual-layer PID control model. Then, serialize the proportion of gas nitrogen valve and liquid nitrogen valve opening duration to generate pulse duty cycle data. The system reads the preset control parameter sequence from the underlying controller's flash memory and extracts the control cycle duration value for the high-frequency action of the solenoid valve. It then retrieves the actual opening time of the gas nitrogen valve and the liquid nitrogen valve for the current cycle, obtained from the previous steps. To overcome the thermal inertia and nonlinearity of the heat sink, a two-layer PID control model is established. The outer loop is a temperature-position PID controller, and the inner loop is a temperature change rate incremental PID controller. First, the difference between the current actual temperature and the set target temperature in the denoised temperature sequence is extracted and input into the outer loop temperature-position PID controller, which outputs the desired target temperature change rate. Then, the derivative of the denoised temperature sequence is calculated to obtain the actual temperature change rate, which is then subtracted from the target temperature change rate and input into the inner loop incremental PID controller. Through iterative calculation, the duty cycle dynamic compensation coefficient is directly output. Using the retrieved gas nitrogen valve opening time as the dividend and the aforementioned control cycle duration value as the divisor, a division operation is performed to obtain the initial proportion of the gas nitrogen valve opening time. This initial proportion is then multiplied and compensated with the duty cycle dynamic compensation coefficient to obtain the final proportion of the gas nitrogen valve opening time. Similarly, the liquid nitrogen valve opening time is used as the dividend, and the control cycle duration is used as the divisor for division. After correction by the duty cycle dynamic compensation coefficient, the final liquid nitrogen valve opening time percentage is obtained. The gas nitrogen valve opening time percentage is placed first, and the liquid nitrogen valve opening time percentage is placed second. A one-dimensional array encapsulation and serialization operation are performed to directly output the pulse duty cycle data. For example, if the control cycle duration is extracted to be 0.100 seconds, and the liquid nitrogen valve opening time recorded in the previous step is 0.030 seconds, the outer loop PID calculates the target cooling rate based on the current temperature deviation, and the inner loop PID calculates the required duty cycle dynamic compensation coefficient of 1.10 based on the actual cooling rate deviation. Dividing the 0.030 seconds of the liquid nitrogen valve by 0.100 seconds gives an initial percentage of 0.30, which is multiplied by the compensation coefficient 1.10 to calculate the final liquid nitrogen valve opening time percentage of 0.33. The gas nitrogen valve is calculated similarly to 0.00. The numbers 0.00 and 0.33 are arranged and encapsulated in sequence to generate a pulse duty cycle data set containing the two elements 0.00 and 0.33.

[0035] S402: Based on pulse duty cycle data, the proportion of gas nitrogen valve and liquid nitrogen valve is input into the decision tree model. According to the internal hierarchical judgment rules of the model, the proportion is compared with the preset condition boundary value layer by layer. Based on the comparison results, the judgment path is selected and the corresponding classification attribute code is extracted to generate an energy competition state identifier. The system receives a pulse duty cycle data set containing the percentage of time the gas nitrogen valve is open and the percentage of time the liquid nitrogen valve is open. This input data undergoes extreme value limiting and data type casting preprocessing to ensure that the percentage values ​​are single-precision floating-point numbers within the standard range of 0 to 1. The preprocessed percentages of gas nitrogen valve opening time and liquid nitrogen valve opening time are then input into a pre-constructed decision tree model. This decision tree model contains a root node for receiving initial data, two hidden intermediate nodes as conditional branches, and four leaf nodes that output the final classification results. Nodes at each level are directly connected via unidirectional binary branching logic. At the root node, the percentage of time the gas nitrogen valve is open is retrieved and compared with a first preset conditional boundary value using a greater than or less than comparison logic. If the value is greater than, the system proceeds to the right-hand heavily competitive branch node; if the value is less than or equal to, the system proceeds to the left-hand lightly competitive branch node. Upon reaching the branch node, the percentage of liquid nitrogen valve opening time is retrieved and compared twice with the second preset condition boundary value. Based on the comparison result, the process directly points to the leaf node at the end, extracting the preset integer classification attribute code within the leaf node as the energy competition state identifier output. For example, substituting the previously obtained percentage of gas nitrogen valve opening time of 0.20 and liquid nitrogen valve opening time of 0.30, the first preset condition boundary value is set to 0.25, and the second preset condition boundary value is set to 0.40. This boundary value is obtained from the statistical data of the first 200 times when both valves were opened simultaneously, causing a sudden jump in pipeline gas resistance pressure. At the root node, 0.20 is compared with 0.25, and it is determined that 0.20 is less than 0.25, thus proceeding to the left branch node. At the left branch node, the percentage of liquid nitrogen valve opening time of 0.30 is extracted and compared with 0.40, and it is determined that 0.30 is less than 0.40, triggering the entry into the leftmost uncompetitive state leaf node. Extract the fixed classification attribute code 0 from the leaf node, and directly generate and output the energy competition state identifier as 0. This value of 0 means that the probability of the current valve action causing pressure competition in the fluid pipeline network is extremely low.

