Compressor refrigerant migration detection method and system based on swimming pool heating and ventilation equipment

Abstract

The invention provides a compressor refrigerant migration detection method and system based on swimming pool heating and ventilation equipment. The method comprises the steps that a heating and ventilation equipment monitoring network is established and used for collecting refrigerant state parameters of all key components in heating and ventilation equipment in real time; acquiring swimming pool environment parameters in real time in the swimming pool environment where the heating and ventilation equipment is located, inputting the swimming pool environment parameters into the corrosion rate prediction model based on the corrosion kinetic equation, and predicting durability parameters of the heating and ventilation equipment in the swimming pool environment; according to the structure of the heating and ventilation equipment, a space-time diagram convolutional network is adopted to establish a refrigerant state transition prediction model of the heating and ventilation equipment; through a refrigerant state transition prediction model, according to the refrigerant state parameters of all the key components, refrigerant transition coefficients and candidate transition paths between all the key components in the heating and ventilation equipment are predicted in real time; and based on the durability parameter, the refrigerant migration coefficient and the candidate migration path, a refrigerant circulation loop in the heating and ventilation equipment is regulated and controlled through a fuzzy control algorithm, and the refrigerant migration phenomenon is actively inhibited.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of data processing, and more particularly, embodiments of the present application relate to a compressor refrigerant migration detection method and system based on a pool heating and cooling equipment. BACKGROUND

[0002] In the pool heating and cooling equipment, after the compressor stops, the refrigerant migrates to the evaporator or the suction pipe due to gravity or pressure difference and accumulates. This phenomenon is called refrigerant migration.

[0003] Refrigerant migration can cause liquid refrigerant to be sucked into the cylinder when it is started again, causing the phenomenon of liquid strike, resulting in mechanical damage to components such as pistons and connecting rods. In addition, refrigerant migration can also cause the actual amount of compressed gas in the compressor to decrease, the compression ratio to decrease, and the energy efficiency to decrease. At the same time, the mixture of refrigerant and lubricating oil can cause oil line blockage or coking, further affecting the stability of the system.

[0004] In related technologies, the crankcase heater continuously heats the lubricating oil, so that the temperature in the crankcase is always higher than the temperature of the evaporator (usually maintaining a temperature difference of 5-10°C), thereby eliminating the driving force of refrigerant migration due to pressure difference. This refrigerant migration suppression method mainly relies on the crankcase heater or the gas-liquid separator, but the heater has high energy consumption and is difficult to maintain for a long time, and the evacuation shutdown operation is complex. Especially in the pool environment with high humidity and high chlorine ion concentration, traditional refrigerant management solutions such as crankcase heaters are difficult to continue running after shutdown, and cannot effectively suppress refrigerant migration. Therefore, it is urgent to design a new refrigerant migration suppression scheme to solve at least one of the above technical problems. SUMMARY

[0005] In this context, embodiments of the present application aim to provide a compressor refrigerant migration detection method and system based on a pool heating and cooling equipment, which can intelligently regulate and control the refrigerant circulation loop in the heating and cooling equipment, actively suppress the refrigerant migration phenomenon, and improve the durability of the pool heating and cooling equipment and prolong the service life of the compressor.

[0006] In a first aspect of the embodiments of the present application, a compressor refrigerant migration detection method based on a pool heating and cooling equipment is provided, comprising:

[0007] Pressure sensors and / or temperature sensors are deployed in key components of the heating and cooling equipment to establish a heating and cooling equipment monitoring network for real-time collection of refrigerant state parameters of each key component in the heating and cooling equipment; the key components at least include: a compressor, an evaporator, and a suction pipe;

[0008] The pool environment parameters are collected in real time in the pool environment where the heating and ventilation equipment is located, and the pool environment parameters at least include temperature, humidity, air quality, chlorine concentration, and ventilation condition, and the pool environment parameters are input into a corrosion rate prediction model based on a corrosion kinetics equation to predict the durability parameters of the heating and ventilation equipment in the pool environment.

[0009] A refrigerant state migration prediction model of the heating and ventilation equipment is established by adopting a space-time graph convolution network according to the structure of the heating and ventilation equipment, wherein a thermodynamics-hydrodynamics coupling model is used to infer the refrigerant migration relationship between the key components in the refrigerant state migration prediction model, so as to simulate the refrigerant circulation loop between the key components.

[0010] Through the refrigerant state migration prediction model, the refrigerant migration coefficient and the candidate migration path between the key components in the heating and ventilation equipment are predicted in real time according to the refrigerant state parameters of the key components, wherein the higher the refrigerant migration coefficient is, the greater the probability of refrigerant migration to the corresponding component is.

[0011] Based on the durability parameters, the refrigerant migration coefficient, and the candidate migration path of the key components, the refrigerant circulation loop in the heating and ventilation equipment is regulated by a fuzzy control algorithm to actively inhibit the refrigerant migration phenomenon.

[0012] In a second aspect of the embodiments of the present application, a compressor refrigerant migration detection system based on a pool heating and ventilation equipment is provided, comprising:

[0013] The acquisition module is configured to deploy pressure sensors and / or temperature sensors in the key components of the heating and ventilation equipment, establish a heating and ventilation equipment monitoring network, and collect refrigerant state parameters of the key components in the heating and ventilation equipment in real time; the key components at least include a compressor, an evaporator, and a suction pipe.

[0014] The environment monitoring module is configured to collect pool environment parameters in real time in the pool environment where the heating and ventilation equipment is located, and the pool environment parameters at least include temperature, humidity, air quality, chlorine concentration, and ventilation condition, and input the pool environment parameters into a corrosion rate prediction model based on a corrosion kinetics equation to predict the durability parameters of the heating and ventilation equipment in the pool environment.

[0015] The prediction module is configured to establish a refrigerant state migration prediction model of the heating and ventilation equipment by adopting a space-time graph convolution network according to the structure of the heating and ventilation equipment, wherein a thermodynamics-hydrodynamics coupling model is used to infer the refrigerant migration relationship between the key components in the refrigerant state migration prediction model, so as to simulate the refrigerant circulation loop between the key components; through the refrigerant state migration prediction model, the refrigerant migration coefficient and the candidate migration path between the key components in the heating and ventilation equipment are predicted in real time according to the refrigerant state parameters of the key components, wherein the higher the refrigerant migration coefficient is, the greater the probability of refrigerant migration to the corresponding component is.

[0016] a control module configured to regulate a refrigerant circulation loop in the heating and ventilation device by a fuzzy control algorithm based on the endurance parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, to actively suppress the refrigerant migration phenomenon.

[0017] In a third aspect of the embodiments of the present application, a terminal device is provided, comprising at least one processor, a memory, and an input-output unit; wherein the memory is configured to store a computer program, and the processor is configured to invoke the computer program stored in the memory to execute the compressor refrigerant migration detection method based on the pool heating and ventilation device according to any one of the first aspect.

[0018] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, comprising instructions which, when executed on a computer, cause the computer to execute the compressor refrigerant migration detection method based on the pool heating and ventilation device according to any one of the first aspect.

[0019] In a fifth aspect of the embodiments of the present application, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the compressor refrigerant migration detection method based on the pool heating and ventilation device according to any one of the first aspect.

[0020] According to the compressor refrigerant migration detection method and system based on the pool heating and cooling equipment according to the embodiment of the application, first, pressure sensors and / or temperature sensors are deployed in key components of the heating and cooling equipment, a heating and cooling equipment monitoring network is established, and refrigerant state parameters of each key component in the heating and cooling equipment are collected in real time; the key components at least include a compressor, an evaporator, and a suction pipe. Further, pool environment parameters are collected in real time in the pool environment where the heating and cooling equipment is located, the pool environment parameters at least include temperature, humidity, air quality, chlorine concentration, and ventilation condition, and the pool environment parameters are input into a corrosion rate prediction model based on a corrosion kinetics equation to predict a durability parameter of the heating and cooling equipment in the pool environment. Then, a refrigerant state migration prediction model of the heating and cooling equipment is established according to the structure of the heating and cooling equipment by using a space-time graph convolution network. The thermodynamic-hydrodynamic coupling model is used to infer the refrigerant migration relationship between each key component in the refrigerant state migration prediction model, so as to simulate the refrigerant circulation loop between each key component. Then, through the refrigerant state migration prediction model, the refrigerant migration coefficient and the candidate migration path between each key component in the heating and cooling equipment are predicted in real time according to the refrigerant state parameters of each key component; the higher the refrigerant migration coefficient is, the greater the probability of refrigerant migration to the corresponding component is. Finally, based on the durability parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, the refrigerant circulation loop in the heating and cooling equipment is regulated by using a fuzzy control algorithm to actively inhibit the refrigerant migration phenomenon. Through the embodiment of the application, the refrigerant migration phenomenon of the compressor can be inhibited in the running state or the shutdown state, and the mechanical damage, the decrease of the energy efficiency ratio, the oil path blockage, and the like of the heating and cooling equipment caused by the refrigerant migration can be avoided, and the stability of the equipment system is improved. Especially, the embodiment of the application can adapt to the high-frequency start-stop characteristics of the pool equipment in the high-humidity and high-chlorine environment of the pool, improve the durability of the pool heating and cooling equipment, and prolong the service life of the compressor. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flowchart of a compressor refrigerant migration detection method based on a pool heating and cooling equipment according to the application is shown.

[0022] Figure 2 A structural diagram of a compressor refrigerant migration detection system based on a pool heating and cooling equipment according to the application is shown.

[0023] Figure 3 A structural diagram of a medium according to the embodiment of the application is shown. DETAILED DESCRIPTION

[0024] The following refers to Figure 1 , Figure 1A flowchart of a compressor refrigerant migration detection method based on a pool heating and ventilation device is provided for an embodiment of the present application. It should be noted that the embodiments of the present application can be applied to any applicable use scenario and / or maintenance scenario of underwater operation systems.

[0025] To address at least one of the foregoing issues, embodiments of the present application provide a compressor refrigerant migration detection method and system based on a pool heating and ventilation device.

[0026] Figure 1 The flow of a compressor refrigerant migration detection method based on a pool heating and ventilation device provided by an embodiment of the present application is shown, which includes:

[0027] Step S101, deploying pressure sensors and / or temperature sensors in key components of the heating and ventilation device, establishing a heating and ventilation device monitoring network for real-time collection of refrigerant state parameters of each key component in the heating and ventilation device.

[0028] The heating and ventilation device refers to a system and related equipment that serves heating, ventilation, and air conditioning in a building. Embodiments of the present application mainly involve heating and ventilation devices in a pool or pool venue. In a pool environment, in addition to achieving basic heating, ventilation, and air conditioning functions, the heating and ventilation device also needs to cope with the influence of factors such as high humidity and high chlorine environment specific to the pool, to ensure indoor air quality and suitable temperature conditions.

[0029] In the embodiments of the present application, the key components at least include: compressor, evaporator, and suction pipe. These components are closely related to the sensitivity of core function, refrigerant state change, and high-risk nature of refrigerant migration in the refrigerant circulation system. The compressor is the power core of the refrigerant circulation system, which compresses and transports refrigerant to drive the entire refrigerant circulation. During the compression process, the pressure and temperature of the refrigerant will change significantly. If the refrigerant migrates abnormally, such as the presence of liquid refrigerant entering the compressor, it will cause the working conditions of the compressor to deteriorate. Liquid refrigerant is incompressible, and entering the compressor may cause liquid hammer phenomenon, damaging the piston, valve, and other components of the compressor, seriously affecting the normal operation and service life of the compressor. Therefore, real-time monitoring of the state parameters of the refrigerant in the compressor can timely detect signs of refrigerant migration, ensuring the safe and stable operation of the compressor.

