Ultralow-temperature refrigerating system and control method

Abstract

The ultralow-temperature refrigerating system and the control method have the beneficial effects that the temperature stability and the energy efficiency are remarkably improved, and the future target refrigerating capacity is accurately predicted by collecting and preprocessing multi-dimensional operation parameters in real time and constructing a dynamic thermal load prediction model based on multivariate nonlinear regression; the predicted value is compared with the real-time refrigerating capacity to generate a feedforward control instruction, the frequency of a high-low-pressure-stage compressor and the opening degree of an electronic expansion valve are adjusted in advance, and temperature fluctuation caused by thermal load sudden change is effectively restrained; meanwhile, a control strategy is fused with an energy efficiency optimization objective function, high-pressure loop and low-pressure loop cooperative adjustment is achieved, and unit refrigerating capacity power consumption is minimized; in addition, by periodically updating the model weight coefficient online, the adaptability of long-term operation of the system is guaranteed, and the manual intervention requirement is remarkably reduced.

Description

Technical Field

[0001] This invention relates to the field of refrigeration technology, and in particular to an ultra-low temperature refrigeration system and control method. Background Technology

[0002] Cryogenic refrigeration systems (e.g., below -80°C) are widely used in medical, scientific research, and industrial fields (e.g., for superconducting equipment or biological sample preservation). Their core requirement is maintaining the long-term stability of the temperature of the object being cooled. Existing systems typically employ a cascaded refrigeration structure, containing a high-pressure stage and a low-pressure stage with dual loops, thermally coupled through an interstage heat exchanger. During system operation, key parameters such as evaporator / condenser temperature, compressor suction and discharge pressure, and electronic expansion valve opening (e.g., low-pressure stage evaporator outlet temperature, high-pressure stage condenser outlet temperature, and high and low-pressure side outlet temperatures of the interstage heat exchanger) need to be monitored in real time, and the refrigeration capacity is adjusted via a variable-frequency compressor and electronic expansion valve. However, because the heat load of the object being cooled often fluctuates drastically due to sudden changes in the external environment or internal conditions (e.g., equipment start-up and shutdown, sample placement), traditional control methods mainly rely on feedback adjustment mechanisms based on current operating parameters (e.g., PID control), which exhibit significant lag.

[0003] Problems with the existing structure include:

[0004] First, the lag in feedback control leads to temperature fluctuations. When the heat load changes abruptly, the system must wait for the temperature of the object being cooled to shift before initiating adjustments (such as adjusting the compressor frequency or expansion valve opening). At this time, the cooling capacity response delay may be as long as several minutes, causing the temperature of the object being cooled to overshoot or oscillate (typically fluctuating within ±2℃), affecting the stability of the equipment (such as the inactivation of biological samples).

[0005] Secondly, parameter coupling leads to a decrease in energy efficiency. The high-pressure stage and low-pressure stage circuits are strongly coupled through interstage heat exchangers. Adjusting a single parameter (such as changing only the low-pressure stage compressor frequency) may disrupt the interstage matching, resulting in an abnormal increase in the high-pressure stage discharge pressure or an excessively low low-pressure stage suction pressure (Note: This phenomenon can easily trigger compressor protection shutdown or a deterioration in the energy efficiency ratio). In addition, existing methods rely on manual experience to set control parameters, making it difficult to adapt to complex changes in operating conditions (such as a sudden rise in ambient temperature or a step increase in load), requiring frequent manual intervention to maintain stable operation.

[0006] Therefore, the following problems urgently need to be solved in this field: how to predict the trend of heat load changes in advance and actively adjust the cooling capacity before temperature fluctuations occur; how to achieve coordinated optimization of high and low temperature circuit parameters to improve system energy efficiency while ensuring temperature accuracy; and how to build an adaptive control model to reduce reliance on human experience. Summary of the Invention

[0007] The purpose of this invention is to provide an ultra-low temperature refrigeration system and control method to solve the problems existing in the prior art.

