Air-cooled simulation environment laboratory precision temperature and humidity control refrigeration cycle system

Abstract

The application discloses a precision temperature and humidity adjusting refrigeration cycle system of a forced air simulation environment laboratory, and relates to the technical field of laboratory refrigeration equipment.The system comprises seven modules, i.e., a refrigeration and dehumidification integrated circuit, a three-stage cold energy heat recovery collaborative module, a dynamic coupling self-learning decoupling unit and the like, and a closed-loop collaborative network is constructed through an industrial Ethernet; the refrigeration and dehumidification energy is complementary, the temperature and humidity are decoupled to eliminate interference, and the flow field is self-adaptive to optimize the temperature field; the cold energy is adjustable in a full range and the waste heat is recovered in stages, and the energy consumption and precision are dynamically balanced; the working condition is pre-judged to shorten the switching time, and the fault is classified for self-recovery to guarantee operation.The application solves the pain points of the traditional system, such as temperature and humidity adjusting conflict, high energy consumption, slow response and insufficient reliability, realizes the cooperation of precise control, deep energy saving, fast response and stable operation, provides a reliable simulation environment for performance testing of forced air equipment, and guarantees the accuracy of test data.

Description

Technical Field

[0001] This invention relates to the field of laboratory refrigeration equipment technology, and in particular to a refrigeration cycle system for precise temperature and humidity control in an air-cooled simulated environment laboratory. Background Technology

[0002] As a core platform for performance testing of air-cooled equipment in industries such as scientific research, electronics, and chemical engineering, the air-cooled simulation environment laboratory's refrigeration cycle system's temperature and humidity control accuracy, load adaptability, energy-saving level, and operational stability directly determine the reliability of test data and the efficiency of the testing process. With the continuous improvement of industry requirements for product performance testing, the existing air-cooled simulation environment laboratory's refrigeration cycle system has gradually exposed many technical shortcomings, making it difficult to meet the testing needs of high precision, wide load, and low energy consumption.

[0003] In terms of temperature and humidity control, traditional refrigeration cycle systems often use independent loop designs for refrigeration and dehumidification, which lack an energy coordination mechanism. This not only wastes energy but also easily leads to regulation conflicts, causing significant temperature fluctuations during dehumidification. At the same time, there is significant coupling interference in temperature and humidity regulation, which traditional control methods cannot dynamically capture and eliminate, thus limiting the accuracy of temperature and humidity control. In addition, the air supply structure is mostly a fixed design, which cannot adaptively adjust according to changes in the flow field in the test room, resulting in uneven local temperature field distribution. This, in turn, affects the consistency of environmental parameters around the test product and reduces the reliability and repeatability of test data.

[0004] In terms of load adaptation and energy saving, the existing refrigeration systems mostly use segmented regulation or simple frequency conversion regulation for cooling capacity adjustment. The cooling capacity coverage is limited, making it difficult to accurately match the dynamically changing load demand of the laboratory. The cooling capacity is insufficient at high loads and redundant at low loads. The application of condensation heat recovery technology is limited, and it can only achieve partial waste heat recovery. Moreover, it is not designed in conjunction with cooling capacity adjustment, resulting in low waste heat utilization and high system energy consumption. At the same time, traditional systems cannot balance regulation accuracy and operating energy consumption. They often sacrifice energy consumption in pursuit of high accuracy or give up some control accuracy in order to reduce energy consumption, making it difficult to achieve a dynamic balance between the two.

[0005] In terms of operational response and reliability, traditional systems lack the ability to predict load changes. When switching operating conditions, they can only passively respond to load changes, resulting in significant parameter oscillations and excessively long stabilization times, which affects testing efficiency. In addition, system fault diagnosis often relies on simple threshold judgments, which have low identification accuracy. They cannot automatically handle minor faults and require manual intervention for maintenance. For serious faults, they lack a rapid emergency response mechanism, which can easily lead to test interruptions or even equipment damage. This not only increases maintenance costs but also affects the continuity of testing work.

[0006] Therefore, this application proposes a refrigeration cycle system for precise temperature and humidity control in an air-cooled simulated environment laboratory. Summary of the Invention

[0007] One objective of this invention is to propose a precise temperature and humidity control refrigeration cycle system for air-cooled simulated environment laboratories. This invention addresses the technical pain points of traditional air-cooled simulated environment laboratory refrigeration cycle systems, such as conflicting temperature and humidity control, uneven local environments, limited cooling capacity adaptability, difficulty in balancing energy consumption and accuracy, slow operational response, and insufficient reliability. It achieves precise and coordinated control of temperature and humidity, effective elimination of coupling interference, and optimization of temperature field uniformity, achieving a balance between wide-load dynamic adaptation and deep energy saving. At the same time, it improves the response speed of operating condition switching, the continuity of system operation, and the convenience of maintenance, providing a stable, efficient, and reliable simulation environment for the performance testing of air-cooled equipment, and ensuring the accuracy and repeatability of test data.

[0008] According to an embodiment of the present invention, a precise temperature and humidity control refrigeration cycle system for an air-cooled simulated environment laboratory includes an integrated refrigeration and dehumidification loop, a three-stage cold and heat recovery collaborative module, a dynamic coupling self-learning decoupling unit, a flow field cold energy linkage optimization structure and a working condition migration prediction system. Each module constructs a millisecond-level closed-loop collaborative network through an industrial Ethernet network.