[0030] The evaporator is a key component for refrigerant evaporation and heat exchange. In the evaporator, the refrigerant absorbs heat from the surrounding environment and evaporates, achieving the effect of refrigeration. The state of the refrigerant in the evaporator is very complex, and its pressure, temperature and phase will change with the evaporation process. If the refrigerant migration is abnormal, such as insufficient refrigerant flow or uneven distribution, it will affect the heat exchange efficiency of the evaporator, leading to a decrease in refrigeration effect. In addition, abnormal migration of refrigerant in the evaporator can also cause local overheating or supercooling, damaging the heat exchange tubes and other components of the evaporator. Since the refrigerant in the evaporator is in a low-pressure, low-temperature state, the temperature difference with the external environment is large, and the phase change of the refrigerant is relatively violent, so the refrigerant in the evaporator is more prone to migration.

[0031] The suction pipe is an important channel connecting the compressor and the evaporator, which is responsible for transporting the low-temperature and low-pressure refrigerant evaporated in the evaporator to the compressor. The state of the refrigerant in the suction pipe directly affects the suction effect and running stability of the compressor. If there is liquid refrigerant or poor refrigerant flow in the suction pipe, it will cause unstable suction pressure of the compressor, affecting the normal operation of the compressor. In addition, the migration of refrigerant in the suction pipe can also cause insufficient refrigerant intake of the compressor, reducing the refrigeration or heating effect. The suction pipe is usually in a relatively low temperature and pressure environment, and factors such as its pipe diameter and length also affect the flow characteristics of the refrigerant. Therefore, the suction pipe is one of the sensitive parts of refrigerant migration, and is prone to problems such as refrigerant accumulation and poor flow. By monitoring the refrigerant state parameters of the suction pipe in real time, refrigerant migration phenomena can be detected in a timely manner, and appropriate measures can be taken to adjust and ensure the normal flow of refrigerant in the suction pipe and the normal suction of the compressor.

[0032] In addition to the compressor, evaporator, suction pipe and other key components, there are also some other components that have an important influence on the state of the refrigerant and the operation of the equipment, which also belong to the category of key components. For example, the condenser, liquid storage tank, expansion valve.

[0033] For example, when deploying pressure sensors and / or temperature sensors to establish a heating and ventilation equipment monitoring network in 101, factors such as the installation location, installation method and compatibility with components of the sensors need to be considered comprehensively. For the compressor, pressure sensors and temperature sensors are usually installed on the inlet and outlet pipes of the compressor. Installing sensors on the inlet pipe can monitor the pressure and temperature of the refrigerant sucked into the compressor, and understand the state of the refrigerant entering the compressor. Installing sensors on the outlet pipe can monitor the pressure and temperature of the refrigerant discharged by the compressor, and master the working output of the compressor. When installing, ensure that the sensor is tightly connected to the pipe to avoid leakage and other problems, and consider the protection level of the sensor to adapt to the high humidity and other conditions in the pool environment.

[0034] For the evaporator, sensors are installed at some key positions of its inlet and outlet pipes and the evaporator body. The sensors on the inlet pipe of the evaporator are used to monitor the state of the refrigerant entering the evaporator, and the sensors on the outlet pipe are used to monitor the state of the refrigerant after heat exchange in the evaporator. The sensors installed on the evaporator body can more directly understand the distribution and state change of the refrigerant inside the evaporator. During installation, attention should be paid to the position of the sensors that will not affect the normal heat exchange function of the evaporator, and good fixation and sealing work of the sensors should be done.

[0035] The sensors installed on the suction pipe can monitor the pressure and temperature of the refrigerant in the suction pipe in real time, which helps to understand the suction state of the compressor and the initial state of the entire refrigerant circulation. During installation, good contact between the sensor and the suction pipe should be ensured, and the influence of factors such as vibration of the suction pipe on the sensor should be considered, and appropriate damping measures should be taken.

[0036] In addition, for other key components such as the condenser, liquid tank and expansion valve, sensors will also be installed according to their structure and functional characteristics. For example, sensors are installed on the inlet and outlet pipes of the condenser to monitor the state change of the refrigerant during condensation. Sensors are installed on the inlet or outlet or inside the liquid tank to understand the refrigerant in the liquid tank. Sensors are installed on the front and rear pipes of the expansion valve to monitor the pressure and temperature change of the refrigerant before and after the expansion valve. Thus, a monitoring network that can comprehensively and real-time monitor the state of the refrigerant in the HVAC equipment is established.

[0037] In step S102, the pool environment parameters are collected in real time in the pool environment where the HVAC equipment is located, and the pool environment parameters are input into the corrosion rate prediction model based on the corrosion kinetics equation to predict the durability parameters of the HVAC equipment in the pool environment.

[0038] In the embodiments of the present application, the pool environment parameters at least include temperature, humidity, air quality, chlorine concentration, and ventilation condition.

[0039] It can be understood that the corrosion rate prediction model based on the corrosion kinetics equation is an algorithmic model for predicting the corrosion rate and durability parameters of HVAC equipment in a specific environment, such as a swimming pool environment. Based on the corrosion kinetics equation, the model takes into account the influence of various environmental factors on metal corrosion and can accurately predict the corrosion of metal surfaces under different environmental conditions and the remaining service life of the equipment. The model mainly includes the following structures: first, the data preprocessing module, which is used to process the input swimming pool environment parameters, eliminate the dimensional differences between different parameters, and map various environmental parameters to a unified numerical interval, so that the subsequent model can accurately process and analyze these data. The second is the corrosion rate prediction layer, which is based on the electrochemical corrosion mechanism and is the core part of the model, which is used to predict the corrosion rate of the metal surface according to the preprocessed environmental parameters. As can be seen, the main function of the model is to predict the corrosion rate of the metal surface of the HVAC equipment in the corresponding environment according to the swimming pool environment parameters, and then calculate the cumulative corrosion depth of the key components of the equipment, and finally generate the durability parameters reflecting the remaining life of the equipment. Through these functions, the corrosion of the equipment in a specific environment can be evaluated in advance, providing a scientific basis for the maintenance and management of the equipment, and helping to take timely measures to prevent the equipment from being damaged due to corrosion and prolong the service life of the equipment.

[0040] Exemplarily, when constructing the corrosion rate prediction model based on the corrosion kinetics equation, first, based on the basic principles of electrochemical corrosion, the corrosion process of metal in electrolyte solution (such as swimming pool water) is studied in depth to determine the main factors affecting the corrosion rate, such as temperature, humidity, chlorine concentration, etc. Then, through a large number of experimental studies, collect data on metal corrosion under different environmental conditions, and use these data to calibrate and verify the model. At the same time, combined with historical data analysis and regression, determine various parameters in the model, such as material characteristic parameters, environmental sensitivity factors and corrosion inhibition coefficients, etc. These parameters reflect the characteristics of the metal material itself and the influence degree of environmental factors on corrosion. After repeated experiments and optimization, a model that can accurately predict the corrosion rate and durability parameters is finally constructed.

[0041] In practical application, the corrosion rate prediction model can adopt at least one of the following models: electrochemical kinetics basic model (mixed potential theory model), corrosion rate empirical model (Arrhenius modified model), diffusion control corrosion model (Fick's law coupling model), and multi-physical field coupling model (thermal-flow-electrochemical coupling model), combined with machine learning algorithm to construct.

[0042] Exemplarily, based on the Arrhenius equation in chemical reaction kinetics combined with the temperature dependence of metal corrosion, an environmental sensitivity factor (such as chloride ion concentration, humidity) is introduced to modify the corrosion rate. Thus, a corrosion rate prediction model is constructed in combination with a time series analysis algorithm. The Arrhenius equation can be expressed as: where v is the corrosion rate, A is the pre-exponential factor, Ea is the activation energy, R is the gas constant, T is the thermodynamic temperature, f(CL - ) and f(RH) are the correction functions of chloride ion concentration and relative humidity, respectively. For example, temperature T affects the chemical reaction rate The higher the temperature, the greater the reaction rate, and the faster the reaction speed.

[0043] As an optional embodiment, in 102, the pool environment parameters are input into a corrosion rate prediction model based on a corrosion kinetics equation to predict the durability parameter of the heating and ventilation equipment in the pool environment, including: normalizing the pool environment parameters to eliminate the dimensional differences of different parameters, and mapping the temperature, humidity, chlorine concentration, particulate matter concentration and ventilation rate in the pool environment parameters to a unified numerical interval; based on the electrochemical corrosion mechanism, a corrosion rate prediction layer of the metal surface corrosion rate is established; the normalized pool environment parameters are input into the corrosion rate prediction layer to predict the metal surface corrosion rate under the current pool environment; based on the metal surface corrosion rate, the time is integrated to calculate the cumulative corrosion depth of the key components of the equipment; and according to the ratio between the service life of the equipment and the cumulative corrosion depth of the key components of the equipment, the durability parameter reflecting the remaining life of the equipment is generated.

[0044] In the embodiments of the present application, the model parameters of the corrosion rate prediction layer are determined through experimental calibration and historical data regression analysis. It can be understood that experimental calibration is to measure the corrosion rate under different environmental conditions (such as a specific temperature, chloride ion concentration, humidity, etc.) by conducting controllable experiments on the corrosion process of metal materials in a laboratory environment, thereby obtaining the basic value of the model parameters. This method can directly observe the physical and chemical phenomena of the interaction between materials and the environment, and ensure the matching of the parameters and the actual corrosion mechanism. The historical data regression analysis is to use long-term accumulated equipment operation data or field monitoring data to mine the quantitative relationship between the environmental parameters and the corrosion rate through statistical methods (such as multiple linear regression, machine learning fitting, etc.), and further optimize and verify the parameters calibrated by experiments. In the examples of the present application, experimental calibration provides a benchmark value at the mechanism level, and historical data regression corrects the adaptability of the model to complex environmental changes through big data of actual scenes, so that the parameters are more suitable for the operating environment of the real equipment.

[0045] In the embodiments of the present application, the model parameters of the corrosion rate prediction layer include material characteristic parameters, environmental sensitive factors, and corrosion inhibition coefficients.

[0046] It can be understood that the model parameters of the corrosion rate prediction layer are composed of three types of core variables. First, the material characteristic parameters, which reflect the influence of the inherent properties of the metal material itself on corrosion, such as the composition of the metal (such as the content of alloying elements), the microstructure (such as grain size, surface roughness), hardness, and corrosion-resistant coating characteristics, etc. The corrosion resistance of different materials differs significantly, for example, stainless steel is more corrosion-resistant than ordinary carbon steel due to the presence of chromium, so the material characteristic parameters are one of the basic inputs for model prediction. Second, the environmental sensitive factors, which describe the direct influence of external environmental conditions on the corrosion rate, such as the concentration of chloride ions (the main component of disinfectants in swimming pools), environmental humidity, temperature, air quality (such as the content of oxidizing gases), and ventilation conditions, etc. Taking chloride ions as an example, an increase in its concentration can destroy the passivation film on the metal surface, significantly accelerating the corrosion process, so the environmental sensitive factors are key variables in the model that dynamically adjust the corrosion rate. Finally, the corrosion inhibition coefficients, which reflect the effect of external intervention measures to inhibit corrosion, such as the concentration of corrosion inhibitors used, the integrity of the corrosion-resistant coating, the intensity of the cathodic protection current, etc. The introduction of corrosion inhibition coefficients can quantify the slowing effect of human intervention on the corrosion rate, for example, after adding corrosion inhibitors to the swimming pool water, even if the concentration of chloride ions is high, the corrosion rate may be significantly inhibited, thereby prolonging the service life of the equipment.