[0008] To achieve the above objectives, the present invention provides the following solution:

[0009] This invention provides a control method for an ultra-low temperature refrigeration system, comprising the following steps:

[0010] S1. Data Acquisition and Preprocessing: Multiple sensors deployed within the cryogenic refrigeration system are used to collect raw operating status parameters in real-time and periodically. These parameters include at least: evaporator outlet temperature of the low-pressure stage refrigeration circuit, suction pressure of the low-pressure stage compressor, discharge pressure of the low-pressure stage compressor, condenser outlet temperature of the high-pressure stage refrigeration circuit, suction pressure of the high-pressure stage compressor, discharge pressure of the high-pressure stage compressor, high-pressure side outlet temperature of the interstage heat exchanger, low-pressure side outlet temperature of the interstage heat exchanger, current operating frequency of the low-pressure stage compressor, current operating frequency of the high-pressure stage compressor, current opening degree of the low-pressure stage electronic expansion valve, current opening degree of the high-pressure stage electronic expansion valve, current temperature of the object being cooled, and ambient temperature. Data preprocessing is performed on the collected raw operating status parameters. This preprocessing includes data cleaning to remove outliers, data filtering to smooth noise, and data normalization to eliminate the dimensional differences between parameters, thereby generating a standardized historical status dataset.

[0011] S2. Prediction Model Construction: Based on the generated standardized historical state dataset, a dynamic heat load prediction model is constructed to predict the target cooling capacity required by the system at future times. The model construction process includes: calculating the actual heat load at historical times based on the evaporator outlet temperature, the current operating frequency of the low-pressure stage compressor, the current operating frequency of the high-pressure stage compressor, and the temperature change rate of the cooled object from the historical state dataset; using the actual heat load at historical times as the dependent variable, and selecting a set of preset key influencing factors as independent variables from the standardized historical state dataset. The key influencing factors include at least the normalized low-pressure stage evaporator outlet temperature, the low-pressure stage compressor suction pressure, the high-pressure stage condenser outlet temperature, the temperature change rate of the cooled object, and the ambient temperature; using a multivariate nonlinear regression analysis algorithm, the mapping relationship between the dependent and independent variables is fitted to determine the structure of the dynamic heat load prediction model and a set of model weight coefficients.

[0012] S3. Target Cooling Capacity Prediction: During the operation of the cryogenic refrigeration system, a set of real-time operating parameters, collected and preprocessed at the current moment, are input into the dynamic heat load prediction model constructed in step S2. Through model calculation, the target cooling capacity within the future prediction time window is obtained. The formula for calculating the target cooling capacity is:

[0013] Qpred (t+Δt)=F(W,X(t))=Σ(w i *x i (t))+w c ;

[0014] Among them, Q pred (t+Δt) represents the predicted target cooling capacity, F is the nonlinear mapping function of the dynamic heat load prediction model, and W is the set of model weight coefficients, which is determined in step S2; X(t) is the vector of key influencing factors at the current time t as model input, and w i For the i-th key influence factor x i The weighting coefficients of (t), w c This is the constant bias term of the model;

[0015] S4. Control command generation and execution: The target cooling capacity predicted in step S3 is compared with the current real-time cooling capacity calculated based on the current operating state of the system to obtain the cooling capacity deviation. Based on the cooling capacity deviation, the target operating frequency for the high-pressure stage compressor and the low-pressure stage compressor, as well as the target opening degree for the high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve are calculated through the preset control strategy adjustment module. The corresponding control commands are generated and sent to the corresponding frequency converters and valve actuators to perform the adjustment of the target operating frequency and the target opening degree. In this way, the cooling capacity of the system is adjusted in a feedforward manner before the actual heat load changes significantly, so as to maintain the stability of the temperature of the object being cooled.

[0016] Preferably, in step S1, the data preprocessing operation further includes: using the Kalman filter algorithm to process the temperature and pressure sensor data containing significant random noise, in order to improve the accuracy and stability of the data and provide high-quality data input for subsequent model construction.

[0017] Preferably, in step S2, the multivariate nonlinear regression analysis algorithm is a gradient boosting decision tree algorithm or a support vector regression algorithm, which is used to effectively capture the complex nonlinear relationship between various operating parameters and heat load.

[0018] Preferably, in step S4, when the control strategy adjustment module calculates the target operating frequency and the target opening degree, it also comprehensively considers the objective function of energy efficiency optimization, so as to maximize the overall unit cooling capacity power consumption of the system while meeting the target cooling capacity, thereby achieving energy-saving operation.

[0019] Preferably, it further includes:

[0020] S5. Online model update: During continuous system operation, the latest actual heat load and its corresponding real-time operating parameters are periodically added to the historical state dataset, and step S2 is re-executed to iteratively optimize and update the model weight coefficients, thereby ensuring the long-term adaptability of the model to changes in system state.