[0009] The integrated refrigeration and dehumidification circuit adopts a shared finned heat exchange channel design to achieve complementary energy for refrigeration and dehumidification. The three-stage cold and heat recovery collaborative module combines stepless adjustment, constant volume replenishment, and condensation heat cascade recovery to cover a wide cold capacity range of 10%-100%. The dynamic coupling self-learning decoupling unit updates the coupling coefficient based on real-time data iteration to eliminate temperature and humidity regulation interference. The flow field cold capacity linkage optimization structure optimizes the temperature field uniformity through real-time CFD simulation and adaptive adjustment of the physical structure. The operating condition migration prediction system integrates BP neural network and transfer learning to output load demand in advance.

[0010] It also includes an energy consumption accuracy dynamic balancing module and a fault self-diagnosis and self-healing module. The energy consumption accuracy dynamic balancing module dynamically allocates the ratio of cooling output and heat recovery through configurable priority coefficients and optimization algorithms, and adaptively switches between accuracy-first and energy consumption-first modes. The fault self-diagnosis and self-healing module realizes fault classification diagnosis and self-healing response by constructing a fault feature library and using deep learning recognition algorithms.

[0011] Furthermore, the integrated refrigeration and dehumidification circuit includes an integrated composite heat exchanger, in which a refrigeration channel and a dehumidification channel are arranged in parallel. The two channels share a sawtooth and corrugated composite fin structure. The working mode is switched by an electromagnetic three-way valve. In the refrigeration mode, the dehumidification channel is closed by a sealing baffle. In the dehumidification mode, the refrigeration channel outputs 30%-50% of the rated cooling capacity as the dehumidification cooling capacity support.

[0012] Furthermore, in the three-stage cooling and heat recovery collaborative module, the stepless adjustment unit uses a combination of a variable frequency compressor and an electronic expansion valve, achieving continuous adjustment of cooling capacity from 10% to 100% through a PID algorithm. The constant-capacity replenishment unit connects 2-4 constant-capacity compressors in parallel, automatically starting and stopping according to the cooling capacity demand threshold. The condensing heat cascade recovery unit includes a primary plate heat exchanger, secondary heat pipes, and terminal waste heat recovery coils, used for heating the hot water tank, preheating the supply air, and preheating the humidifier, respectively, through a formula... Dynamically allocate and recover heat;

[0013] in, , , These are the weighting coefficients for the three levels of recycling. , , ,and , , , They are three-stage heat recovery systems.

[0014] Furthermore, the coupling coefficient of the dynamically coupled self-learning decoupling unit can be updated in real time, and its update formula is as follows:

[0015]

[0016] in, The initial coupling coefficients are... For runtime, For integration variables, , The weights are dynamic, and the self-learning iteration cycle is 5-10 minutes. The least squares method is used to optimize the algorithm by selecting the last 30 sets of valid historical data. , When the temperature and humidity coupling deviation exceeds ±0.5℃ or ±3%, the gain of the corresponding regulating branch will be automatically reduced by 20%-30% to eliminate regulation interference.

[0017] Furthermore, the flow field cooling capacity linkage optimization structure includes a deformable air supply duct, a double-layer perforated plate assembly, an adjustable guide plate, and a mixer. The deformable air supply duct adjusts its cross-sectional shape via a piezoelectric ceramic actuator. The upper layer of the double-layer perforated plate has an adjustable opening ratio of 30%-50%, while the lower layer has a fixed opening ratio of 40%. The guide plate has an adjustable angle of 0°-45°. The mixer is positioned before the air outlet of the air handling unit and contains spiral guide vanes. The CFD real-time simulation update frequency is 1-2 times / min. The intelligent control unit, based on sensor array data, uses a formula... Dynamically match fan speed with cooling capacity;

[0018] in For flow coefficient, This is the real-time cross-sectional area of ​​the pipeline. For the supply air pressure difference, This refers to air density.

[0019] Furthermore, in the aforementioned operating condition migration prediction system, the criteria for determining similar operating conditions are a temperature and humidity setpoint deviation of ≤ ±2℃ and a cooling demand deviation of ≤ ±10%. The cooling capacity variation patterns of similar operating conditions are extracted using a transfer learning algorithm, and the prediction formula is optimized as follows:

[0020]

[0021] in, Weights are assigned to historical data. , As the current data weight, , , To migrate data weights, ,and , To migrate the operating condition cooling data, This is data related to environmental interference.

[0022] Furthermore, the priority coefficient P of the energy consumption accuracy dynamic balance module can be manually set by the user or automatically adapted by the system according to the test task. When P ≥ 0.7, the system enters the accuracy priority mode, ensuring temperature control accuracy ≤ ±0.1℃ and humidity control accuracy ≤ ±2% by increasing the cooling capacity adjustment frequency and sensor sampling frequency. When P < 0.7, the system enters the energy consumption priority mode, optimizing the ratio of cooling capacity output to heat recovery. The optimization formula is as follows:

[0023]

[0024] in, This is the energy consumption baseline value in precision-priority mode. The target energy consumption value in the energy consumption priority mode is reduced compared to the precision priority mode by increasing the heat recovery weight and reducing redundant cooling output.

[0025] Furthermore, the sensor array includes 16-24 Pt100A-grade temperature sensors and 4-6 high-precision humidity sensors. The temperature sensors are evenly distributed within a 1m radius around the unit in the test room, with a measuring point set every 0.5m to form a three-dimensional monitoring network. The humidity sensors are installed on the supply air surface, return air surface, and the center of the test area, and the data acquisition frequency is 10-20Hz. The data is transmitted in real time to the edge computing module of the intelligent control unit. After filtering and noise reduction preprocessing, it provides accurate data support for the adjustment of each module.