[0047] These three types of parameters jointly act on the corrosion rate prediction layer, achieving accurate mapping from environmental input to corrosion rate output. The material characteristic parameters determine the inherent corrosion resistance of the metal in a particular environment, the environmental sensitive factors reflect the intensity of external pressure, and the corrosion inhibition coefficients represent the effect of protective measures. The parameter values determined through experimental calibration and historical data regression analysis can capture the non-linear relationships between these variables. For example, when the concentration of chloride ions exceeds a certain threshold, even if the material itself is highly corrosion-resistant, the corrosion rate may still rise sharply due to the strong oxidizing nature of chloride ions. However, if there is a high concentration of corrosion inhibitors present, the upward trend of the corrosion rate may be inhibited. This dynamic relationship enables the model to adapt to complex working condition changes in the swimming pool environment (such as water quality fluctuations, seasonal temperature changes), providing a reliable theoretical basis for subsequent durability parameter calculation and refrigerant migration inhibition strategy formulation. Ultimately, the accuracy of the model parameters directly determines the precision of the corrosion rate prediction, which in turn affects the reliability of the equipment remaining life assessment, providing scientific support for the preventive maintenance and life optimization of heating and ventilation equipment.

[0048] Specifically, in the above optional embodiment, when the pool environment parameters are input into the corrosion rate prediction model based on the corrosion kinetics equation to predict the durability parameters of the heating and ventilation equipment, the pool environment parameters need to be normalized first. Different environmental parameters have different dimensions and orders of magnitude, for example, temperature may be in Celsius, humidity may be in percentage, chlorine concentration may be in milligrams per liter, etc. These differences will affect the processing and analysis of data by the model. Through normalization, these parameters are mapped to a unified numerical interval, eliminating the influence of dimensional differences. For example, in actual operation, all parameters may be mapped to the interval of 0 to 1. Thus, the model can more accurately process and analyze these data, avoiding calculation errors caused by different dimensions, and improving the accuracy and stability of the model.

[0049] Next, a corrosion rate prediction layer of the corrosion rate of the metal surface is established based on the electrochemical corrosion mechanism. Electrochemical corrosion refers to the corrosion process of a metal in an electrolyte solution due to electrode reactions. In this prediction layer, the influence of the chemical properties of the metal, environmental factors (such as temperature, humidity, chlorine concentration, etc.) on the corrosion rate is considered. Here, according to the electrochemical principle, a mathematical relationship between the corrosion rate of the metal and the environmental parameters is established. For example, an increase in temperature may accelerate the diffusion speed of metal atoms, thereby accelerating the corrosion process; the presence of chloride ions may destroy the passivation film on the metal surface, promoting the occurrence of corrosion. In this way, the model can predict the corrosion rate of the metal surface according to the input environmental parameters. Thus, it can accurately reflect the corrosion of the metal under different environments, providing a reliable basis for subsequent calculation of cumulative corrosion depth and durability parameters.

[0050] Then, the normalized pool environment parameters are input into the corrosion rate prediction layer to predict the corrosion rate of the metal surface under the current pool environment. This step is based on the mathematical model of the corrosion rate prediction layer established in the previous step. The model calculates the corrosion rate of the metal surface according to the input environmental parameters according to the pre-determined mathematical relationship. For example, when the input temperature, humidity, chlorine concentration, etc. parameters are high, the corrosion rate calculated by the model may increase accordingly. Thus, the corrosion rate of the metal under the current environment can be obtained in real time and accurately, providing timely information for the maintenance and management of the equipment.

[0051] Subsequently, the cumulative corrosion depth of the critical components of the equipment is calculated by integrating the metal surface corrosion rate with respect to time. Specifically, the corrosion rate represents the amount of corrosion of the metal surface per unit time. By integrating the corrosion rate over a period of time, the cumulative corrosion of the metal surface over that period of time, i.e. the cumulative corrosion depth, can be obtained. For example, when the corrosion rate data is obtained in the form of discrete sampling at a time interval (e.g. recording the corrosion rate value every hour), the calculation of the cumulative corrosion depth needs to convert the continuous time integration into numerical accumulation at discrete time points. The core logic is to estimate the corrosion amount in each small time interval by piecewise estimation, and then accumulate these small amounts to obtain the total corrosion depth. Specifically, the entire observation period is divided into several adjacent time intervals, and the length of each interval is a fixed time step (such as 1 hour). For each interval, it is assumed that the corrosion rate changes approximately linearly within the interval, and the corrosion amount in the interval can be estimated by the product of the average value of the corrosion rate at the two endpoints of the interval and the time step. The estimated values of all intervals are accumulated, and the cumulative corrosion depth in the entire observation period is obtained. The advantage of this method is that it does not require complex continuous function expressions, but only relies on the actual collected discrete data, which is especially suitable for experimental measurement data or the discretization results of numerical simulation output. Thus, the discretization method simplifies the calculation complexity in practical applications and is suitable for processing experimental or monitoring data.

[0052] In another example, an extended calculation principle is adopted that takes into account the influence of dynamic environmental parameters. When the corrosion rate is dynamically affected by environmental parameters (such as chloride ion concentration, temperature, etc.), the calculation of the cumulative corrosion depth needs to be based on a real-time correlation model between the environmental parameters and the corrosion rate. The core logic is to regard the corrosion rate in continuous time as a function of the environmental parameters, and to reflect the cumulative effect of environmental changes on corrosion by integrating this function in the time dimension. Specifically, the corrosion rate is described by a mathematical model driven by environmental parameters (such as an electrochemical kinetics model considering the influence of chloride ion concentration and temperature), and the real-time changes in environmental parameters directly change the value of the corrosion rate in the model. In the calculation, the time needs to be divided into infinitesimal intervals, and in each infinitesimal interval, the corrosion rate is determined by the environmental parameters at that time through the model, and then the corrosion amount in each infinitesimal interval (corrosion rate multiplied by infinitesimal time) is accumulated. This process is essentially a continuous integration of the corrosion rate under the dynamic changes of environmental parameters, which can accurately reflect the cumulative effect of environmental fluctuations (such as chloride ion concentration peaks, temperature surges) on equipment corrosion. In this way, the long-term corrosion behavior of the equipment under complex environmental conditions can be more accurately evaluated, providing a scientific basis for durability prediction and maintenance decisions. Thus, the dynamic model integration is more consistent with real environmental changes, improving the accuracy of corrosion prediction, and both of them together support the accurate evaluation of the corrosion state of the equipment. In this way, the corrosion degree of the critical components of the equipment over a period of time can be accurately reflected, providing an important basis for evaluating the remaining life of the equipment.

[0053] Finally, according to the ratio between the service life of the equipment and the cumulative corrosion depth of the key components of the equipment, a durability parameter reflecting the remaining life of the equipment is generated. The principle is that the service life of the equipment is determined according to the performance of the material and the use requirement at the design time, and the cumulative corrosion depth reflects the degree of corrosion that the equipment has suffered in the actual use process. By calculating the ratio of the two, the proportion of the remaining service life of the equipment, i.e. the durability parameter, can be obtained. For example, if the designed service life of the equipment is 10 years, and the cumulative corrosion depth corresponds to a proportion of 0.2 after a period of use, then the durability parameter of the equipment is 0.8, indicating that the equipment still has 80% of the service life. Thus, the remaining life of the equipment under the current environment is intuitively reflected, providing a clear reference for the maintenance, replacement and other decisions of the equipment.

[0054] Further optionally, after 102, in combination with the real-time corrosion data fed back by the sensor network, the frequency band signal of the corrosion-sensitive frequency band energy is extracted through wavelet packet transform, so as to ensure that the model adapts to any dynamic scene in the pool water quality fluctuation and the disinfectant dosage change. Further, through the Q-learning algorithm, the mapping relationship between the running parameters and the refrigerant control strategy is dynamically optimized according to the frequency band signal, and by comparing the deviation between the actual corrosion monitoring data and the model prediction value, the environmental sensitive factor and the inhibition coefficient in the corrosion rate prediction model are dynamically adjusted to update the model parameters of the corrosion rate prediction model in real time.

[0055] Here, after completing the durability parameter prediction based on the pool environment parameters, a real-time dynamic optimization mechanism is further introduced to cope with the complex changes of the pool environment. First, through the sensor network deployed in the heating and ventilation equipment and the pool environment, real-time data reflecting the corrosion state of the equipment are continuously collected, which directly reflects the corrosion of the current water quality conditions on the equipment. In order to efficiently extract the key features related to corrosion from these complex signals, wavelet packet transform technology is used for frequency domain analysis of real-time corrosion data. Wavelet packet transform can decompose signals into different frequency bands, accurately identify sensitive frequency bands strongly related to the corrosion process, and extract energy features in these frequency bands as core indicators reflecting water quality fluctuations and corrosion state. This process enables the system to sensitively capture the instantaneous changes of pool water quality, such as sudden changes in chloride ion concentration, adjustments in disinfectant dosage, or interference from suspended particles in water, thereby ensuring that the corrosion rate prediction model is always synchronized with the current water quality conditions.

[0056] In particular, from the frequency characteristics, the sensitive frequency band is usually in a specific frequency range that matches the characteristic frequencies of the physical and chemical changes occurring during the corrosion process. The corrosion process involves chemical reactions on the metal surface, ion migration, and possible changes in microstructure, which generate characteristic signals in a certain frequency range. For example, when metal corrodes, the diffusion of ions in the electrolyte solution, the rupture and repair of the oxide film on the metal surface, and other processes may correspond to vibrations or electromagnetic signals of specific frequencies. The sensitive frequency band can capture these frequency components related to the nature of corrosion, while filtering out interference frequencies unrelated to corrosion. In the pool environment, corrosion-related signals caused by factors such as the corrosion of metal by chloride ions and changes in dissolved oxygen in water will be reflected in a specific frequency range, and the sensitive frequency band can accurately locate and extract these signals.

[0057] In terms of energy distribution, the signal energy in the sensitive frequency band has obvious concentration. When corrosion occurs, the signal energy generated by the related physical and chemical changes will concentrate in certain specific frequency ranges, making the energy of these frequency bands significantly higher than that of other frequency bands. This is because the characteristic changes during the corrosion process have certain regularity and periodicity, which produce stronger signal responses at specific frequencies. By extracting the sensitive frequency band energy signal through wavelet packet transform, these energy concentration characteristics can be highlighted, thus more clearly reflecting the degree and state of corrosion. For example, when the corrosion rate increases, the energy in the sensitive frequency band may significantly increase; when corrosion is inhibited, the energy of the sensitive frequency band will correspondingly decrease. This energy distribution characteristic makes the sensitive frequency band an important indicator for evaluating the corrosion state. In terms of time-frequency characteristics, the sensitive frequency band has good time-frequency localization characteristics. The corrosion process is a dynamic process, and its signal characteristics change over time. The sensitive frequency band can accurately analyze the signal in both time and frequency dimensions, capturing the frequency changes of the corrosion signal at different times and reflecting the evolution of different frequency components over time. This allows the system to track the development of the corrosion process in real time and promptly detect changes in the corrosion state. For example, when the pool water quality suddenly changes, the frequency components and energy distribution of the corrosion signal will change rapidly, and the sensitive frequency band can capture these changes in a timely manner, providing real-time corrosion information for the system to take appropriate measures.