[0021] The present invention also provides an ultra-low temperature refrigeration system, comprising:

[0022] The high-pressure stage refrigeration circuit includes, in sequence, a high-pressure stage compressor, a high-pressure stage condenser, a high-pressure stage electronic expansion valve, and a high-pressure stage evaporator coil;

[0023] The low-pressure stage refrigeration circuit includes, in sequence, a low-pressure stage compressor, a low-pressure stage condenser coil, a low-pressure stage electronic expansion valve, and a low-pressure stage evaporator for refrigerating the object being cooled.

[0024] An interstage heat exchanger, wherein the high-pressure stage evaporator coil and the low-pressure stage condenser coil are both disposed inside the interstage heat exchanger, for the purpose of realizing heat transfer between the high-pressure stage refrigeration circuit and the low-pressure stage refrigeration circuit;

[0025] A status sensor is deployed at key node locations in the high-pressure stage refrigeration circuit and the low-pressure stage refrigeration circuit to collect raw operating status parameters in real time. The raw operating status parameters include at least the low-pressure stage evaporator outlet temperature, the low-pressure stage compressor suction pressure, the low-pressure stage compressor discharge pressure, the high-pressure stage condenser outlet temperature, the high-pressure stage compressor suction pressure, the high-pressure stage compressor discharge pressure, the current temperature of the object being cooled, and the ambient temperature.

[0026] The central control unit is electrically connected to the status sensor, the high-pressure stage compressor, the low-pressure stage compressor, the high-pressure stage electronic expansion valve, and the low-pressure stage electronic expansion valve. The central control unit is used to execute the control method of the cryogenic refrigeration system.

[0027] Preferably, the central control unit includes:

[0028] The data processing module is used to receive and process the raw operating state parameters collected by the state sensor, and perform data cleaning, filtering and normalization operations to generate a standardized historical state dataset.

[0029] The model building module is used to construct a dynamic heat load prediction model for predicting future target cooling capacity based on the standardized historical state dataset and using a preset regression analysis algorithm, and to determine its model weight coefficients.

[0030] The prediction calculation module is used to input the real-time collected operating parameters into the dynamic heat load prediction model, and perform calculations according to the preset mathematical formula to output the target cooling capacity prediction value for future time.

[0031] The control decision module is used to calculate control commands for adjusting the operating frequency of the high-pressure stage compressor and the low-pressure stage compressor, as well as the opening degree of the high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve, based on the deviation between the target cooling capacity prediction value and the current real-time cooling capacity.

[0032] Preferably, both the high-pressure stage compressor and the low-pressure stage compressor are variable frequency scroll compressors, and their operating frequencies are continuously adjusted by receiving analog or digital communication signals from the central control unit.

[0033] Preferably, both the high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve are stepper motor driven electronic expansion valves, and their valve core opening is precisely adjusted in steps by receiving pulse signals from the central control unit for fine control of refrigerant flow.

[0034] Preferably, the central control unit further includes a memory and a processor. The memory stores the standardized historical state dataset, the structure and weight coefficients of the dynamic heat load prediction model, and the control method compiled into executable instructions. The processor is configured to read and execute the executable instructions in the memory to drive the data processing module, model building module, prediction calculation module, and control decision module to work together.

[0035] The present invention achieves the following beneficial technical effects compared to the prior art:

[0036] This invention provides an ultra-low temperature refrigeration system and control method, which significantly improves temperature stability and energy efficiency. By collecting and preprocessing multi-dimensional operating parameters in real time, a dynamic heat load prediction model based on multivariate nonlinear regression is constructed to accurately predict the future target cooling capacity. Then, the predicted value is compared with the real-time cooling capacity to generate feedforward control commands, which adjust the frequency of the high and low pressure stage compressors and the opening of the electronic expansion valve in advance, effectively suppressing temperature fluctuations caused by sudden changes in heat load. At the same time, the control strategy integrates an energy efficiency optimization objective function to achieve coordinated regulation of the high and low pressure loops, minimizing power consumption per unit cooling capacity. In addition, by periodically updating the model weight coefficients online, the system's long-term adaptability is ensured, significantly reducing the need for manual intervention. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the 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.

[0038] Figure 1 The flowchart shows the control method of the cryogenic refrigeration system provided by the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] The purpose of this invention is to provide an ultra-low temperature refrigeration system and control method to solve the problems existing in the prior art.