[0036] S403: Call the phase adjustment flag according to the energy competition status flag, verify the trigger sequence and phase adjustment point position corresponding to the status flag, calculate the trigger timing offset, and allocate the action duration of multiple valves according to the offset and feedforward compensation to generate basic valve control commands. The output integer 0 is retrieved as the energy competition state identifier, and the phase adjustment identifier generated in the previous stage, namely the binary sequence 1001, is extracted. The binary sequence 1001 is decoded bit by bit; the first digit 1 represents gas-phase priority action, and the last digit 1 represents setting the phase adjustment point position at the beginning of the control cycle. Based on this decoding rule, the specific triggering sequence is determined to be that the gas nitrogen valve is triggered early and the liquid nitrogen valve is triggered late, and the adjustment point position is anchored at the absolute zero moment of a single cycle. The internally pre-stored basic phase compensation time value and the multiplication weight coefficient bound to the energy competition state identifier are called. The basic phase compensation time value is multiplied by the multiplication weight coefficient, and the multiplication operation is performed to obtain the specific triggering timing offset value. Simultaneously, a feedforward compensation strategy is introduced, using the gas-liquid response lag level difference in the phase adjustment identifier as a feedforward interference observation variable, inputting it into a preset feedforward gain function to calculate the feedforward time correction. The system performs an early gas phase opening operation by subtracting the trigger timing offset from the original trigger time of the gas nitrogen valve, and a delayed liquid phase opening operation by adding the trigger timing offset to the original trigger time of the liquid nitrogen valve. The original multi-valve action duration is then redistributed by adding the feedforward time correction, ultimately generating a basic valve control command containing the new trigger time node and compensation duration. For example, after decoding and determining that early gas phase triggering is required, the basic phase compensation time value is set to 0.005 seconds, and the trigger timing offset is calculated to be 0.005 seconds. If cryogenic hysteresis is detected in the liquid nitrogen pipeline (due to phase adjustment flag analysis), the feedforward gain function calculates a feedforward time correction of +0.003 seconds. Adding 0.005 seconds to the original trigger point of the liquid nitrogen valve at 0.020 seconds yields a new trigger point at 0.025 seconds; and adding the +0.003-second feedforward to the original action duration. Through the combined action of feedforward compensation and dual-layer PID, not only is the back pressure peak staggered in the time domain, but the energy supply also fills the gap in cooling loss caused by pipeline lag in advance.

[0037] Please see Figure 6 The specific steps of S5 are as follows: S501: Extract the time series of extreme temperature points on the surface of the heat sink in the thermal vacuum environment simulation based on the denoised temperature sequence. Perform adjacent difference operation on the temperature at multiple sampling times in the sequence. Mark the time index corresponding to the difference exceeding the preset temperature change threshold and reconstruct it in time order to generate a temperature extreme value time series index set. The system receives temperature data from a network of high-precision copper-constantan thermocouple sensors inside the thermal vacuum chamber at a frequency of 10 times per second. This data is then processed using a sliding window weighted algorithm to generate a single-precision floating-point denoised temperature sequence. Temperature values ​​at each discrete sampling moment are extracted sequentially along the time axis. The current temperature value is compared with the temperature values ​​at the previous and next adjacent sampling moments. If the current temperature value is greater than both the previous and next adjacent temperatures, the current timestamp and its corresponding temperature value are extracted as a single data pair. All extracted data pairs are integrated to generate a time series of extreme temperature points on the heat sink surface in the thermal vacuum environment simulation. Along the generated time series, the temperature values ​​at the previous extreme point and the immediately following extreme point are extracted sequentially. The absolute difference between the temperature values ​​at the next extreme point and the previous extreme point is calculated, completing the adjacent difference operation. The preset temperature change threshold, fixed in the underlying memory, is read. This threshold is set to 2.5 degrees Celsius based on fatigue test data of the thermal stress deformation critical point of beryllium copper alloy, the core material of the cabin. The calculated absolute difference is compared with 2.5 degrees Celsius. If the difference is greater than 2.5 degrees Celsius, a high-risk physical fluctuation is determined to have occurred at that time point. Substituting the actual collected data, the first extreme point temperature is -150.0 degrees Celsius, occurring at the 50th sampling time point, and the immediately following extreme point temperature is -146.0 degrees Celsius, occurring at the 85th sampling time point. Subtracting -150.0 degrees Celsius from -146.0 degrees Celsius yields an absolute difference of 4.0 degrees Celsius. Since 4.0 degrees Celsius is greater than the critical threshold of 2.5 degrees Celsius, the array indices of the time dimension containing these two temperature values, namely 50 and 85, are directly extracted and marked as key points. All marked discrete-time index values ​​are placed into a 1D array, and a reconstruction operation based on ascending order of numerical values ​​from low to high is forced to be performed, directly outputting a temperature extreme value time series index set containing all fluctuation point time series.