[0026] Furthermore, the fault self-diagnosis and self-healing module includes a fault feature library, a deep learning recognition unit, and a hierarchical response unit;

[0027] The fault feature library pre-stores 10-15 types of typical fault feature parameters for core components such as compressors, valves, sensors, and heat exchangers;

[0028] The deep learning recognition unit extracts feature vectors from real-time sensor data and matches them with a feature library.

[0029] The graded response unit automatically adjusts the drive parameters or corrects the data offset to achieve self-healing for minor faults, and switches to the backup circuit within 1 second for serious faults, while triggering an audible and visual alarm and recording the fault time, type and data for easy maintenance later.

[0030] Furthermore, the intelligent control unit is equipped with an industrial-grade PLC controller and an edge computing module, and integrates LabVIEW control software to support remote monitoring, parameter adjustment, and data export; and realizes data interaction between modules through the Modbus TCP protocol, with a closed-loop control cycle of ≤100ms.

[0031] The intelligent control unit also has a built-in sub-module for full lifecycle management of equipment, which can continuously record operating parameters, energy consumption data, fault information and maintenance records, and generate system optimization suggestions through data analysis.

[0032] The beneficial effects of this invention are:

[0033] 1. This invention achieves the unity of temperature and humidity coordinated control, coupling interference elimination and temperature field uniform optimization by using the energy complementary design of the integrated cooling and dehumidification circuit, the interference blocking mechanism of the dynamic coupling self-learning decoupling unit, and the adaptive adjustment of the flow field cooling capacity linkage optimization structure. This provides a stable and accurate simulation environment for the test products, ensures the reliability and repeatability of the test data, and solves the pain points of temperature and humidity regulation conflicts and uneven local environment in traditional systems.

[0034] 2. This invention relies on the full-range cooling capacity adjustment and tiered waste heat recovery design of the three-stage cooling and heat recovery collaborative module, combined with the mode adaptive optimization of the energy consumption accuracy dynamic balance module. It can accurately match the dynamic load demand of the laboratory, maximize the utilization of recovered energy, and reduce redundant energy consumption, achieving the dual goals of wide load coverage and deep energy saving. It breaks through the problem of the limitations of traditional system cooling capacity adjustment and the inability to balance energy consumption and accuracy.

[0035] 3. In this invention, the advance load prediction capability of the working condition migration prediction system enables the system to adjust its operating status in advance, shorten the working condition switching stabilization time, and, with the graded response mechanism of the fault self-diagnosis and self-healing module, automatically heals minor faults and quickly switches to backup circuits for serious faults, effectively avoiding test interruptions and equipment damage, and significantly improving the continuity of system operation, response speed and maintenance convenience. Attached Figure Description

[0036] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0037] Figure 1 This is a schematic diagram of the overall framework structure of a refrigeration cycle system for precise temperature and humidity control in a wind-cooled simulated environment laboratory proposed in this invention.

[0038] Figure 2 This is a schematic diagram of the operation process of a refrigeration cycle system for precise temperature and humidity control in a wind-cooled simulated environment laboratory proposed in this invention. Detailed Implementation

[0039] To make the technical means and objectives and effects of the present invention easier to understand, the embodiments of the present invention will be described in detail below with reference to specific illustrations.

[0040] like Figure 1-2 As shown, the present invention discloses a refrigeration cycle system for precise temperature and humidity control in an air-cooled simulated environment laboratory, including an integrated refrigeration and dehumidification loop, a three-stage cold and heat recovery collaborative module, a dynamic coupling self-learning decoupling unit, a flow field cold energy linkage optimization structure, an operating condition migration prediction system, an energy consumption accuracy dynamic balance module, and a fault self-diagnosis and self-healing module.

[0041] Each module constructs a millisecond-level closed-loop collaborative network via industrial Ethernet, with a network transmission rate of ≥100Mbps and a data transmission latency of ≤50ms, ensuring real-time command interaction and data feedback between modules. The intelligent control unit, serving as the system's central hub, is equipped with an industrial-grade PLC controller and an edge computing module, integrating LabVIEW control software. The software includes four core modules: real-time monitoring, parameter setting, data export, and fault alarm. It supports data interaction between modules via the Modbus TCP protocol, with a closed-loop control cycle of ≤100ms, ensuring rapid response to control commands.

[0042] Meanwhile, the intelligent control unit has a built-in sub-module for managing the entire life cycle of the equipment. It records system operating parameters such as temperature and humidity, cooling output, and energy consumption data every minute, and records fault information and maintenance records in real time. Fault information includes fault codes, fault causes, and occurrence times, while maintenance records include maintenance time, content, and personnel. Through data analysis, it generates system optimization suggestions, such as adjusting the condensing heat recovery weight coefficient based on long-term operating data and optimizing the sensor sampling frequency, to further improve the system's operational stability and energy-saving effect.

[0043] The integrated refrigeration and dehumidification circuit is based on an integrated composite heat exchanger, with two independent channels arranged in parallel inside: a refrigeration channel and a dehumidification channel.

[0044] Specifically, the two channels share a serrated and corrugated composite fin structure, and the composite fin design can improve the heat exchange efficiency compared with traditional straight fins.

[0045] Channel switching is achieved through an industrial-grade high-frequency electromagnetic three-way valve. In cooling mode, the electromagnetic three-way valve is energized, driving the sealing baffle to close the dehumidification channel, and the cooling channel operates at full load. The refrigerant flows through the channel via copper pipes, and the concentrated output of cooling capacity meets the laboratory's cooling requirements. In dehumidification mode, the electromagnetic three-way valve is de-energized, the sealing baffle opens the dehumidification channel, and the cooling channel outputs 30%-50% of the rated cooling capacity through the flow regulating valve. The refrigerant flow rate is controlled at 5-10 m³ / h, providing stable cooling capacity support for the dehumidification process and preventing excessive temperature drop due to dehumidification.