[0041] To enable those skilled in the art to fully understand and reproduce the technical solution of this invention, the following description is provided in conjunction with the appendix. Figure 1 The present invention provides a more detailed description of the cryogenic refrigeration system and its control method, along with several preferred embodiments. All values ​​for temperature, pressure, flow rate, frequency, valve opening, etc., mentioned herein are for illustrative purposes only and do not constitute a limitation on the scope of protection of the present invention.

[0042] Example 1:

[0043] The cryogenic refrigeration system in this embodiment consists of a high-pressure stage refrigeration circuit, a low-pressure stage refrigeration circuit, an interstage heat exchanger, a status sensor array, and a central control unit.

[0044] The high-pressure stage refrigeration circuit uses R404A refrigerant. Along the refrigerant flow direction, a high-pressure stage variable frequency scroll compressor, a high-pressure stage microchannel condenser, a high-pressure stage electronic expansion valve, and a high-pressure stage evaporator coil located within the interstage heat exchanger are connected in series. A subcooling section is configured at the condenser outlet to ensure the high-pressure side refrigerant is in a stable subcooled state before entering the electronic expansion valve. The rated displacement of the high-pressure stage compressor is 1.3 m³ / s. 3 The speed is adjusted continuously by the 0-10V analog signal output by the central control unit, with a frequency range of 30-90Hz.

[0045] The low-pressure stage refrigeration circuit uses R23 refrigerant. Along the refrigerant flow direction, a low-pressure stage variable frequency scroll compressor, a low-pressure stage condenser coil located within the interstage heat exchanger, a low-pressure stage electronic expansion valve, and a plate-fin evaporator directly attached to the object being cooled (in this embodiment, the wall of the -86℃ biobank cavity) are connected in series. The low-pressure stage compressor has a rated discharge capacity of 0.8 m³ / s. 3 / h, with a frequency conversion range of 30-90Hz. The two-stage loops perform reverse coupling heat exchange in the interstage heat exchanger. The refrigerant in the high-pressure stage evaporator coil absorbs heat during evaporation, providing condensation cooling capacity for the refrigerant in the low-pressure stage condenser coil, thus achieving cascaded refrigeration.

[0046] The state sensor array includes:

[0047] Temperature sensors: Pt100 platinum resistance thermometers, respectively arranged at the outlet of the low-pressure stage evaporator, the outlet of the high-pressure stage condenser, the outlets of the high and low pressure sides of the interstage heat exchanger, the center of the cavity of the object being cooled, and the ambient temperature measuring point, with a measurement uncertainty of ±0.1℃;

[0048] Pressure sensor: silicon resonant type, range 0-4MPa, respectively arranged at the suction and discharge ends of the low-pressure stage compressor and the suction and discharge ends of the high-pressure stage compressor, with a measurement uncertainty of ±0.25%FS;

[0049] Frequency detection: Directly output from the Hall sensor inside the compressor inverter;

[0050] Valve opening detection: The continuous opening value of 0-500 steps is obtained through the step number feedback of the electronic expansion valve stepper motor driver.

[0051] The central control unit uses an industrial-grade ARM Cortex-A72 dual-core processor, equipped with a Linux real-time kernel, and has 512MB of DDR4 RAM and 8GB of eMMC memory. It communicates with various sensors and drivers via a Modbus-RTU bus to complete data acquisition, model calculation, and control command output.

[0052] The specific implementation steps of the control method are as follows: Figure 1 As shown, after the system is powered on for the first time, the central control unit automatically executes the initialization script to complete hardware self-test and sensor calibration, and then proceeds to the following steps:

[0053] S1. Data Acquisition and Preprocessing

[0054] The central control unit cyclically collects all the above-mentioned raw operating status parameters at a frequency of 2Hz. The collected raw data first undergoes outlier removal: if a temperature or pressure value deviates from the midpoint of its sliding window by more than 3σ within 5 consecutive sampling cycles, it is marked as an outlier and replaced with the midpoint of the window; subsequently, a Kalman filter is used to denoise the temperature and pressure signals. The process noise covariance matrix Q and the observation noise covariance matrix R of the filter are set according to the previous calibration experiments, with Q = 1 × 10 for the temperature channel. -4 ℃ 2 R = 1 × 10 -3 ℃ 2 Pressure channel Q = 2 × 10 -3 MPa 2 R = 4 × 10 -3 MPa 2 After filtering, all parameters are normalized according to their respective historical extreme values, mapped to the [0,1] interval, forming a standardized historical state dataset, which is then appended to the SQLite database in real time.