[0058] In addition, the sensitive frequency band is closely related to the physicochemical mechanism of corrosion. Different types and mechanisms of corrosion produce signals with different characteristics, and the sensitive frequency band can extract signal features corresponding to these specific corrosion mechanisms. For example, uniform corrosion and localized corrosion (such as pitting corrosion and crevice corrosion) may produce signals in different frequency ranges, and the sensitive frequency band can accurately distinguish different corrosion types and degrees based on these differences. At the same time, the sensitive frequency band can also reflect the changes in the microstructure of the metal surface during corrosion, such as the growth and rupture of the oxide film, the dissolution and redeposition of metal crystals, etc. These microstructure changes will produce characteristic signals in specific frequency ranges, and by analyzing the signals of the sensitive frequency band, we can gain a deeper understanding of the micro-mechanism of corrosion.

[0059] Finally, the sensitive frequency band has certain stability and repeatability. Under the same corrosion conditions, the position and energy distribution of the sensitive frequency band are relatively stable, which allows the system to establish a reliable corrosion evaluation model based on historical data and experience. Even in different devices and environments, as long as the corrosion mechanism is similar, the characteristics of the sensitive frequency band have certain repeatability, ensuring the accuracy and universality of corrosion monitoring and evaluation. This stability and repeatability make the sensitive frequency band a reliable feature indicator in the field of corrosion monitoring, providing strong support for ensuring the long-term stable operation of equipment.

[0060] Based on the acquisition of the energy signal of the corrosion sensitive frequency band, the Q-learning algorithm in the reinforcement learning framework is further introduced to build a dynamic mapping relationship optimization mechanism between the operating parameters and the refrigerant control strategy. Q-learning learns how to adjust the operating mode (such as start-stop frequency, cooling / heating mode switching) and refrigerant flow control strategy of the equipment under different water quality conditions to minimize corrosion risk and maintain high-efficiency operation through continuous interaction and trial-and-error with the environment. Specifically, according to the real-time extraction of the corrosion sensitive frequency band signal, the effect of the current strategy is evaluated, and the strategy parameters are dynamically adjusted so that the equipment can quickly respond to water quality fluctuations. For example, when the concentration of chloride ions is detected to be rising, leading to an increase in corrosion risk, the system will automatically optimize the operating parameters, reducing the number of high-frequency start-stop of the compressor or adjusting the refrigerant flow distribution, thereby reducing the probability of localized corrosion. This real-time feedback-based optimization mechanism significantly improves the system's adaptability to dynamic environments.

[0061] To ensure the long-term accuracy of the corrosion rate prediction model, a dynamic model parameter updating mechanism is also established. By continuously comparing the actual corrosion monitoring data collected by the sensor network with the model prediction values, the system can identify the source of the prediction error of the model under the current environment. If it is found that the deviation between the prediction value and the actual value exceeds the set threshold, the system will trigger the model parameter correction process, focusing on adjusting the environmental sensitive factors (such as the influence weight of chloride ion concentration on corrosion rate) and corrosion inhibition coefficients (such as the corrosion inhibitor effect correction parameter). This process combines real-time feedback data and historical trend analysis to ensure that the model always reflects the latest water quality conditions and corrosion rules. For example, when the pool disinfectant formula is adjusted, causing changes in the chloride ion corrosion mechanism, the model can quickly adapt to the new corrosion dynamics characteristics through parameter updating, avoiding prediction failure due to model lag. Through the closed-loop process, the complete dynamic adaptation capability from environmental perception to control strategy optimization to model self-updating is achieved, significantly improving the durability and reliability of the heating and ventilation equipment in complex pool environments.

[0062] Further optionally, after 102, the running parameters of the heating and ventilation equipment can also be collected in real time through the equipment control system. The running parameters at least include: equipment running time, running mode, start and stop frequency, compressor load fluctuation amplitude. Then, the running parameters and the durability parameters are correlated to establish a multi-dimensional influence factor matrix. The influence factors at least include: the correlation factor between the start and stop frequency and the corrosion rate of the compressor components obtained by historical data regression analysis, and the influence weight factor of the refrigeration mode proportion on the corrosion rate of the compressor components. A dynamic weight ratio model is constructed based on the multi-dimensional influence factor matrix to adjust the model parameter weight in the corrosion rate prediction model.

[0063] Specifically, after completing the durability parameter prediction based on the environmental parameters, multi-dimensional correlation analysis of the equipment running state is further introduced to improve the accuracy of corrosion prediction. The running parameters of the heating and ventilation equipment are collected in real time through the equipment control system, including the cumulative running time of the equipment, the current running mode (such as refrigeration, heating or standby), the start and stop frequency per unit time, and the fluctuation amplitude of the compressor load. Running time is directly related to the fatigue accumulation effect of metal components, and long-time running may accelerate material aging. Running mode switching changes the refrigerant flow path and pressure distribution, and the stress state of metal components is significantly different under different modes. High-frequency start and stop can cause metal components to experience repeated thermal expansion and contraction and pressure impact, which may induce local corrosion. Compressor load fluctuation directly affects the severity of internal pressure changes, and excessive fluctuation may damage the metal surface protective film and exacerbate the corrosion risk. Real-time monitoring of these running parameters provides key data support for understanding the impact of equipment dynamic behavior on corrosion.

[0064] To quantify the correlation between operating parameters and corrosion rate, the system uses historical data regression analysis method to construct a multi-dimensional influence factor matrix. This matrix integrates the correlation factors between start-stop frequency and compressor component corrosion rate, such as high-frequency start-stop may cause metal surface fatigue crack propagation, thereby accelerating the corrosion process. At the same time, it covers the influence weight factor of refrigeration mode proportion on corrosion rate, and the difference in refrigerant flow state and temperature distribution under different modes will change the electrochemical environment of the metal surface. Through regression analysis, the specific contribution of each operating parameter to the corrosion rate can be determined, such as the specific percentage of compressor component corrosion rate that may increase for each certain number of increase in start-stop frequency, or the influence amplitude of corrosion rate caused by the change of refrigeration mode proportion. These quantitative relationships lay the foundation for subsequent dynamic adjustment of model parameters.

[0065] Based on the multi-dimensional influence factor matrix, the system constructs a dynamic weight ratio model to adjust the parameter weight distribution in the corrosion rate prediction model in real time. According to the characteristics of the current equipment operating state, this model dynamically optimizes the importance proportion of each influencing factor in the model. When a significant increase in start-stop frequency is detected, the model will automatically increase the weight of the correlation factor to more sensitively capture the acceleration effect of high-frequency start-stop on corrosion. When the refrigeration mode proportion increases, the corresponding weight factor is enhanced to reflect the influence of refrigerant circulation state change on the corrosion path. This adaptive adjustment mechanism enables the corrosion rate prediction model to more accurately match the actual operating conditions of the equipment, significantly improving the timeliness and accuracy of the prediction results. By deeply coupling operating parameters and durability parameters, the system realizes the leap from static environment evaluation to dynamic running state monitoring, providing more reliable data support for preventive maintenance and life management of equipment.

[0066] It should be noted that the multi-dimensional influence factor matrix is a data model constructed by systematically analyzing the correlation between the device operating parameters and the corrosion rate, and its core goal is to quantify the influence degree of different operating states on the corrosion process of metal components. The matrix takes the key parameters of device operation as the input dimension, and takes the corrosion rate change as the output response, and extracts the influence weight and interaction law of each parameter through statistical regression and machine learning method. For example, the construction of the matrix depends on the long-term collection of historical operation data and the corresponding corrosion monitoring records. These data cover the device operation time, start-stop frequency, operation mode switching record, and compressor load fluctuation parameters, and are associated with the corrosion rate measurement value in the corresponding time period. Through regression analysis technology, the most significant factors affecting the corrosion rate can be identified. For example, there is a clear positive correlation between the start-stop frequency and the corrosion rate of the compressor components. High-frequency start-stop will cause the metal surface to bear repeated thermal stress and mechanical impact, accelerating the damage of the protective film and the initiation of local corrosion. The refrigeration mode proportion is also confirmed as an important influencing factor. Frequent switching between refrigeration / heating modes can change the refrigerant flow state and metal surface temperature distribution, thereby affecting the electrochemical corrosion process. Each influence factor in the multi-dimensional influence factor matrix is assigned a clear quantitative weight, reflecting its specific contribution to the corrosion rate. The influence factor of start-stop frequency may reflect the percentage increase in corrosion rate for each certain number of start-stop; the influence factor of refrigeration mode proportion may reflect the linear or nonlinear correlation between the proportion of different mode running time and the corrosion rate. In addition, the matrix also captures the interaction effects between parameters, such as the synergistic acceleration of corrosion caused by long-time running and high-frequency start-stop. These weights and interaction relationships are verified and optimized through multi-dimensional regression models to ensure that the matrix can accurately reflect the corrosion law under actual working conditions. The multi-dimensional influence factor matrix formed eventually becomes the basis for dynamically adjusting the corrosion rate prediction model. The system looks up the corresponding influence factor value in the matrix according to the real-time parameters of the current device operation, and adjusts the weight distribution of each environmental parameter and operating parameter in the model accordingly. This dynamic adjustment mechanism enables the corrosion prediction model to adapt to changes in the operating state of the device in real time, significantly improving the prediction accuracy and reliability, and providing a scientific basis for preventive maintenance decisions.

[0067] In step S103, a refrigerant state migration prediction model of the heating and ventilation equipment is established by adopting a space-time diagram convolution network according to the structure of the heating and ventilation equipment. In the embodiment of the present application, a thermodynamics-hydrodynamics coupling model is used to infer the refrigerant migration relationship between each key component in the refrigerant state migration prediction model, so as to simulate the refrigerant circulation loop between each key component.

[0068] In the construction of the refrigerant state migration prediction model for HVAC equipment, the spatio-temporal graph convolution network (ST-GCN) realizes accurate simulation of complex refrigerant circulation loops by fusing equipment structure features and refrigerant dynamic behavior. First, the model constructs a graph network based on the equipment pipeline topology, abstracting key components such as compressors, heat exchangers, and expansion valves as nodes, and the refrigerant flow direction and heat exchange relationship between components as edges. Through graph convolution operations, the network can capture the spatial distribution characteristics of refrigerant among different components, such as the pressure gradient change of refrigerant in the condenser and the phase distribution law in the evaporator. At the same time, the model introduces a time-dimension extended graph convolution layer, enabling it to learn the dynamic laws of refrigerant state evolution over time, such as refrigerant flow mutations caused by high-frequency start-stop and phase change speed changes caused by load fluctuations. This spatio-temporal feature fusion mechanism enables the model to not only analyze the static distribution characteristics of refrigerant but also track its dynamic migration path.