[0046] The three-stage cooling and heat recovery collaborative module achieves a wide cooling capacity coverage of 10%-100% through a three-stage mechanism of stepless adjustment, constant volume replenishment, and cascade recovery of condensation heat, precisely matching the dynamic load requirements of the laboratory.

[0047] The stepless adjustment unit uses a combination of a variable frequency compressor and an electronic expansion valve. It adjusts the opening of the electronic expansion valve and the compressor speed in real time through a PID algorithm. The PID parameters are set as proportional coefficient Kp=1.2-1.5, integral coefficient Ki=0.05-0.1, and derivative coefficient Kd=0.1-0.2 to ensure that the cooling capacity is continuously adjustable from 10% (3.7kW) to 100% (37kW), with an adjustment accuracy of ≤±1% of the rated cooling capacity.

[0048] The constant-capacity supplement unit is connected in parallel with 2-4 50HP semi-hermetic reciprocating compressors. The system calculates the cooling demand threshold based on the average cooling demand fluctuation over the past 5 minutes. When the cooling output of the stepless adjustment unit is close to the upper limit of the threshold for 3 minutes and still cannot meet the load demand, the system automatically starts one constant-capacity compressor to supplement the cooling capacity. When the load decreases and the cooling demand is lower than the lower limit of the threshold for 3 minutes, the constant-capacity compressors are shut down in sequence to avoid cooling capacity redundancy.

[0049] The condensation heat recovery unit consists of a primary plate heat exchanger, secondary heat pipes, and terminal waste heat recovery coils. The primary plate heat exchanger recovers 70%-80% of the condensation heat. The inlet and outlet temperatures of the hot water tank are collected by a water temperature sensor, and the recovered heat is calculated in conjunction with data from a flow sensor. ,in The specific heat capacity of water, For traffic, The primary heat pipe is a copper water heat pipe that recovers 15%-20% of the condensation heat. The heat exchange efficiency is adjusted based on feedback data from a temperature sensor inside the duct, and used for preheating the supply air to ensure the supply air temperature is ≤5℃ different from the indoor temperature, reducing temperature fluctuations. The terminal waste heat recovery coil is a spiral coil that recovers 5%-10% of the condensation heat, used for preheating the humidifier to raise the humidification water temperature to 20-30℃, improving humidification efficiency by ≥10%.

[0050] Through formula Dynamically allocate three-stage heat recovery, of which , , ,and The system dynamically adjusts the weighting coefficients through an intelligent control unit based on real-time data such as hot water tank level, air supply temperature, and humidification requirements to ensure maximum utilization of recovered heat.

[0051] The dynamic coupling self-learning decoupling unit eliminates mutual interference during temperature and humidity regulation by updating the coupling coefficient in real time. The formula for updating the coupling coefficient is:

[0052]

[0053] Initial coupling coefficient The settings are based on the initial temperature and humidity fluctuations in the laboratory. If the initial temperature and humidity fluctuations are ≤±1℃ and ±5%, then... Take 0.1-0.2; if the fluctuation is > ±1℃ or ±5%, then Set the value to 0.2-0.3 to ensure that coupling interference can be quickly suppressed in the initial stage.

[0054] In the formula, t is the system running time. For integration variables, , The weights are dynamic, corresponding to the humidity change rate, respectively. With temperature change rate The influence weight.

[0055] The self-learning iteration cycle of this unit is set to 5-10 minutes. Within each iteration cycle, the system automatically collects 30 sets of temperature and humidity control data, removes outlier data (deviation > ±3σ) using the 3σ criterion, and retains valid data for optimization. , The value is optimized by using the least squares method to construct the objective function. ,in, , For the temperature and humidity control deviation of the i-th data set, solve for the deviation that makes... smallest , The value enables dynamic weight adaptive optimization.

[0056] When the system detects a temperature and humidity coupling deviation exceeding ±0.5℃ or ±3%, it automatically reduces the gain of the corresponding regulation branch by 20%-30%. For example, when humidity regulation causes temperature fluctuations of ±0.6℃, the PID proportional coefficient of the humidity regulation branch is reduced from 1.2 to 0.96 to avoid further interference of humidity regulation on temperature. Conversely, when temperature regulation causes humidity fluctuations of ±4%, the gain of the temperature regulation branch is reduced by 25% to ensure independent and accurate control of temperature and humidity.

[0057] The flow field cooling capacity linkage optimization structure optimizes the temperature field uniformity in the test room through adaptive adjustment of the physical structure and real-time CFD simulation, ensuring the reliability of the test data.

[0058] The deformable air supply duct in this structure is made of high-strength aluminum alloy. Four piezoelectric ceramic actuators are evenly arranged around the circumference of the duct, which can switch between circular and elliptical cross sections with a cross-sectional area variation range of ±20% to adapt to wind speed adjustment under different cooling requirements.

[0059] The double-layer perforated plate assembly is made of stainless steel. The opening rate of the upper perforated plate is adjustable from 30% to 50%, while the opening rate of the lower perforated plate is fixed at 40%. The hole diameter is 3-5mm and the hole spacing is 10-15mm.

[0060] The adjustable air deflector is made of ABS material and is driven by a stepper motor. The adjustable angle is 0°-45° with an adjustment accuracy of 0.5°, which is used to precisely control the air delivery direction.