[0055] S2. Construction of Dynamic Heat Load Prediction Model

[0056] After the system has accumulated 24 hours of operation, the central control unit extracts the most recent 6 hours of continuous data from the database as training samples. First, the actual heat load is calculated using the evaporator outlet temperature change rate, compressor frequency, and the temperature change rate of the object being cooled.

[0057] Subsequently, the low-pressure stage evaporator outlet temperature, low-pressure stage compressor suction pressure, high-pressure stage condenser outlet temperature, the temperature change rate of the cooled object, and ambient temperature were selected as key influencing factors to form an independent variable vector. With the actual heat load as the dependent variable, a gradient boosting decision tree algorithm was used for nonlinear regression. The model hyperparameters were determined through grid search: tree depth 6, learning rate 0.05, subsampling rate 0.8, and number of iterations 200. After training, the model weight coefficients W and bias b were stored in the control unit's JSON model file, with the file header containing a version number and checksum to prevent abnormal writes.

[0058] S3, Target Cooling Capacity Prediction

[0059] During normal system operation, the central control unit collects the latest real-time operating parameters every 10 seconds, preprocesses them using the same procedure as in step 1, and then inputs them into the gradient boosting decision tree model, outputting the target cooling capacity for the next 60 seconds. The prediction results are smoothed using cubic spline interpolation to prevent abrupt changes.

[0060] The formula for calculating the target cooling capacity is:

[0061] Q pred (t+Δt)=F(W,X(t))=Σ(wi *x i (t))+w c ;

[0062] Among them, Q pred (t+Δt) represents the predicted target cooling capacity, F is the nonlinear mapping function of the dynamic heat load prediction model, and W is the set of model weight coefficients, which is determined in step S2; X(t) is the vector of key influencing factors at the current time t as model input, and w i For the i-th key influence factor x i The weighting coefficients of (t), w c This is the constant bias term of the model.

[0063] S4. Control Command Generation and Execution

[0064] The central control unit calculates the current real-time cooling capacity (using the same calculation method as in step 2) and compares it with the target cooling capacity to obtain the cooling capacity deviation ΔQ. If |ΔQ|>50W, the control strategy adjustment module is triggered. This module solves for the optimal compressor frequency and valve opening based on the energy balance and energy efficiency optimization objective function. The optimization algorithm adopts improved particle swarm optimization (PSO), with a population size of 20 and 30 iterations. The PSO outputs the target frequency of the high-pressure stage compressor, the target frequency of the low-pressure stage compressor, the target number of steps for the high-pressure stage electronic expansion valve, and the target number of steps for the low-pressure stage electronic expansion valve. The central control unit sends 0-10V analog signals and pulse sequences to the compressor inverter and stepper motor driver respectively to achieve stepless speed regulation and step-by-step valve position adjustment. The entire control cycle is less than 2s, which is much smaller than the lag time of 5-8 minutes of traditional feedback control, effectively suppressing temperature fluctuations.

[0065] S5, Online Model Update

[0066] Every 168 hours of system operation, the central control unit automatically appends the latest week's operational data to the historical state dataset and re-executes step 2, model training and validation. If the new model's prediction error (MAE) is reduced by more than 5% compared to the old model, the new model replaces the old model; otherwise, the old model is retained, and only its internal statistical features are updated to achieve long-term model adaptability.

[0067] In a 25°C environmental chamber, the system of this embodiment and a control system using traditional PID feedback control were run in parallel for 72 hours, with the set temperature of the object being cooled at -86°C. Typical disturbances were simulated during the experiment: 20 kg of room-temperature samples were added to the sample library every 12 hours, and the door was opened and closed for 30 seconds every 3 hours. The results showed that the temperature fluctuation range of the system of this invention was ±0.3°C, with a maximum instantaneous deviation of 0.5°C; the temperature fluctuation range of the control system was ±2.1°C, with a maximum instantaneous deviation of 3.4°C. Furthermore, the average power consumption per unit cooling capacity of the system of this invention was reduced by 12.7%.

[0068] As can be seen from the above specific implementation methods, the present invention significantly improves the temperature stability and energy efficiency of the cryogenic refrigeration system by integrating high-fidelity data acquisition, nonlinear heat load prediction, feedforward collaborative control and online model updating, and the operation process does not require manual intervention, thus achieving reliable support for the needs of medical, scientific research and industrial cryogenic applications.