[0038] S502: Extract the time interval sequence of adjacent extreme points based on the temperature extreme value time series index set, perform a reciprocal transformation on the time interval sequence and perform a difference operation with the preset control cycle reference frequency, map the difference result to the corresponding interval in the frequency correction coefficient table and assign coefficient values ​​to generate pulse frequency correction parameters. The output temperature extreme value time series index set is retrieved. This set encapsulates the array index values ​​of high-risk sampling points arranged in ascending chronological order. Two immediately adjacent time index values ​​within this set are extracted sequentially. Based on the data acquisition reference frequency of 10 times per second, the time span interval of a single sampling point is calculated to be 0.1 seconds. The number of interval sampling points is obtained by subtracting the previous time index value from the subsequent time index value. This number is then multiplied by 0.1 seconds to calculate the absolute time span between the two extreme points. All independently calculated time span values ​​are concatenated sequentially to generate a time interval sequence of adjacent extreme points. For each independent time span value in this time interval sequence, a division operation is performed using a constant 1 as the dividend and the time span value as the divisor to obtain the physical oscillation frequency value corresponding to that time span, completing the reciprocal transformation operation of the time interval sequence. The preset control cycle reference frequency of the thermodynamic control platform, calibrated at the factory, is extracted; its preset value is 10.0 Hz. Subtract the previously calculated physical oscillation frequency from the 10.0 Hz baseline value to obtain the absolute difference. Substitute the specific variables obtained earlier into the calculation, extract the adjacent indices 85 and 50, subtract 50 from 85 to obtain 35 interval sampling points, multiply 35 by 0.1 seconds to obtain the time interval between adjacent extreme points as 3.5 seconds. Divide the constant 1 by 3.5 seconds for reciprocal transformation, and retain two decimal places to obtain the current extreme physical oscillation frequency as 0.29 Hz. Subtract 0.28 Hz from the baseline frequency of 10.0 Hz to obtain the absolute difference calculation result as 9.72 Hz. Read the preset frequency correction coefficient table in the flash memory area, which was formulated by nonlinear fitting of 500 hours of liquid nitrogen injection compensation experimental data. The table defines a multiplication coefficient of 1.2 for absolute differences in the range of 0 Hz to 5.0 Hz, and a multiplication coefficient of 1.5 for differences in the range of 5.0 Hz to 10.0 Hz. The calculated difference result of 9.72 Hz is substituted into the coefficient table for comparison and addressing to determine that it falls exactly within the numerical range of 5.0 Hz to 10.0 Hz. The value of 1.5 bound to this range is directly extracted and the coefficient is assigned. 1.5 is stored as the final physical variable and the pulse frequency correction parameter is generated and sent to the bus driver layer.