[0061] The mixer is located in front of the air outlet of the air handling unit and has spiral guide vanes inside, which can make the airflow fully mixed and avoid uneven local wind speed.

[0062] The CFD real-time simulation uses ANSYS Fluent software, with structured mesh generation and a mesh size ≤ 5mm. The simulation update frequency is 1-2 times / min. During the simulation, real-time data collected by the sensor array, such as temperature, humidity, wind speed, and pressure, are input, and the predicted flow field distribution results are output.

[0063] The sensor array includes 16-24 Pt100A-grade temperature sensors and 4-6 high-precision humidity sensors. The temperature sensors are evenly distributed within a 1m radius around the unit in the test room, with a measuring point set every 0.5m. Their three-dimensional coordinates are x=0-1m, y=0-1m, and z=0-1m, forming a three-dimensional monitoring network.

[0064] Humidity sensors were installed on the supply air surface, return air surface, and center of the test area, with a data acquisition frequency of 10-20Hz.

[0065] The collected data is transmitted to the edge computing module of the intelligent control unit in real time. The Kalman filter algorithm is used to reduce noise, and the preprocessing delay is ≤50ms, providing accurate data input for flow field optimization.

[0066] The intelligent control unit uses sensor data and CFD simulation results, through formulas... Dynamically matching fan speed and cooling capacity, where This is the flow coefficient, with a value ranging from 0.85 to 0.95. The specific value can be dynamically adjusted based on the pipe cross-sectional shape. The real-time cross-sectional area of ​​the pipe is calculated using the displacement of a piezoelectric ceramic actuator. The supply air pressure differential is measured by a differential pressure sensor. The density is calculated based on real-time temperature and humidity. , Atmospheric pressure, The gas constant of air. For absolute temperature, the wind speed uniformity in the test room is ultimately ensured to be ≤ ±0.2 m / s, and the temperature field uniformity is ≤ ±0.5℃.

[0067] The operating condition transition prediction system integrates BP neural network and transfer learning algorithm to output the predicted value of cooling demand 5-10 minutes in advance, allowing sufficient time for system adjustment and avoiding lag in response to operating condition switching.

[0068] The system first sets similar operating condition judgment criteria. When the temperature and humidity setpoint deviation is ≤ ±2℃ and the cooling capacity demand deviation is ≤ ±10%, the similar operating condition matching process is divided into two steps:

[0069] The first step is to use the K-nearest neighbor algorithm (K=5) to select the 5 historical operating conditions that are closest to the current temperature and humidity setpoints;

[0070] The second step is to calculate the similarity between the cooling demand of each group of historical operating conditions and the current operating conditions. Operating conditions with a similarity ≥ 0.8 were selected as migration operating conditions, and their cooling capacity variation patterns were extracted as migration operating condition cooling capacity data. .

[0071] The BP neural network adopts a four-layer structure of "5-10-5-1". The input layer has 5 neurons, which are the current temperature and humidity, historical average cooling load, environmental disturbance data, and cooling load under migration conditions. The hidden layer has two layers, with the first layer having 10 neurons and the second layer having 5 neurons. The output layer has 1 neuron, which is the predicted value of cooling load. The activation function used is ReLU, the training iterations are 1000, and the learning rate is 0.001.

[0072] Transfer learning employs a fine-tuning strategy, freezing the weights of the first two layers of the pre-trained model and only fine-tuning the weights of the last two layers to adapt to the 500 new working condition data sets of the current system, thereby shortening the training time and improving the prediction accuracy.

[0073] Environmental interference data It includes five dimensions: outdoor dry-bulb temperature, outdoor wet-bulb temperature, atmospheric pressure, power grid voltage fluctuation, and the number of times the laboratory door is opened and closed, all collected in real time by corresponding sensors.

[0074] The prediction formula is ,in This is the weight of historical data, taken as 0.4 when the operating conditions are stable and 0.3 when the operating conditions fluctuate. This represents the current data weight; when the proportion of real-time cooling demand is high, it is taken as 0.5. This is the weight of the interference factor; it is set to 0.15 when environmental fluctuations are large. This is the weight of the migration data; it is set to 0.1 when the similarity of the migration conditions is high, and... .

[0075] Tests show that the prediction error of this system is ≤±3%, which is 10%-15% higher than the prediction accuracy of a single BP neural network. For example, in the event of sudden changes in outdoor temperature and humidity, it can still accurately predict changes in cooling demand and start the constant-capacity compressor or adjust the opening of the electronic expansion valve in advance to ensure that the stable switching time of the operating conditions is ≤30 minutes.

[0076] The energy consumption and accuracy dynamic balance module achieves a dynamic balance between accuracy and energy consumption through a configurable priority coefficient P, with a value range of 0-1, to meet the needs of different testing tasks.

[0077] There are two ways to set the priority coefficient P:

[0078] One option is for users to manually set values ​​through the LabVIEW software interface. For example, for high-precision testing of scientific research prototypes, P can be manually set to 0.9, and for routine quality inspection, P can be set to 0.5.

[0079] Secondly, the system automatically adapts to the test task type. The system has a built-in task type library, which includes three types: high-precision test, regular test, and quick screening test. After the user selects the task type, the system automatically matches the corresponding P value: P=0.7-1.0 for high-precision test, P=0.4-0.7 for regular test, and P=0-0.4 for quick screening test.

[0080] When P ≥ 0.7, the system enters the precision priority mode. At this time, the cooling capacity adjustment frequency increases from 1 time / second to 5 times / second, the sensor sampling frequency increases from 10Hz to 20Hz, and the proportional coefficient Kp of the PID algorithm increases by 20%, ensuring that the temperature control accuracy is ≤ ±0.1℃ and the humidity control accuracy is ≤ ±2%.