[0069] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0070] It should be noted that the components mentioned in the above embodiments are all general standard parts or components known to those skilled in the art. Their structures and principles can be learned by those skilled in the art through technical manuals or conventional experimental methods.

[0071] This invention has illustrated its principles and implementation methods using specific examples. The descriptions of these embodiments are merely illustrative of the method and its core ideas; furthermore, those skilled in the art will recognize that modifications may be made to the specific implementation methods and application scope based on the principles of this invention. Therefore, the content of this specification should not be construed as limiting the invention.

Claims

1. A control method of an ultralow temperature refrigeration system, characterized by, The method comprises the following steps: S1, data acquisition and preprocessing, through a plurality of sensors deployed in the ultra-low temperature refrigeration system, real-time and periodic acquisition of original operating state parameters, the original operating state parameters at least including: low-pressure stage refrigeration circuit evaporator outlet temperature, low-pressure stage compressor suction pressure, low-pressure stage compressor discharge pressure, high-pressure stage refrigeration circuit condenser outlet temperature, high-pressure stage compressor suction pressure, high-pressure stage compressor discharge pressure, inter-stage heat exchanger high-pressure side outlet temperature, inter-stage heat exchanger low-pressure side outlet temperature, low-pressure stage compressor current operating frequency, high-pressure stage compressor current operating frequency, low-pressure stage electronic expansion valve current opening degree, high-pressure stage electronic expansion valve current opening degree, current temperature of the cooled object and ambient temperature; performing data preprocessing operation on the collected original operating state parameters, the data preprocessing operation including data cleaning to eliminate outliers, data filtering to smooth noise and data normalization to eliminate the influence of different parameters, thereby generating a set of standardized historical state data; S2, prediction model construction, based on the generated standardized historical state data set, a dynamic heat load prediction model for predicting the target refrigeration capacity required by the system at the future time is constructed; the model construction process includes: calculating the actual heat load at the historical time according to the evaporator outlet temperature, low-pressure stage compressor current operating frequency, high-pressure stage compressor current operating frequency and temperature change rate of the cooled object in the historical state data set; taking the actual heat load at the historical time as the dependent variable, and selecting a set of pre-set key influence factors from the standardized historical state data set as the independent variables, the key influence factors at least including the normalized low-pressure stage evaporator outlet temperature, low-pressure stage compressor suction pressure, high-pressure stage condenser outlet temperature, temperature change rate of the cooled object and ambient temperature; adopting multivariate nonlinear regression analysis algorithm, fitting the mapping relationship between the dependent variable and the independent variable, thereby determining the structure of the dynamic heat load prediction model and a set of model weight coefficients; S3, target refrigeration capacity prediction, in the process of operation of the ultra-low temperature refrigeration system, a set of real-time operating parameters collected and preprocessed at the current time are input into the dynamic heat load prediction model constructed in step S2, and the target refrigeration capacity in the future prediction time window is obtained through model operation; the calculation formula of the target refrigeration capacity is: Q pred (t+Δt) = F(W,X(t)) =∑(w i *x i (t))+w c ; wherein Q pred (t + At) is the predicted target cooling capacity, F is a non-linear mapping function of the dynamic thermal load prediction model, W is a set of model weight coefficients, which is determined in step S2; X(t) is a vector of key influencing factors at the current time t as model input, w i is the weight coefficient corresponding to the i-th key influencing factor x i (t); w c is a constant bias term of the model. S4, control instruction generation and execution, comparing the target refrigeration capacity predicted in step S3 with the current real-time refrigeration capacity calculated based on the current operating state of the system to obtain the refrigeration capacity deviation; according to the refrigeration capacity deviation, through the pre-set control strategy adjustment module, the target operating frequency for the high-pressure stage compressor and the low-pressure stage compressor, and the target opening degree for the high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve are calculated; corresponding control instructions are generated and sent to the corresponding frequency converter and valve driver to execute the adjustment of the target operating frequency and the target opening degree, thereby feeding forwardly adjusting the refrigeration capacity of the system before the actual heat load changes significantly, so as to maintain the stability of the temperature of the cooled object.

2. The control method of an ultra-low temperature refrigeration system according to claim 1, characterized by, In step S1, the data preprocessing operation further includes: processing temperature and pressure sensor data containing significant random noise using a Kalman filtering algorithm to improve the accuracy and stability of the data and provide high-quality data input for subsequent model construction.