[0069] To enhance the model's ability to represent the physical characteristics of refrigerant, the system introduces a thermodynamics-hydrodynamics coupled model as a physical constraint. By establishing a joint solution framework for the refrigerant state equation (such as the pressure-temperature-enthalpy relationship) and the fluid dynamics equation (such as the Navier-Stokes equation), the model accurately simulates the flow pattern, phase change process, and heat exchange efficiency of refrigerant in the pipeline. For example, at the high-pressure region of the compressor outlet, the model predicts the refrigerant flow velocity distribution by solving the turbulent flow equation; in the phase change region of the evaporator, the model captures the interface evolution of liquid and gas phases by combining the phase field model. These physical field simulation results serve as input features for the spatio-temporal graph convolution network, significantly improving the model's prediction accuracy of refrigerant migration paths. At the same time, the energy conservation constraint provided by the coupled model (such as the balance relationship between refrigerant enthalpy change and compressor power consumption) effectively prevents physically unreasonable predictions that may occur in purely data-driven models.

[0070] During the model training phase, the system uses a transfer learning mechanism to jointly optimize the device historical operation data and physical simulation results. Using a reinforcement learning framework, the model dynamically adjusts the attention weights of the graph convolution network, enabling it to focus on nodes and edges that significantly affect refrigerant migration. For example, when detecting abnormal condenser outlet temperature, the model automatically enhances the feature transmission weight of the edges connected to this node to more sensitively capture refrigerant pressure fluctuation signals. This hybrid modeling method not only inherits the interpretability advantage of physical models but also captures nonlinear coupling effects in actual working conditions through data-driven methods, such as the influence of refrigerant impurity content on flow resistance and the decline in heat exchange efficiency caused by environmental temperature changes. The final refrigerant state migration prediction model can provide key decision support for equipment operation parameter optimization, fault warning, and energy efficiency improvement.

[0071] In an optional embodiment, in 103, a refrigerant state migration prediction model of a heating and cooling equipment is established according to a space-time graph convolution network based on a heating and cooling equipment structure, including: based on a pipeline layout and a component connection relationship of the heating and cooling equipment, a topological structure of a refrigerant circulation network is constructed, key components are abstracted as nodes in a graph network, and pipeline connections between the key components are abstracted as edges; the key components at least include: a compressor, an evaporator, and a suction pipe; each node is given a physical attribute label, and the physical attribute label includes a component type, a material characteristic, and an operating state parameter; thermodynamic and fluid mechanics data are integrated to establish a dynamic coupling model of refrigerant migration; wherein, based on refrigerant phase change characteristics, energy exchange rates at each node are calculated, and based on Bernoulli equation and Reynolds number, flow velocity distribution and pressure drop characteristics of the refrigerant are analyzed; a baseline path of refrigerant migration is generated through finite element simulation as an initial migration mode library of the graph network; a space-time graph convolution network ST-GCN is used to model the refrigerant migration process; through a graph convolution layer, refrigerant state parameters of adjacent nodes are aggregated to generate a refrigerant migration potential field between components; through a time series convolution layer, a hysteresis effect of the refrigerant migration is captured; an environment sensitive factor is introduced to dynamically adjust migration weights, and a refrigerant adsorption coefficient on a metal surface is corrected in real time according to pool environment parameters; based on historical corrosion data, migration inhibition weight parameters of high corrosion risk areas are dynamically allocated to construct a refrigerant state migration prediction model.

[0072] Specifically, in the construction of the refrigerant state migration prediction model of the heating and cooling equipment, the space-time graph convolution network (ST-GCN) realizes fine modeling of the refrigerant migration path by fusing the equipment topological structure and dynamic physical characteristics. The model first establishes a topological structure of a refrigerant circulation network based on the equipment pipeline layout and the component connection relationship, abstracts key components such as compressors, evaporators, and suction pipes as nodes in a graph network, and abstracts pipeline connections between the components as edges. Each node is given a physical attribute label, including a component type (such as a compressor component), a material characteristic (such as a copper pipe thermal conductivity), and a real-time operating state parameter (such as a compressor outlet pressure). This structured modeling method converts the complex refrigerant circulation system into a calculable graph structure, providing a clear physical mapping basis for subsequent analysis.

[0073] To enhance the model's ability to characterize the dynamic behavior of the refrigerant, the system integrates thermodynamic and fluid mechanics data to establish a dynamic coupling model. Based on the refrigerant phase change characteristics (such as the gas-liquid two-phase conversion of R32 refrigerant in the evaporator), the energy exchange rates at each node are calculated to accurately reflect the influence of refrigerant state changes on the migration path; at the same time, the flow velocity distribution is analyzed through Bernoulli equation, the flow state (laminar or turbulent) in the pipeline is evaluated combining with Reynolds number, and the migration resistance parameters between nodes are optimized based on the pressure drop characteristics. The fusion of these physical field data enables the model to accurately simulate the phase change and flow characteristics of the refrigerant in the complex pipeline, such as the prediction of liquid knock risk caused by flow velocity sudden change in the suction pipeline.

[0074] During the model construction process, finite element simulation is introduced to generate refrigerant migration benchmark paths as the initial migration pattern library for the graph network. Through simulation, typical migration trajectories of refrigerant under different operating conditions are obtained, such as the path characteristics of liquid refrigerant migrating to the evaporator due to gravity after the compressor stops. These benchmark paths serve as prior knowledge for the graph convolution network, helping the network quickly converge to reasonable migration patterns. ST-GCN aggregates the refrigerant state parameters (such as pressure, temperature) of adjacent nodes through graph convolution layers to generate a refrigerant migration potential field between components, quantifying the migration tendency of different paths; the time convolution layer captures the hysteresis effect of refrigerant migration, such as the propagation delay phenomenon of refrigerant pressure fluctuations in the pipeline after the compressor starts and stops.

[0075] To adapt to the dynamic changes of the pool environment, the model introduces environment-sensitive factors to dynamically adjust the migration weight. For example, when detecting that the pool water temperature rise leads to a decrease in evaporator heat exchange efficiency, the model automatically enhances the migration weight between the evaporator node and the compressor, reflecting the adjustment needs of the refrigerant circulation path. At the same time, based on historical corrosion data (such as corrosion records caused by liquid impact on the compressor shell), the migration inhibition weight parameters of high corrosion risk areas are dynamically allocated, such as reducing the refrigerant retention probability in the suction line elbow where liquid is prone to accumulate, thereby reducing the adsorption time of corrosive media on the metal surface. This adaptive adjustment mechanism enables the model to respond in real time to changes in device operating state and environmental conditions, providing dynamic decision support for refrigerant migration path optimization and corrosion risk prevention and control.

[0076] The final refrigerant state migration prediction model, through spatio-temporal feature fusion and physical constraint enhancement, not only realizes the accurate simulation of refrigerant circulation paths, but also predicts potential risk areas based on corrosion mechanisms. For example, when the model detects that frequent start-stop of the compressor leads to abnormal refrigerant migration path, it can provide early warning of possible lubrication system pollution or component corrosion, providing a scientific basis for equipment maintenance. This data-driven and physical model combined method breaks through the limitations of traditional empirical formulas, significantly improving the accuracy and engineering practicability of refrigerant migration prediction under complex operating conditions.

[0077] Further optionally, in order to improve the model performance of the refrigerant state migration prediction model, the probability distribution of each candidate migration path is calculated through the forward propagation of the graph network; the path probability is defined as the weighted sum of the inter-node migration potential energy, and the weight is determined by the refrigerant property parameters and environmental factors; the Monte Carlo sampling method is used to generate multiple high-probability migration paths as candidate schemes for the refrigerant redistribution strategy. Further, the model prediction results are verified by the actual refrigerant state data collected by the sensor network, and the prediction error is calculated. If the deviation between the predicted migration path and the actual detected path exceeds the threshold, the model parameter correction mechanism is triggered. The migration learning technology is used to migrate the model training results under historical working conditions to similar equipment, thereby shortening the convergence time of refrigerant migration prediction for new equipment.

[0078] Taking a pool heating system as an example, when the ambient temperature suddenly rises, causing the refrigerant evaporation rate to be abnormal, the model calculates the probability distribution of the refrigerant migration path between each component based on the graph network topology. For example, the weight of the path from the compressor outlet to the condenser may increase due to the increase in ambient temperature (due to the increase in refrigerant saturation pressure), while the weight of the path from the evaporator to the expansion valve may decrease due to liquid level fluctuations. The Monte Carlo sampling method generates ten high-probability migration paths in this scenario, including the regular circulation path, the bypass pressure relief path, and the emergency diversion path. The system selects the optimal scheme after comprehensively evaluating the energy consumption and corrosion risk of each path. In the model verification stage, the refrigerant pressure and temperature data collected by the sensors in real time are compared with the prediction results. If the actual pressure drop value of a certain migration path is 15% higher than the predicted value, the system determines that there is an abnormal deviation and triggers parameter correction. At this time, the model adjusts the attention weight of the graph convolution layer through back propagation, enhances the sensitivity to the evaporator surface fouling resistance parameter, and reduces the weight proportion of the environmental humidity factor, so that the subsequent prediction is more consistent with the actual working condition. It can be seen that the migration learning technology significantly improves the model deployment efficiency. In practical applications, when deploying a new model of pool unit, the historical training results of the same series of equipment in similar climate zones are directly reused. For example, when migrating the refrigerant migration model of a hotel unit in area a to a similar device in area b, only the adjustment coefficient of the environmental sensitive factor needs to be adjusted for the local high-temperature dry environment, and the model convergence time is greatly shortened. This cross-device knowledge migration mechanism effectively solves the problem of refrigerant migration prediction in small sample scenarios and provides technical support for multi-unit collaborative optimization.

[0079] In step S104, the refrigerant state migration prediction model is used to predict the refrigerant migration coefficient between each key component and the candidate migration path in real time according to the refrigerant state parameters of each key component. The higher the refrigerant migration coefficient, the higher the probability of refrigerant migration to the corresponding component.

[0080] As an optional embodiment, in 104, the refrigerant state migration prediction model is used to predict the refrigerant migration coefficient and the candidate migration path between the key components in the heating and ventilation equipment in real time according to the refrigerant state parameters of the key components, including: the ST-GCN-based refrigerant state migration prediction model uses the refrigerant migration potential field, combines the statistical weight of the historical migration path, and calculates the refrigerant migration coefficient between the key components; the migration potential field is generated by aggregating the refrigerant state parameters of adjacent nodes through a graph convolution layer, reflecting the migration tendency of the refrigerant between the components. The real-time migration rate is introduced as a correction factor, and the time series convolution layer of the refrigerant state migration prediction model is used to capture the hysteresis effect of the refrigerant migration, and the matching degree of the refrigerant migration coefficient and the real-time working condition is dynamically adjusted. The ST-GCN-based potential migration path is generated, combined with the refrigerant migration coefficient, the corrosion resistance of the component material, and the current environmental parameters, and the potential migration path is scored in multiple dimensions. The path probability distribution of the potential migration path is generated through Monte Carlo simulation, the confidence interval of each potential migration path is calculated, the confidence is determined by the matching degree of the path probability density function and the historical data, and the environmental sensitive factor is introduced to dynamically correct the confidence weight. According to the upper limit value of the confidence interval, the potential migration path is arranged in descending order, and the potential migration path with the highest confidence is selected as the candidate migration path with high priority. When a high-corrosion-risk area is detected, the ST-GCN-based historical corrosion data dynamically improves the migration inhibition weight of the current area, and the candidate migration path bypassing the high-risk area is preferentially selected.