[0081] For example, in the high-precision testing of a certain electronic component, with the indoor temperature set to 24℃ and humidity to 50%, the system can achieve temperature fluctuation ≤ ±0.08℃ and humidity fluctuation ≤ ±1.5% through this mode.

[0082] When P < 0.7, the system enters the energy-priority mode, as indicated by the formula. Optimize operating parameters, including The energy consumption baseline value in precision-priority mode can be obtained through historical data statistics, such as under a cooling capacity requirement of 400kW. , The target energy consumption value under the energy-priority mode can be calculated based on the heat recovery potential, for example, under a cooling capacity demand of 400kW. .

[0083] The system increases the weight of condensation heat recovery, such as It increased from 0.7 to 0.8. By increasing the value from 0.15 to 0.18, reducing redundant cooling output, and shutting down redundant sensors, energy consumption is reduced by 20%-30% compared to the precision-priority mode. Actual testing shows that with a cooling demand of 400kW, the actual energy consumption of the energy-priority mode is 155kW, which is 22.5% lower than that of the precision-priority mode, significantly improving operational economy.

[0084] The fault self-diagnosis and self-healing module improves system reliability and reduces maintenance costs through a fault feature library, deep learning recognition, and a hierarchical response mechanism.

[0085] This module first constructs a fault characteristic library, pre-storing 10-15 typical fault characteristic parameters for core components such as compressors, valves, sensors, and heat exchangers, specifically including:

[0086] The following faults are listed: current fault (compressor current ≥ 150A for 3 seconds), overheating fault (compressor discharge temperature ≥ 120℃ for 5 seconds), stuck fault (valve drive current ≥ 5A and valve position unchanged for 5 seconds), drift fault (sensor measurement value deviation ≥ ±0.5℃ / ±5% for 10 seconds), and leakage fault (heat exchanger inlet / outlet pressure difference ≤ 0.1MPa for 10 seconds). Each type of fault corresponds to a unique fault code.

[0087] The deep learning recognition unit adopts a CNN convolutional neural network model. The input is the feature vector of real-time sensor data, which includes feature values ​​in 10 dimensions such as temperature, humidity, pressure, current, and voltage. Through model training, the training set contains 5,000 sets of fault data and 10,000 sets of normal data, and the fault recognition accuracy is ≥95%.

[0088] During training, data augmentation techniques, such as adding Gaussian noise and translation transformation, are used to expand the fault data samples and avoid model overfitting.

[0089] The hierarchical response unit performs targeted processing based on the fault identification results:

[0090] For minor faults, such as valve jamming and sensor drift, the self-healing mechanism is activated. When the valve is jammed, the system automatically increases the valve drive current from 3A to 8A in increments of 0.5A for 2 seconds, while simultaneously outputting a pulse signal to impact the valve stem. If the valve position returns to normal after 3 consecutive adjustments, the self-healing is considered successful, and the fault code and self-healing process are recorded.

[0091] When the sensor drifts, the system calculates the drift amount based on calibration data from the past 30 days. , =Average value - Calibration value, obtained through the formula: Correction= Measurement- If the current measurement value is corrected and the deviation after correction is ≤ ±0.2℃ / ±3%, then the self-healing is considered successful.

[0092] For serious faults, such as compressor shutdown and heat exchanger leakage, the system automatically switches to the backup circuit within 1 second, starts the backup constant-volume compressor and the independent electric heating / dehumidification branch, ensures that the temperature and humidity fluctuation in the laboratory is ≤±1℃ / ±5%, triggers audible and visual alarms, and pops up a fault prompt window through the LabVIEW software interface, recording the fault time, type, data and backup circuit start status, so that maintenance personnel can quickly locate the problem later.

[0093] Working principle:

[0094] After the system starts up, the intelligent control unit first completes initialization, loading the preset temperature and humidity setpoints, priority coefficient P, and initial coupling coefficient. The system establishes communication connections with seven major modules, including the integrated refrigeration and dehumidification loop and the three-stage cold and heat recovery collaborative module, through the Modbus TCP protocol. Each module performs self-checks, and after the fault self-diagnosis and self-healing module confirms that there are no faults by comparing with the fault feature database, the system enters standby mode.

[0095] Users select the test task type or manually set the priority coefficient P through the LabVIEW software interface. The intelligent control unit automatically matches the corresponding operating mode according to the task type. If P ≥ 0.7, the accuracy priority mode is activated, increasing the cooling capacity adjustment frequency to 5 times / second and the sensor sampling frequency to 20Hz; if P < 0.7, the energy consumption priority mode is activated, according to the formula... Calculate the target energy consumption value and set the heat recovery weighting coefficient. , , .

[0096] Subsequently, the integrated cooling and dehumidification circuit switches its operating mode according to the test requirements. In cooling mode, the electromagnetic three-way valve drives the sealing baffle to close the dehumidification channel, and the cooling channel operates at full load. In dehumidification mode, the sealing baffle opens, and the cooling channel outputs 30%-50% of its rated cooling capacity as support for dehumidification. The shared sawtooth and corrugated composite fins achieve energy complementarity.

[0097] Simultaneously, the three-stage cooling and heat recovery coordination module is activated. The stepless adjustment unit uses a PID algorithm to adjust the speed of the variable frequency compressor and the opening of the electronic expansion valve, achieving continuous adjustment of cooling capacity from 10% to 100%. The constant-capacity replenishment unit automatically starts and stops 2-4 parallel constant-capacity compressors according to the cooling demand threshold. The condensing heat cascade recovery unit recovers condensing heat through primary plate heat exchangers, secondary heat pipes, and terminal waste heat recovery coils, according to the formula... Dynamically allocate heat to the hot water tank, preheat the air supply, and preheat the humidifier.