3. The control method of an ultra-low temperature refrigeration system according to claim 1, characterized by, In step S2, the multivariate nonlinear regression analysis algorithm is a gradient boosting decision tree algorithm or a support vector regression algorithm, which is used to effectively capture the complex nonlinear relationship between the operating parameters and the thermal load.

4. The control method of an ultra-low temperature refrigeration system according to claim 1, characterized by, In step S4, the control strategy adjustment module also considers the objective function of energy efficiency optimization when calculating the target operating frequency and target opening, so that the overall unit refrigeration power consumption is maximized under the premise of meeting the target refrigeration capacity, thereby achieving energy-saving operation.

5. The control method of an ultra-low temperature refrigeration system according to claim 1, characterized by, Also includes: S5, online model updating, periodically supplementing the latest actual thermal load and its corresponding real-time operating parameters to the historical state data set during the continuous operation of the system, and re-executing step S2 to iteratively optimize and update the model weight coefficients, to ensure the long-term adaptability of the model to system state changes.

6. An ultra-low temperature refrigeration system, characterized by, Includes: a high-pressure stage refrigeration circuit, which sequentially includes a high-pressure stage compressor, a high-pressure stage condenser, a high-pressure stage electronic expansion valve, and a high-pressure stage evaporating coil; a low-pressure stage refrigeration circuit, which sequentially includes a low-pressure stage compressor, a low-pressure stage condensing coil, a low-pressure stage electronic expansion valve, and a low-pressure stage evaporator for refrigerating the cooled object; an inter-stage heat exchanger, the high-pressure stage evaporating coil and the low-pressure stage condensing coil are arranged inside the inter-stage heat exchanger, for realizing heat transfer between the high-pressure stage refrigeration circuit and the low-pressure stage refrigeration circuit; a state sensor, which is deployed at key node positions of the high-pressure stage refrigeration circuit and the low-pressure stage refrigeration circuit, for real-time acquisition of original operating state parameters, the original operating state parameters at least including low-pressure stage evaporator outlet temperature, low-pressure stage compressor suction pressure, low-pressure stage compressor discharge pressure, high-pressure stage condenser outlet temperature, high-pressure stage compressor suction pressure, high-pressure stage compressor discharge pressure, current temperature of the cooled object, and ambient temperature; a central control unit, which is electrically connected with the state sensor, the high-pressure stage compressor, the low-pressure stage compressor, the high-pressure stage electronic expansion valve, and the low-pressure stage electronic expansion valve, and is used to execute the control method of the ultra-low temperature refrigeration system.

7. The ultra-low temperature refrigeration system of claim 6, wherein, The central control unit includes: a data processing module, for receiving and processing the original operating state parameters collected by the state sensor, performing data cleaning, filtering, and normalization operations to generate a standardized historical state data set; a model construction module, for constructing a dynamic thermal load prediction model for predicting future target refrigeration capacity based on the standardized historical state data set using a preset regression analysis algorithm, and determining its model weight coefficients; The prediction operation module is configured to input the real-time collected operation parameter into the dynamic thermal load prediction model, and perform operation according to a preset mathematical formula to output a target refrigerating capacity prediction value at a future time; The control decision module is configured to calculate a control instruction for adjusting the operation frequency of the high-pressure stage compressor and the low-pressure stage compressor and the opening degree of the high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve according to the deviation between the target refrigerating capacity prediction value and the current real-time refrigerating capacity.

8. The ultra-low temperature refrigeration system of claim 6, wherein, The high-pressure stage compressor and the low-pressure stage compressor are both variable frequency scroll compressors, and the operation frequency thereof is continuously adjusted by receiving an analog quantity or digital communication signal from the central control unit.

9. The ultra-low temperature refrigeration system of claim 6, wherein, The high-pressure stage electronic expansion valve and the low-pressure stage electronic expansion valve are both electronic expansion valves driven by stepping motors, and the opening degree of the valve core thereof is precisely step-adjusted by receiving a pulse signal from the central control unit, for fine control of the refrigerant flow.

10. The ultra-low temperature refrigeration system of claim 7, wherein, The central control unit further comprises a memory and a processor, the memory stores the standardized historical state data set, the structure and weight coefficient of the dynamic thermal load prediction model, and the control method compiled as executable instructions; The processor is configured to read and execute the executable instructions in the memory to drive the data processing module, the model construction module, the prediction operation module and the control decision module to work cooperatively.

Images