[0039] S503: Based on the pulse frequency correction parameter, call the original pulse frequency and timing sequence in the basic valve control instruction, perform frequency correction on the product operation of the original pulse frequency, and resample and rearrange the timing sequence according to the period length corresponding to the correction frequency to generate the target switching level instruction; The system receives a calculated pulse frequency correction parameter of 1.5 and simultaneously extracts the original pulse frequency value encapsulated within the basic valve control instructions and the timing sequence corresponding to the gas-liquid dual-path valve actions from the underlying hardware execution stack. Using this original pulse frequency value as the multiplicand and the aforementioned pulse frequency correction parameter as the multiplier, the system directly performs a multiplication operation to calculate an updated target operating frequency value, completing the frequency correction processing for the underlying injection action control cycle. A reciprocal conversion calculation is then performed on this corrected target operating frequency, specifically by dividing the target operating frequency value by a constant 1, to derive the updated absolute time length of a single control cycle. The timing sequence from the original basic instructions is extracted, which records the theoretical opening duration of each valve within the original control cycle. The updated single control cycle length is divided by the original single control cycle length to obtain a time axis scaling factor. The original scheduled opening duration of each valve in the timing sequence is multiplied by the time axis scaling factor to calculate the corrected valve action duration. This new duration data is then overwritten into the new underlying drive control array according to the original action sequence, generating the final target switching level command. Substituting the preceding specific operating parameters for continuous calculation, the original pulse frequency value in the basic valve control command is extracted as 10.0 Hz, corresponding to an original control cycle length of 0.100 seconds. The original opening duration of the nitrogen valve within this cycle is 0.020 seconds. Multiplying 10.0 Hz by the pulse frequency correction parameter 1.5 yields a corrected pulse frequency of 15.0 Hz. Dividing this by a constant 1 gives a new single-cycle length of approximately 0.066 seconds for the corrected frequency. Dividing this new cycle length of 0.066 seconds by the original cycle length of 0.100 seconds yields a single-precision time axis scaling factor of 0.66. The original opening duration of the nitrogen valve, 0.020 seconds, was retrieved and multiplied by a scaling factor of 0.66. After resampling, the new opening duration of the nitrogen valve was calculated to be 0.013 seconds. The underlying timing status register bits were directly updated according to the effective high-level continuous pulse width of 0.013 seconds, completing the resampling and rearrangement of the time-domain pulse output, directly generating and sending the target switching level command to the relay array. The result of 0.013 seconds in this example indicates that during this high-frequency oscillation phase, the physical mass of a single fluid injection was adaptively compressed by 35%.

[0040] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for temperature control using gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink, characterized in that, Includes the following steps: S1: Collect the temperature sensor voltage signal of the simulated heat sink surface in the thermal vacuum environment to obtain the temperature sequence, record the opening time of the gas nitrogen solenoid valve and the liquid nitrogen injection valve to obtain the opening time of the gas nitrogen valve and the liquid nitrogen valve, and denoise the temperature sequence to generate a denoised temperature sequence. S2: Based on the opening time of the gas nitrogen valve and the liquid nitrogen valve, retrieve the denoised temperature sequence to extract the start time of the gas nitrogen and liquid nitrogen response, calculate the gas nitrogen response lag time in combination with the gas nitrogen valve opening time, calculate the difference between the liquid nitrogen valve opening time and generate the liquid nitrogen response lag time. S3: Obtain the lag judgment interval sequence based on the preset time interval and calculate the gas nitrogen lag level with the gas nitrogen response lag duration; calculate the liquid nitrogen lag level with the liquid nitrogen response lag duration and the lag judgment interval sequence and compare it with the gas nitrogen lag level to generate a phase adjustment identifier. S4: Extract preset control parameters to obtain the control cycle duration and calculate the pulse duty cycle through a dual-layer PID control model. Classify the pulse duty cycle, calculate the energy competition state identifier, and analyze the timing offset with the phase adjustment identifier. Combine the feedforward compensation to allocate the valve action duration and generate basic valve control commands.

2. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The denoised temperature sequence includes temperature data points, time stamps, and noise filtering indicators. The liquid nitrogen response hysteresis duration includes response start time, valve opening time difference, and hysteresis amount. The phase adjustment indicator includes gas nitrogen hysteresis level, liquid nitrogen hysteresis level, and level comparison result. The basic valve control command includes pulse duty cycle, energy competition status indicator, timing offset, and valve action duration.

3. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a simulated heat sink in a thermal vacuum environment according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Acquire temperature sensor voltage signals from the simulated heat sink surface in a thermal vacuum environment to obtain temperature sequences and align them with time; record the opening times of the gas nitrogen solenoid valve and liquid nitrogen injection valve to obtain the opening times of the gas nitrogen valve and liquid nitrogen valve; map the voltages of multiple sampling points with a unified time identifier to obtain a time-aligned temperature and voltage sequence. S102: Based on the time-aligned temperature-voltage sequence, the voltage difference between adjacent sampling points is compared with a preset voltage mutation threshold, the sampling points corresponding to the positions exceeding the voltage mutation threshold are marked, and the neighborhood mean is replaced for the marked sampling points to obtain the abnormal suppression temperature sequence. S103: Based on the abnormal suppression temperature sequence, a sliding window weighting operation is performed on the continuous sampling points. The multiple sampling points within the window are weighted and summed according to the weight coefficient of the distance from the center point, and the summing result is used to replace the sampling value at the center of the window to generate a denoised temperature sequence.

4. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 3, characterized in that, The voltage mutation threshold is determined by acquiring the reference temperature sensor voltage signal of the heat sink when the gas nitrogen solenoid valve and liquid nitrogen injection valve are not turned on, obtaining a reference voltage sequence, calculating the standard deviation of the voltage values ​​at multiple sampling points in the reference voltage sequence, obtaining the reference voltage standard deviation, and multiplying the reference voltage standard deviation with a preset tolerance coefficient.

5. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the opening time of the nitrogen valve and the denoised temperature sequence, perform time index matching, calculate the difference between the timestamps of multiple sampling points in the sequence and the opening time of the nitrogen valve, select the turning point where the difference sign changes from negative to positive as the response judgment point and mark the timestamp to obtain the nitrogen response start time point. S202: Construct a time difference sequence based on the start time of the gas nitrogen response and the opening time of the gas nitrogen valve, perform a subtraction operation on the corresponding time values ​​of the two and record the difference result, unify the dimensions of the difference result, and obtain the gas nitrogen response lag time. S203: Based on the liquid nitrogen valve opening time and the denoised temperature sequence, perform a synchronization time retrieval, calculate the difference between the sequence timestamp and the liquid nitrogen valve opening time point by point, extract the time of the first sampling point where the difference turns from negative to positive and subtract it from the liquid nitrogen valve opening time to obtain the liquid nitrogen response lag time.

6. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Obtain the preset time interval parameter sequence, sort the boundary values ​​of multiple time periods in order, calculate the interval length of the difference between adjacent boundaries, map multiple intervals into a lag decision interval label set according to the time span, and then encode all interval labels to establish lag decision interval sequence data. S302: Based on the lag determination interval sequence data, call the gas nitrogen response lag time, compare it with the upper and lower limits of multiple intervals and locate the interval index to which it belongs, and at the same time perform the same interval matching on the liquid nitrogen response lag time, respectively map the matching index to the level identifier, and obtain the gas nitrogen and liquid nitrogen lag level group. S303: Based on the gas nitrogen and liquid nitrogen hysteresis level group, the difference between the gas nitrogen and liquid nitrogen level codes is determined, and the difference sign and amplitude are mapped to the identifier. The determination result is converted into the corresponding phase category code sequence to obtain the phase adjustment identifier.

7. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Extract preset control parameters to obtain control cycle duration, retrieve gas nitrogen valve opening time and liquid nitrogen valve opening time, and calculate the ratio and compensation conversion with control cycle duration through a dual-layer PID control model. Then, serialize the proportion of gas nitrogen valve and liquid nitrogen valve opening duration to generate pulse duty cycle data. S402: Based on the pulse duty cycle data, the proportion of gas nitrogen valve and liquid nitrogen valve are input into the decision tree model. According to the internal hierarchical judgment rules of the model, the proportion is compared with the preset condition boundary value layer by layer. Based on the comparison results, the judgment path is selected and the corresponding classification attribute code is extracted to generate an energy competition state identifier. S403: Based on the energy competition status identifier, call the phase adjustment identifier, verify the trigger sequence and phase adjustment point position corresponding to the status identifier, calculate the trigger timing offset, and allocate the action duration of multiple valves according to the offset and feedforward compensation to generate basic valve control commands.

8. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 7, characterized in that, The decision tree model consists of a root-level condition decision node, a multi-level branch addressing rule base, and a set of terminal state classification labels.

9. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The method further includes: S5: Retrieve the time of the extreme temperature point on the surface of the simulated heat sink in the thermal vacuum environment from the denoised temperature sequence, calculate the pulse frequency correction parameter based on the time of the extreme temperature point, and perform frequency and timing adjustment calculation on the basic valve control command through the pulse frequency correction parameter to generate the target switching level command. The target switching level command includes frequency correction parameters, timing adjustment results, and control signal levels.

10. The method for temperature control of gaseous nitrogen heating and liquid nitrogen cooling in a thermal vacuum environment simulating a heat sink according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the denoised temperature sequence, extract the time series of extreme temperature points on the surface of the heat sink in the simulated thermal vacuum environment. Perform adjacent difference operations on the temperature at multiple sampling times in the sequence. Mark the time index corresponding to the difference exceeding the preset temperature change threshold and reconstruct it in time order to generate a temperature extreme value time series index set. S502: Extract the time interval sequence of adjacent extreme points based on the temperature extreme value time series index set, perform a reciprocal transformation on the time interval sequence and perform a difference operation with the preset control cycle reference frequency, map the difference result to the corresponding interval in the frequency correction coefficient table and assign coefficient values ​​to generate pulse frequency correction parameters. S503: Based on the pulse frequency correction parameter, call the original pulse frequency and timing sequence in the basic valve control instruction, perform frequency correction on the product operation of the original pulse frequency, and resample and rearrange the timing sequence according to the period length corresponding to the correction frequency to generate the target switching level instruction.

Images