[0098] Dynamically coupled self-learning decoupling unit according to formula The coupling coefficient is updated in real time, with a self-learning iteration cycle of 5-10 minutes. The least squares method is used to optimize the data by selecting the last 30 sets of valid historical data. , When the temperature and humidity coupling deviation exceeds ±0.5℃ or ±3%, the corresponding adjustment branch gain is automatically reduced by 20%-30% to eliminate interference.

[0099] In the flow field cooling capacity linkage optimization structure, the deformable air supply duct adjusts its cross-sectional shape via a piezoelectric ceramic actuator, the double-layer perforated plate assembly adjusts the upper layer's opening ratio, the adjustable guide plate adjusts its angle within the 0°-45° range, and the helical guide vanes of the mixer ensure thorough airflow mixing. CFD real-time simulation updates the flow field prediction results every 1-2 times / min. The intelligent control unit, based on data collected by a three-dimensional monitoring network consisting of 16-24 Pt100A-grade temperature sensors and 4-6 high-precision humidity sensors, calculates the flow field predictions according to the formula... Dynamically match wind speed and cooling capacity to ensure temperature field uniformity ≤ ±0.5℃.

[0100] The operating condition migration prediction system filters similar operating conditions based on the criteria of temperature and humidity setpoint deviation ≤ ±2℃ and cooling demand deviation ≤ ±10%, and extracts the cooling capacity data of the migrating operating conditions through a transfer learning algorithm. Combined with a BP neural network according to the formula The system outputs the predicted cooling demand value 5-10 minutes in advance, and the intelligent control unit adjusts the compressor's operating status and valve opening based on the prediction results.

[0101] During operation, the energy consumption accuracy dynamic balance module monitors the temperature and humidity control accuracy and energy consumption data in real time. If the accuracy does not meet the standard, it automatically increases the cooling capacity adjustment gain. If the energy consumption exceeds the target value, it increases the heat recovery weight. The fault self-diagnosis and self-healing module continuously extracts sensor data feature vectors and matches them with the fault feature library. When it identifies minor faults such as valve jamming or sensor drift, it automatically adjusts the drive parameters or corrects the data offset to achieve self-healing. When it identifies serious faults such as compressor shutdown or heat exchanger leakage, it switches to the backup circuit within 1 second and triggers an audible and visual alarm, and records the fault information.

[0102] The intelligent control unit's equipment lifecycle management submodule continuously records operating parameters, energy consumption data, fault information, and maintenance records. It generates system optimization suggestions through data analysis. Each module achieves real-time data interaction through a millisecond-level closed-loop collaborative network built via industrial Ethernet, ensuring accurate, energy-saving, and stable operation of the system throughout the entire process until the test task is completed. Users can export test data and operation reports through software, and the system automatically shuts down or enters standby mode.

[0103] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A refrigeration cycle system for precise temperature and humidity control in a wind-cooled simulated environment laboratory, characterized in that, It includes an integrated cooling and dehumidification loop, a three-stage cold and heat recovery collaborative module, a dynamic coupling self-learning decoupling unit, a flow field cold energy linkage optimization structure and a working condition migration prediction system. Each module is connected to a millisecond-level closed-loop collaborative network via industrial Ethernet. The integrated refrigeration and dehumidification circuit adopts a shared finned heat exchange channel design to achieve complementary energy for refrigeration and dehumidification. The three-stage cold and heat recovery collaborative module combines stepless adjustment, constant volume replenishment, and condensation heat cascade recovery to cover a wide cold capacity range of 10%-100%. The dynamic coupling self-learning decoupling unit updates the coupling coefficient based on real-time data iteration to eliminate temperature and humidity regulation interference. The flow field cold capacity linkage optimization structure optimizes the temperature field uniformity through real-time CFD simulation and adaptive adjustment of the physical structure. The operating condition migration prediction system integrates BP neural network and transfer learning to output load demand in advance. It also includes an energy consumption accuracy dynamic balance module and a fault self-diagnosis and self-healing module. The energy consumption accuracy dynamic balance module dynamically allocates the ratio of cooling output and heat recovery through configurable priority coefficients and optimization algorithms, and adaptively switches between accuracy priority and energy consumption priority modes. The fault self-diagnosis and self-healing module realizes fault classification diagnosis and self-healing response by constructing a fault feature library and deep learning recognition algorithm. In the three-stage cold and heat recovery collaborative module, the condensation heat cascade recovery unit includes a primary plate heat exchanger, a secondary heat pipe, and a terminal waste heat recovery coil, which are used for heating the hot water tank, preheating the supply air, and preheating the humidifier, respectively, and dynamically distribute the recovered heat according to the following formula: Dynamically allocate and recover heat; in, , , These are the weighting coefficients for the three levels of recycling. , , ,and , , , Each stage involves three levels of heat recovery. The coupling coefficient of the dynamically coupled self-learning decoupling unit can be updated in real time, and its update formula is as follows: ； in, The initial coupling coefficients are... For runtime, For integration variables, , The weights are dynamic, and the self-learning iteration cycle is 5-10 minutes. The least squares method is used to optimize the algorithm by selecting the last 30 sets of valid historical data. , When the temperature and humidity coupling deviation exceeds ±0.5℃ or ±3%, the gain of the corresponding regulating branch will be automatically reduced by 20%-30% to eliminate regulation interference. The flow field cooling capacity linkage optimization structure includes a deformable air supply duct, a double-layer orifice plate assembly, an adjustable guide plate, and a mixer. The CFD real-time simulation update frequency is 1-2 times / min. The intelligent control unit dynamically matches the wind speed and cooling capacity based on sensor array data using the following formula: ； in For flow coefficient, This is the real-time cross-sectional area of ​​the pipeline. For the supply air pressure difference, air density; In the aforementioned operating condition migration prediction system, the criteria for determining similar operating conditions are a temperature and humidity setpoint deviation of ≤ ±2℃ and a cooling demand deviation of ≤ ±10%. The cooling demand variation patterns of similar operating conditions are extracted using a transfer learning algorithm, and the prediction formula is optimized as follows: ； in, Weights are assigned to historical data. , As the current data weight, , , To migrate data weights, ,and , To migrate the operating condition cooling data, This is data related to environmental interference.

2. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The integrated refrigeration and dehumidification circuit includes an integrated composite heat exchanger. The heat exchanger has a refrigeration channel and a dehumidification channel arranged in parallel. The two channels share a sawtooth and corrugated composite fin structure. The working mode is switched by an electromagnetic three-way valve. In the refrigeration mode, the dehumidification channel is closed by a sealing baffle. In the dehumidification mode, the refrigeration channel outputs 30%-50% of the rated cooling capacity as the dehumidification cooling capacity support.

3. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, In the three-stage cooling and heat recovery collaborative module, the stepless adjustment unit adopts a combination of variable frequency compressor and electronic expansion valve, and achieves continuous adjustment of cooling capacity from 10% to 100% through PID algorithm. The constant volume supplement unit connects 2-4 constant volume compressors in parallel, and automatically starts and stops according to the cooling capacity demand threshold.

4. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The initial coupling coefficient The settings are based on the initial temperature and humidity fluctuations in the laboratory: if the initial temperature and humidity fluctuations are ≤±1℃ and ±5%, then... Take 0.1-0.2; if the fluctuation is > ±1℃ or ±5%, then Take 0.2-0.

3.

5. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The deformable air supply duct adjusts its cross-sectional shape via a piezoelectric ceramic actuator, with a cross-sectional area variation range of ±20%; the upper layer of the double-layer perforated plate has an adjustable opening rate of 30%-50%, while the lower layer has a fixed opening rate of 40%; the guide plate has an adjustable angle of 0°-45°; and the mixer is located in front of the air outlet of the air handling unit and has spiral guide vanes inside.

6. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The BP neural network adopts a four-layer structure of "5-10-5-1". The input layer has 5 neurons, which are the current temperature and humidity, historical average cooling capacity, environmental interference data and migration condition cooling capacity. Transfer learning employs a fine-tuning strategy, freezing the weights of the first two layers of the pre-trained model and only fine-tuning the weights of the last two layers to adapt to the newly added operating data of the current system.

7. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The priority coefficient P of the energy consumption accuracy dynamic balancing module can be manually set by the user or automatically adapted by the system according to the test task. When P ≥ 0.7, the system enters the accuracy priority mode, which ensures temperature control accuracy ≤ ±0.1℃ and humidity control accuracy ≤ ±2% by increasing the cooling capacity adjustment frequency and sensor sampling frequency. When P < 0.7, the system enters the energy consumption priority mode, which optimizes the ratio of cooling capacity output to heat recovery. The optimization formula is as follows: ； in, This is the energy consumption baseline value in precision-priority mode. The target energy consumption value in the energy consumption priority mode is reduced compared to the precision priority mode by increasing the heat recovery weight and reducing redundant cooling output.

8. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 5, characterized in that, The sensor array includes 16-24 Pt100A-grade temperature sensors and 4-6 high-precision humidity sensors. The temperature sensors are evenly distributed within a 1m radius around the unit in the test room, with a measuring point set every 0.5m to form a three-dimensional monitoring network. The humidity sensors are installed on the supply air surface, return air surface, and the center of the test area, and the data acquisition frequency is 10-20Hz. The data is transmitted in real time to the edge computing module of the intelligent control unit. After filtering and noise reduction preprocessing, it provides accurate data support for the adjustment of each module.

9. The air-cooled simulated environment laboratory precise temperature and humidity control refrigeration cycle system according to claim 1, characterized in that, The fault self-diagnosis and self-healing module includes a fault feature library, a deep learning recognition unit, and a hierarchical response unit. The fault feature library pre-stores 10-15 types of typical fault feature parameters for core components such as compressors, valves, sensors, and heat exchangers; The deep learning recognition unit extracts feature vectors from real-time sensor data and matches them with a feature library. The graded response unit automatically adjusts the drive parameters or corrects the data offset to achieve self-healing for minor faults, and switches to the backup circuit within 1 second for serious faults, while triggering an audible and visual alarm and recording the fault time, type and data for easy maintenance later.

10. A precise temperature and humidity control refrigeration cycle system for an air-cooled simulated environment laboratory according to claim 5, characterized in that, The intelligent control unit is equipped with an industrial-grade PLC controller and an edge computing module, and integrates LabVIEW control software to support remote monitoring, parameter adjustment and data export. Furthermore, data interaction between modules is achieved through the Modbus TCP protocol, with a closed-loop control cycle of ≤100ms; The intelligent control unit also has a built-in sub-module for full lifecycle management of equipment, which can continuously record operating parameters, energy consumption data, fault information and maintenance records, and generate system optimization suggestions through data analysis.

Images