[0081] In the above embodiment, during the refrigerant migration prediction process of the heating and ventilation equipment, the ST-GCN-based refrigerant state migration prediction model realizes accurate prediction of the migration path through multi-dimensional dynamic analysis. First, the model constructs a refrigerant migration potential field, which aggregates and compresses the refrigerant state parameters (such as pressure, temperature, and flow rate) of key components such as compressors and evaporators through a graph convolution layer, quantifying the migration tendency of the refrigerant between components. For example, when the outlet pressure of the evaporator abnormally rises, the migration potential field will increase the migration weight between the compressor and the condenser, reflecting the trend of the refrigerant accelerating the flow due to the pressure difference. At the same time, the model introduces the real-time migration rate as a correction factor, and uses the time series convolution layer to capture the hysteresis effect of the refrigerant migration. For example, after the compressor starts and stops, the refrigerant pressure fluctuation needs to propagate through the pipeline delay before affecting the downstream components, and dynamic adjustment of the migration coefficient can make the prediction more consistent with the actual working condition.

[0082] In the candidate migration path generation stage, the refrigerant state migration prediction model generates multiple potential migration paths based on the graph structure of ST-GCN, and combines component material corrosion resistance, environmental temperature and humidity, and other parameters for multi-dimensional scoring. Taking a pool heating system as an example, when it is detected that the evaporator surface is scaled and the heat transfer coefficient is reduced, the refrigerant state migration prediction model will reduce the path score through the component, and at the same time increase the migration weight of the bypass pipeline. Monte Carlo simulation further generates path probability distribution, and the confidence interval is calculated by matching the path probability density function with historical data. For example, a certain path has a confidence level exceeding 0.8 in 90% of the historical data, and is listed as a high-priority candidate path. When the environmental parameters change suddenly (such as a sudden drop in outdoor temperature), the environmental sensitive factor dynamically corrects the confidence weight, enhancing the adaptability to real-time working conditions.

[0083] Further optionally, for high corrosion risk areas, the refrigerant state migration prediction model dynamically adjusts the migration inhibition strategy through historical corrosion data. For example, if it is detected that the compressor shell is frequently corroded due to liquid knock, the migration inhibition weight of the area will be increased, and the refrigerant circulation path bypassing the compressor will be preferentially selected. At the same time, combined with the refrigerant migration coefficient and the material corrosion resistance score, high corrosion probability nodes such as copper expansion valves in chlorine-containing refrigerant environments are automatically avoided. This dynamic path optimization mechanism not only reduces the corrosion probability of key components, but also balances system energy consumption and safety performance through intelligent switching of migration paths, providing protection for long-term stable operation of heating and ventilation equipment.

[0084] Exemplarily, in the ST-GCN-based refrigerant state migration prediction model, the calculation of the refrigerant migration coefficient is realized by dynamically optimizing the fusion of spatio-temporal features and historical statistical rules. Firstly, the refrigerant migration potential field is constructed, and the refrigerant state parameters (such as pressure, temperature and flow rate) of key components such as compressors and evaporators are aggregated and compressed through the graph convolution layer to quantify the migration tendency between components. For example, when the outlet pressure of the evaporator abnormally rises, the migration potential field will enhance the migration weight between the compressor and the condenser, reflecting the trend of refrigerant flowing faster due to the pressure difference. At the same time, the model introduces the statistical weight of the historical migration path as prior knowledge, such as the frequency of the path between certain components in historical data or the success rate of migration, and dynamically adjusts the feature transmission weight of the graph convolution layer through the attention mechanism. Specifically, the high migration tendency path output by the migration potential field is weighted and fused with the high frequency historical path, and if the directions are consistent, the migration coefficient will be significantly improved, otherwise the conflicting path will be suppressed through the confidence correction mechanism. Further, the Monte Carlo sampling method is used to generate multiple high-probability migration paths in this process, and the refrigerant migration coefficient is combined with the material corrosion resistance score of the component for multi-dimensional evaluation. For example, although a certain path has a high migration potential, when it passes through a high corrosion risk area, the model will reduce its confidence and trigger the migration inhibition weight adjustment. The time convolution layer further captures the hysteresis effect of refrigerant migration, such as the propagation delay of pressure fluctuations in the pipeline after the compressor starts and stops, and dynamically corrects the migration coefficient through historical time series data training. Finally, the migration coefficient is determined through the joint probability distribution of the potential field output and the statistical weight, and the coefficient value of the high-priority path reflects its comprehensive migration possibility, providing a quantitative basis for the refrigerant redistribution strategy.

[0085] Further optionally, the migration potential field output by the ST-GCN is combined with the refrigerant migration coefficient for joint verification, and the actual refrigerant state data collected by the sensor network is used to verify the prediction result. If the deviation between the predicted path and the actual detected path exceeds a threshold, a model parameter correction mechanism is triggered, and the optimized migration weight is migrated to similar equipment based on the transfer learning technology, thereby shortening the convergence time of refrigerant migration prediction for new equipment. It can be understood that in the verification and optimization process of the refrigerant migration prediction of the heating and ventilation equipment, the spatio-temporal graph convolution network (ST-GCN) continuously improves the model accuracy through multi-dimensional data fusion and dynamic feedback mechanism. The model jointly verifies the migration potential field and the refrigerant migration coefficient. For example, in a pool unit, the migration potential field reflects the refrigerant migration tendency from the compressor to the condenser, and the migration coefficient is calculated based on the real-time collected refrigerant pressure, temperature and flow rate data. When the difference between the two exceeds the preset threshold, the system triggers the parameter correction mechanism, adjusts the attention weight of the graph convolution layer through back propagation, and enhances the sensitivity to abnormal working conditions, such as increasing the weight of the evaporator surface dirt thermal resistance parameter to correct the migration path deviation. This dynamic verification and migration mechanism not only ensures the real-time accuracy of the prediction result, but also significantly reduces the deployment cost of new equipment through cross-device knowledge sharing, providing an intelligent solution for multi-unit collaborative operation.

[0086] In step S105, based on the durability parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, the refrigerant circulation loop in the heating and ventilation equipment is regulated by a fuzzy control algorithm to actively suppress the refrigerant migration phenomenon.

[0087] As an optional embodiment, in step 105, based on the durability parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, the refrigerant circulation loop in the heating and ventilation equipment is regulated by a fuzzy control algorithm, including: based on the durability parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, a fuzzy control algorithm is used to construct a refrigerant redistribution strategy; generating control instructions corresponding to the refrigerant redistribution strategy, and executing the control instructions to cooperatively control the opening degree of the valve group in the heating and ventilation equipment, so as to regulate the refrigerant circulation loop in the heating and ventilation equipment.

[0088] Specifically, in the regulation of HVAC refrigerant circulation loops, the refrigerant redistribution strategy is an active control scheme that guides the flow of refrigerant along a predetermined safe path to suppress abnormal migration by dynamically adjusting system operating parameters. Based on equipment durability parameters (such as compressor cumulative running time, pipeline corrosion level), refrigerant migration coefficients (reflecting the tendency of refrigerant flow between components), and candidate migration paths (a high-probability path set generated by Monte Carlo simulation), a multi-objective optimization model is established through a fuzzy control algorithm. For example, when it is detected that the evaporator surface corrosion-resistant coating is severely worn (durability parameter decreases) and the migration coefficient shows that liquid refrigerant is gathering in this area, the strategy will preferentially generate instructions to reduce the evaporator inlet valve opening degree, while increasing the condenser bypass valve opening degree, guiding the refrigerant to bypass the high-risk area.

[0089] The control instruction generation process corresponding to the refrigerant redistribution strategy involves multi-dimensional parameter fusion and dynamic weight distribution. Taking a pool unit as an example, the system first quantizes the durability parameter into a corrosion risk level (such as 0-5 levels), converts the migration coefficient into a path priority weight (such as priority weight coefficients set from 0.1 to 1.0), and associates the candidate paths with real-time working conditions such as environmental temperature and humidity, compressor load. The fuzzy control engine performs fuzzy reasoning through a pre-set rule base (such as "if the evaporator corrosion level > 3 and the migration coefficient > 0.8, then reduce the corresponding valve opening degree by 20%") to generate a composite instruction containing valve group adjustment amplitude, execution timing, and compensation coefficient. For example, to address the risk of liquid hammer in the suction line, the instruction may require the expansion valve opening degree to be gradually reduced from 70% to 50% within 30 seconds, while the dry filter bypass valve opening degree is increased to 40% to balance the pressure difference.

[0090] The valve opening degree in HVAC refers to the control parameter of the flow passage of the regulating valve in the refrigerant pipeline, usually expressed as a percentage (0% for full closure, 100% for full opening). Taking a commercial multi-split system as an example, the opening adjustment of key components such as four-way valves, electronic expansion valves, and electromagnetic valves directly affects the refrigerant phase change process and flow distribution. When the system detects an abnormal increase in compressor discharge temperature, it may reduce the condenser outlet valve opening degree (e.g., from 80% to 60%) to increase the supercooling degree of the refrigerant, while increasing the evaporator inlet valve opening degree to enhance the heat absorption efficiency, forming a closed-loop control of refrigerant redistribution.

[0091] In the above embodiments, the coordinated control of the control instructions on the valve group is achieved through a hierarchical decision mechanism. At the hardware level, the main control unit parses the strategy instructions into independent adjustment parameters for each valve group and synchronously issues them through an industrial bus (such as CAN or Modbus). At the logic level, a priority arbitration mechanism is used to ensure that critical path adjustment is executed first. For example, when the corrosion inhibition strategy conflicts with the energy optimization strategy, the corrosion risk avoidance instruction (such as immediately closing the bypass valve in the corrosion area) is executed first, and then the energy optimization adjustment is resumed after the risk is eliminated. This dynamic priority mechanism in the pool unit is as follows: during high temperature period, the refrigeration efficiency is prioritized (the valve opening degree corresponding to the increase of condenser fan speed is increased), and during low temperature period, the corrosion protection is focused (the local flow rate of evaporator is reduced).

[0092] The core of the control principle of the above embodiments is to establish a dynamic mapping relationship between the refrigerant migration risk and the valve adjustment. By monitoring the migration coefficient change in real time (such as updating the path probability distribution every 15 seconds), the valve group opening degree combination is continuously optimized, which suppresses abnormal migration while maintaining system energy efficiency. For example, when it is detected that the migration coefficient of a certain pipeline suddenly increases from 0.3 to 0.7, the control unit will start emergency adjustment within 200ms: reduce the upstream valve opening degree to reduce the refrigerant input, while increasing the downstream valve opening degree to speed up the refrigerant discharge, forming a pulse type adjustment waveform, effectively blocking the continuous accumulation of refrigerant in the high-risk area. This adaptive adjustment mechanism based on fuzzy logic can shorten the response time of refrigerant migration suppression compared with traditional PID control, while reducing the wear and tear of components caused by frequent adjustment.

[0093] Further optionally, in the above step, based on the durability parameter, the refrigerant migration coefficient of each key component, and the candidate migration path, a fuzzy control algorithm is used to construct a refrigerant redistribution strategy, including: normalizing the durability parameter, the refrigerant migration coefficient, and the confidence of the candidate migration path to construct the input variable set of the fuzzy control system, wherein the durability parameter reflects the remaining life risk level of the equipment, the refrigerant migration coefficient represents the flow tendency between components, and the confidence of the candidate migration path reflects the path prediction reliability. Based on historical operation data and corrosion kinetics model, a fuzzy rule base is established to map the durability level, migration risk level, and refrigerant redistribution strategy. When the durability parameter is below the threshold and the high-priority migration coefficient path proportion exceeds the preset proportion, the refrigerant redistribution strategy triggers the refrigerant diversion instruction. According to the pool environment parameters, the weight of the fuzzy rule in the refrigerant redistribution strategy is adjusted in real time. When the chlorine concentration increases, the migration inhibition weight of the evaporator fin area is increased, and the refrigerant path of the corrosion-resistant component is preferentially selected. The effectiveness of the refrigerant redistribution strategy is verified by the actual refrigerant distribution data fed back by the sensor network. If the actual migration path obtained based on the actual refrigerant distribution data deviates from the predicted refrigerant migration path of the refrigerant redistribution strategy by more than a set threshold, the fuzzy rule parameters in the refrigerant redistribution strategy are corrected to ensure that the refrigerant redistribution strategy dynamically matches the actual working conditions.

[0094] Specifically, in the construction of the refrigerant redistribution strategy of the heating and cooling equipment, the fuzzy control algorithm realizes precise regulation and control through multi-dimensional parameter fusion and dynamic rule adjustment. First, the durability parameter, the refrigerant migration coefficient, and the confidence of the candidate path are normalized to form the input variable set of the fuzzy controller. For example, the durability parameter is divided into three fuzzy sets of "low risk", "medium risk", and "high risk" according to the remaining life of the equipment, the refrigerant migration coefficient is divided into "low", "medium", and "high" levels according to the flow tendency between components, and the path confidence is quantified into "high confidence", "medium confidence", and "low confidence" intervals based on historical data matching. This normalization process unifies heterogeneous parameters to the standard input domain of the fuzzy control system, laying a foundation for subsequent rule reasoning.

[0095] Based on historical operation data and corrosion kinetics model, the system establishes a fuzzy rule base for the durability level, migration risk level, and refrigerant redistribution strategy. For example, when the durability parameter is in "high risk" and the high migration coefficient path proportion exceeds 70%, the "preferentially inhibit refrigerant flow into the corrosion area" diversion instruction is triggered. The rule base is generated by combining expert experience and data-driven methods. For example, historical data of a certain pool unit shows that the corrosion rate of the evaporator fin area increases by 3 times when the chlorine concentration exceeds 200 ppm, and the corresponding rule will increase the migration inhibition weight of this area. The fuzzy reasoning engine uses Mamdani or T-S type models to generate composite control instructions through weighted average or weighted minimum method, realizing the coordinated adjustment of the refrigerant valve group.

[0096] The dynamic adjustment mechanism of environmental parameters achieves strategy optimization by online evaluation of the impact of external conditions on migration risk. For example, when the chlorine concentration monitoring value of the swimming pool rises from 150 ppm to 250 ppm, the system automatically increases the migration inhibition weight coefficient of the evaporator fin area, making the path selection priority of corrosion-resistant components (such as titanium alloy pipelines) increase by 40%. This dynamic weight adjustment is achieved through environmental sensitive factors in fuzzy rules, for example, in high temperature and high humidity environment, when the condenser surface dew risk increases, the system will reduce the allowed threshold of refrigerant flow rate in this area.

[0097] The above strategy effectiveness verification and correction is completed through a closed-loop feedback mechanism. The sensor network collects real-time refrigerant pressure, temperature and flow rate data, and compares and analyzes with the predicted path. If the deviation between the actual path and the predicted path of a certain migration exceeds 15%, the system triggers the parameter correction process: first, calculate the confidence deviation of the rule through defuzzification, then adjust the membership function parameters or optimize the rule trigger threshold using gradient descent method. For example, after detecting that the actual corrosion rate at the bottom of the evaporator is 20% higher than the predicted value, the system shifts the peak value of the "corrosion risk level" fuzzy set membership function to the high risk area, and reduces the confidence weight of the low migration coefficient path, so that the subsequent strategy is more suitable for the actual working condition. This adaptive correction mechanism makes the refrigerant redistribution strategy remain dynamic optimization throughout the life cycle of the equipment, effectively balancing safety protection and energy efficiency goals.

[0098] By way of example, the fuzzy control algorithm can be a genetic algorithm optimized fuzzy controller. Through genetic algorithm global search quantization factor and membership function parameters, solve the problem of traditional fuzzy control rules relying on experience. For example, in the refrigerant redistribution strategy, the input variable weight can be optimized (such as shifting the membership function peak value of the chlorine concentration monitoring value to the high risk area), so that the strategy convergence time is shortened. Alternatively, an adaptive fuzzy controller. Introduce online learning mechanism to dynamically adjust the rule base to adapt to changes in migration characteristics caused by equipment aging. For example, when detecting that the evaporator corrosion-resistant coating is worn, the system automatically increases the trigger threshold of the migration inhibition rule in this area, while reducing the confidence weight of the low priority path. Alternatively, any one of a Mamdani type fuzzy controller, a Sugeno type fuzzy controller, and a T-S type fuzzy controller. Further optionally, the refrigerant migration control scenario in the present application has the characteristics of multi-source heterogeneous input, strong nonlinearity, time-varying constraint, etc., and therefore a hybrid architecture of T-S type fuzzy controller and genetic algorithm optimization can be used as the fuzzy control algorithm of the foregoing embodiments. Specifically, the T-S model is used to process dynamic processes such as refrigerant phase change and pipeline pressure loss, and the differential term is used to capture the migration hysteresis effect; genetic algorithm is used to optimize input variable weight and rule trigger threshold, solving the problem of rule invalidation caused by sparse historical data; combined with Monte Carlo simulation to generate migration path probability distribution, enhancing the forward-looking of the strategy.

[0099] Further optionally, in the above step, the control instructions corresponding to the refrigerant redistribution strategy are generated, and the control instructions are executed to cooperatively control the valve group opening degree in the heating and ventilation equipment, including: the high-priority migration path corresponding to the pipeline valve opening degree is increased to increase the refrigerant flow to guide the migration direction; the low-priority path valve opening degree is reduced to limit the refrigerant residence time; based on the real-time monitoring data of the refrigerant pressure sensor, a PID algorithm is used to dynamically balance the refrigerant flow rate of each pipeline to prevent system pressure fluctuations caused by sudden changes in valve group opening degree; the valve group adjustment effect is verified through the refrigerant state sensor, if the refrigerant migration path does not reach the expected target, a secondary correction instruction of the valve group opening degree is triggered, and the adjustment parameters are fed back to the fuzzy control algorithm for strategy iteration optimization.

[0100] In the above step, the innovation of the refrigerant redistribution strategy control instruction lies in the deep integration of multi-dimensional dynamic cooperative mechanism and self-adaptive optimization architecture. First, through the differential adjustment of the high-priority path valve opening degree increase and the low-priority path suppression, an active guiding mechanism of refrigerant flow is constructed, for example, in the high-risk area of evaporator corrosion in the pool unit, the valve opening degree of the corrosion-resistant pipeline is dynamically increased to 85%, while the valve opening degree of the corrosion area is reduced to 30%, forming a physical barrier for refrigerant migration. Second, the PID dynamic balancing algorithm based on the pressure sensor is introduced, when the valve group opening degree suddenly changes (such as from 50% to 70%), the pressure fluctuation is compensated through the differential term, the refrigerant flow fluctuation amplitude is controlled within the set range, and the risk of liquid hammer or overheating caused by system pressure imbalance is avoided.

[0101] The sensor network and the closed-loop verification mechanism constitute the core support of strategy optimization. Through the millisecond-level data collection of refrigerant pressure, temperature and flow rate sensors, the deviation between the predicted path and the actual migration trajectory is compared in real time, when the refrigerant aggregation deviation at the bottom of the evaporator is detected to be more than 15%, a secondary correction instruction is triggered, and the gradient adjustment of the valve group opening degree is completed within 200ms. This data-driven closed-loop verification mechanism shortens the response time of strategy correction compared with traditional methods. The self-adaptive iterative optimization architecture realizes strategy evolution through fuzzy control algorithm and migration learning technology.

[0102] In the embodiments of the present application, the refrigerant migration phenomenon of the compressor can be inhibited in the running state or the shutdown state, avoiding mechanical damage, energy efficiency ratio reduction, oil line blockage and other situations caused by refrigerant migration in the heating and ventilation equipment, and improving the stability of the equipment system. Especially, the embodiments of the present application can adapt to the high-frequency start-stop characteristics of the pool equipment in the high-humidity and high-chlorine environment of the pool, improve the durability of the pool heating and ventilation equipment, and prolong the service life of the compressor.

[0103] After introducing the method of the example embodiments of the present application, next, with reference to Figure 2A compressor refrigerant migration detection system based on a pool heating and ventilation device is described for the exemplary embodiments of the present application, which comprises: a collection module for deploying pressure sensors and / or temperature sensors in key components of the heating and ventilation device, establishing a heating and ventilation device monitoring network for real-time collection of refrigerant state parameters of each key component in the heating and ventilation device; the key components at least include: compressor, evaporator, suction pipe; an environmental monitoring module for real-time collection of pool environment parameters in the pool environment where the heating and ventilation device is located, the pool environment parameters at least include: temperature, humidity, air quality, chlorine concentration, ventilation condition, and input the pool environment parameters into a corrosion rate prediction model based on a corrosion kinetics equation to predict the durability parameters of the heating and ventilation device in the pool environment; a prediction module for establishing a refrigerant state migration prediction model of the heating and ventilation device according to the structure of the heating and ventilation device using a spatiotemporal graph convolution network; wherein a thermodynamic-hydrodynamic coupling model is used to infer the refrigerant migration relationship between each key component in the refrigerant state migration prediction model to simulate the refrigerant circulation loop between each key component; through the refrigerant state migration prediction model, according to the refrigerant state parameters of each key component, the refrigerant migration coefficient and the candidate migration path between each key component in the heating and ventilation device are predicted in real time; wherein the higher the refrigerant migration coefficient, the greater the probability of refrigerant migration to the corresponding component; a control module for based on the durability parameters, the refrigerant migration coefficient and the candidate migration path of each key component, the refrigerant circulation loop in the heating and ventilation device is regulated through a fuzzy control algorithm to actively suppress the refrigerant migration phenomenon. The above system can realize each step described in the above method embodiments, and the specific implementation of each step will not be repeated here.

[0104] After introducing the method and system of the exemplary embodiments of the present application, next, a terminal device of the exemplary embodiments of the present application is described, which can realize each step described in the above method embodiments, and the specific implementation of each step will not be repeated here.

[0105] After introducing the method, system and terminal device of the exemplary embodiments of the present application, next, with reference to Figure 3 The computer readable storage medium of the exemplary embodiments of the present application is described, please refer to Figure 3 The computer readable storage medium shown is an optical disc 30, which stores a computer program (i.e. program product) thereon, the computer program will realize each step described in the above method embodiments when running on a processor. The specific implementation of each step will not be repeated here.

[0106] The above-described embodiments are merely specific implementations of the present application, and are used to illustrate the technical solutions of the present application, rather than limit the same. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that any person skilled in the art can make modifications or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some of the technical features within the technical scope disclosed by the present application. The modifications, changes or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for detecting refrigerant migration in a compressor of a swimming pool HVAC system, characterized in that, The method includes: Deploy pressure sensors and / or temperature sensors in key components of HVAC equipment to establish an HVAC equipment monitoring network for real-time acquisition of refrigerant status parameters of each key component in HVAC equipment; key components include at least: compressor, evaporator, and suction pipe; Real-time collection of pool environment parameters in the pool environment where the HVAC equipment is located. The pool environment parameters include at least: temperature, humidity, air quality, chlorine concentration, and ventilation status. The pool environment parameters are then input into a corrosion rate prediction model based on corrosion kinetic equations to predict the durability parameters of the HVAC equipment in the pool environment. Based on the structure of HVAC equipment, a refrigerant state migration prediction model for HVAC equipment is established using a spatiotemporal graph convolutional network. Among them, a thermodynamic-fluid dynamic coupling model is used to infer the refrigerant migration relationship between various key components in the refrigerant state migration prediction model, in order to simulate the refrigerant circulation loop between various key components. The refrigerant state migration prediction model predicts the refrigerant migration coefficient and candidate migration paths between key components in HVAC equipment in real time based on the refrigerant state parameters of each key component. The higher the refrigerant migration coefficient, the greater the probability that the refrigerant will migrate to the corresponding component. Based on the aforementioned durability parameters, the refrigerant migration coefficients of each key component, and candidate migration paths, a fuzzy control algorithm is used to regulate the refrigerant circulation loop in the HVAC equipment to actively suppress refrigerant migration.

2. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 1, characterized in that, The step of inputting the pool environment parameters into a corrosion rate prediction model based on corrosion kinetics equations to predict the durability parameters of HVAC equipment in a pool environment includes: The swimming pool environmental parameters are normalized to eliminate the differences in the dimensions of different parameters, and the temperature, humidity, chlorine concentration, particulate matter concentration and ventilation rate in the swimming pool environmental parameters are mapped to a unified numerical range. Based on the electrochemical corrosion mechanism, a corrosion rate prediction layer for metal surface corrosion rate is established. The model parameters of the corrosion rate prediction layer are determined through experimental calibration and regression analysis of historical data. The model parameters of the corrosion rate prediction layer include material property parameters, environmental sensitivity factors, and corrosion inhibition coefficients. The normalized pool environment parameters are input into the corrosion rate prediction layer to predict the corrosion rate of the metal surface under the current pool environment. The cumulative corrosion depth of key components of the equipment is calculated by integrating the corrosion rate of the metal surface over time. Based on the ratio between the equipment's service life and the cumulative corrosion depth of its critical components, a durability parameter reflecting the equipment's remaining lifespan is generated.

3. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 2, characterized in that, After inputting the pool environment parameters into a corrosion rate prediction model based on corrosion kinetics equations to predict the durability parameters of HVAC equipment in a pool environment, the method further includes: By combining real-time corrosion data fed back from the sensor network, the frequency band signal of the corrosion-sensitive frequency band energy is extracted through wavelet packet transform to ensure that the model can adapt to any dynamic scenario, such as fluctuations in pool water quality and changes in disinfectant dosage. The Q-learning algorithm is used to dynamically optimize the mapping relationship between operating parameters and refrigerant control strategy based on frequency band signals. By comparing the deviation between actual corrosion monitoring data and model prediction values, the environmental sensitivity factor and inhibition coefficient in the corrosion rate prediction model are dynamically adjusted to update the model parameters of the corrosion rate prediction model in real time.

4. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 2, characterized in that, After inputting the pool environment parameters into a corrosion rate prediction model based on corrosion kinetics equations to predict the durability parameters of HVAC equipment in a pool environment, the method further includes: The operating parameters of the HVAC equipment are collected in real time through the equipment control system; the operating parameters include at least: equipment running time, operating mode, start-up and shutdown frequency, and compressor load fluctuation range. The correlation analysis between the operating parameters and durability parameters was performed to establish a multi-dimensional influencing factor matrix; among which, the influencing factors include at least: the correlation factor between start-up and shutdown frequency and corrosion rate of compressor components obtained through regression analysis of historical data, and the weighting factor of the influence of the proportion of cooling mode on the corrosion rate of compressor components. A dynamic weighting model is constructed by combining the multi-dimensional influence factor matrix to adjust the model parameter weights in the corrosion rate prediction model.

5. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 1, characterized in that, The step of establishing a refrigerant state transition prediction model for HVAC equipment using a spatiotemporal graph convolutional network based on the equipment structure includes: Based on the piping layout and component connection relationships of HVAC equipment, a topology structure of the refrigerant circulation network is constructed, and key components are abstracted as nodes in a graph network, and the piping connections between key components are abstracted as edges; key components include at least: compressor, evaporator, and suction pipe; Each node is assigned a physical attribute label, which includes component type, material properties, and operating status parameters. By integrating thermodynamic and fluid dynamic data, a dynamic coupled model for refrigerant migration is established. Specifically, the energy exchange rate at each node is calculated based on the phase change characteristics of the refrigerant, and the refrigerant velocity distribution and pressure drop characteristics are analyzed based on the Bernoulli equation and Reynolds number. The baseline path for refrigerant migration is generated through finite element simulation, serving as the initial migration pattern library for the graph network. The refrigerant migration process is modeled using a spatiotemporal graph convolutional network (ST-GCN). The refrigerant state parameters of adjacent nodes are aggregated through graph convolutional layers to generate the refrigerant migration potential energy field between components. The hysteresis effect of refrigerant migration is captured through temporal convolutional layers. An environmentally sensitive factor is introduced to dynamically adjust the migration weights, and the adsorption coefficient of refrigerant on the metal surface is corrected in real time according to the pool environment parameters. Based on historical corrosion data, migration inhibition weight parameters for high corrosion risk areas are dynamically allocated to construct a refrigerant state migration prediction model.

6. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 1, characterized in that, The refrigerant state migration prediction model, based on the refrigerant state parameters of each key component, predicts in real time the refrigerant migration coefficients and candidate migration paths between key components in the HVAC system, including: The refrigerant state migration prediction model based on ST-GCN uses a refrigerant migration potential energy field and combines the statistical weights of historical migration paths to calculate the refrigerant migration coefficient between key components. The migration potential energy field is generated by aggregating the refrigerant state parameters of adjacent nodes through a graph convolutional layer, reflecting the refrigerant migration tendency between components. By introducing the real-time migration rate as a correction factor, the hysteresis effect of refrigerant migration is captured through the temporal convolutional layer of the refrigerant state migration prediction model, and the matching degree between the refrigerant migration coefficient and the real-time operating conditions is dynamically adjusted. Potential migration paths are generated based on ST-GCN, and multi-dimensional scoring is performed on the potential migration paths by combining refrigerant migration coefficient, component material corrosion resistance and current environmental parameters. The path probability distribution of potential migration paths is generated by Monte Carlo simulation, and the confidence interval of each potential migration path is calculated. The confidence level is determined by the path probability density function and the matching degree of historical data, and the confidence weight is dynamically adjusted by introducing environmental sensitivity factors. Potential migration paths are sorted in descending order based on the upper limit of the confidence interval, and the potential migration path with the highest confidence is selected as the high-priority candidate migration path. When a high corrosion risk area is detected, the migration suppression weight of the current area is dynamically increased based on the historical corrosion data of ST-GCN, and candidate migration paths that bypass the high-risk area are selected first.

7. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 1, characterized in that, The method of regulating the refrigerant circulation loop in HVAC equipment using a fuzzy control algorithm based on the durability parameters, refrigerant migration coefficients of various key components, and candidate migration paths includes: Based on the aforementioned durability parameters, the refrigerant migration coefficients of each key component, and candidate migration paths, a refrigerant redistribution strategy is constructed using a fuzzy control algorithm. The control command corresponding to the refrigerant redistribution strategy is generated, and the control command is executed to coordinate the valve opening in the HVAC equipment to regulate the refrigerant circulation loop in the HVAC equipment.

8. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 7, characterized in that, Based on the durability parameters, the refrigerant migration coefficients of each key component, and candidate migration paths, a refrigerant redistribution strategy is constructed using a fuzzy control algorithm, including: The durability parameters, refrigerant migration coefficient, and confidence scores of candidate migration paths are normalized to construct a set of input variables for the fuzzy control system. The durability parameters reflect the risk level of the equipment's remaining lifespan, the refrigerant migration coefficient characterizes the refrigerant flow tendency between components, and the confidence scores of candidate migration paths reflect the reliability of path prediction. Based on historical operating data and corrosion kinetics model, a mapping relationship between durability level, migration risk level and refrigerant redistribution strategy is established using fuzzy rules; when the durability parameter is below the threshold and the proportion of high-priority migration coefficient paths exceeds the preset ratio, the refrigerant diversion command in the refrigerant redistribution strategy is triggered. The weights of fuzzy rules in the refrigerant redistribution strategy are adjusted in real time based on pool environmental parameters. When the chlorine concentration increases, the migration inhibition weight of the evaporator fin area is increased, and the refrigerant path of corrosion-resistant components is selected first. The effectiveness of the refrigerant redistribution strategy is verified by the actual refrigerant distribution data fed back by the sensor network. If the actual migration path obtained based on the actual refrigerant distribution data deviates from the refrigerant migration path predicted by the refrigerant redistribution strategy by more than a set threshold, the fuzzy rule parameters in the refrigerant redistribution strategy are corrected to ensure that the refrigerant redistribution strategy dynamically matches the actual operating conditions.

9. The method for detecting refrigerant migration in a compressor based on swimming pool HVAC equipment according to claim 7, characterized in that, The process of generating control commands corresponding to the refrigerant redistribution strategy and executing the control commands to coordinate the control of valve group openings in the HVAC equipment includes: High-priority migration paths correspond to increased valve opening in pipelines, increasing refrigerant flow to guide the migration direction; The valve opening of low-priority paths is reduced, limiting the refrigerant residence time; Based on real-time monitoring data from refrigerant pressure sensors, a PID algorithm is used to dynamically balance the refrigerant flow rate in each pipeline to prevent system pressure fluctuations caused by sudden changes in valve opening. The valve group adjustment effect is verified by the refrigerant status sensor. If the refrigerant migration path does not reach the expected target, a secondary correction command for the valve group opening is triggered, and the adjustment parameters are fed back to the fuzzy control algorithm for strategy iterative optimization.

10. A compressor refrigerant migration detection system based on swimming pool HVAC equipment, characterized in that, The system executes the compressor refrigerant migration detection method based on swimming pool HVAC equipment as described in any one of claims 1 to 9, the system comprising: The data acquisition module is used to deploy pressure sensors and / or temperature sensors in key components of HVAC equipment to establish an HVAC equipment monitoring network for real-time acquisition of refrigerant status parameters of each key component in HVAC equipment; key components include at least: compressor, evaporator, and suction pipe; An environmental monitoring module is used to collect swimming pool environmental parameters in real time in the swimming pool environment where the HVAC equipment is located. The swimming pool environmental parameters include at least: temperature, humidity, air quality, chlorine concentration, and ventilation status. The swimming pool environmental parameters are input into a corrosion rate prediction model based on corrosion kinetic equations to predict the durability parameters of the HVAC equipment in the swimming pool environment. The prediction module is used to establish a refrigerant state migration prediction model for HVAC equipment based on the equipment's structure using a spatiotemporal graph convolutional network. Specifically, a thermodynamic-hydrodynamic coupling model is used to infer the refrigerant migration relationships between key components in the prediction model, simulating the refrigerant circulation loops between these components. Based on the refrigerant state parameters of each key component, the prediction model predicts the refrigerant migration coefficients and candidate migration paths between key components in real time. A higher refrigerant migration coefficient indicates a greater probability of refrigerant migrating to the corresponding component. The control module is used to regulate the refrigerant circulation loop in the HVAC equipment based on the durability parameters, the refrigerant migration coefficients of each key component, and candidate migration paths, in order to actively suppress refrigerant migration.

Images