Defrosting control method and system for low-temperature variable-frequency air source heat pump and computer software

Abstract

The invention relates to the technical field of air conditioners, in particular to a defrosting control method and system for a low-temperature variable-frequency air source heat pump and computer software. According to the method, through multi-point temperature monitoring and external environment parameters, the working frequency of the compressor and switching of the four-way valve are dynamically adjusted by applying the prediction model, and the frost layer forming and eliminating process is accurately recognized. And high-frequency defrosting can quickly defrost, and after a frost layer is preliminarily removed, low-frequency maintenance is switched, so that energy consumption is reduced. And in the defrosting process, the energy recovery device is matched, waste heat is introduced into the energy storage system and used in subsequent heating, and the overall energy consumption is further reduced. According to the scheme, the defrosting time is effectively shortened, the temperature reduction range of the water tank is reduced, the energy-saving and heating requirements are met, and wide practical value and popularization prospects are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioners, and particularly to a defrosting control method, a system and computer software for a low-temperature variable-frequency air-source heat pump. Background Art

[0002] Low-temperature variable-frequency air-source heat pumps have broad application prospects in cold regions. They utilize the heat in the air to transfer thermal energy to indoor spaces or water tanks by means of compressor work, and have the advantages of energy conservation, environmental protection, and safety. Compared with air-source heat pumps in normal-temperature environments, the heat exchange efficiency of air-source heat pumps in low-temperature environments will decrease significantly. The main reason is that when the outside temperature is too low, the fin heat exchanger is prone to frosting or icing on the surface, resulting in a sharp decline in heat exchange performance, and thus affecting the overall heating effect.

[0003] Currently, the defrosting methods of low-temperature variable-frequency air-source heat pumps mostly adopt the following common technical solutions:

[0004] 1. Timed defrosting: Forcefully switch to the defrosting mode at a preset time interval, and defrost by means of compressor reverse rotation or four-way valve switching. Although it is easy to implement, there are problems of untimely defrosting or excessive defrosting, resulting in high energy consumption and unsatisfactory defrosting effects;

[0005] 2. Temperature-triggered defrosting: Monitor the temperature of the heat exchanger coil, and defrost when the set threshold is reached. This method can reduce unnecessary defrosting losses to a certain extent, but it lacks accurate control over the thickness and dynamic development trend of the frost layer, and cannot balance energy consumption and defrosting efficiency;

[0006] 3. Pressure / temperature difference detection defrosting: Detect the pressure at the inlet and outlet of the heat exchanger or the temperature difference to determine whether the heat exchanger is frosted. When the difference reaches the threshold, trigger defrosting. This method is more flexible than timed or single-temperature-triggered defrosting, but it cannot simultaneously consider the dynamic coupling of external environmental changes and system operating conditions, and there may still be energy consumption waste or incomplete defrosting.

[0007] Traditional defrosting control methods usually set parameters such as defrosting timing and defrosting frequency relatively fixed, and cannot be adjusted in real time according to the actual frost layer thickness and system operating conditions, resulting in frequent or excessive defrosting, increasing the compressor load and wasting energy. At the same time, since the heat absorption in low-temperature environments is limited, long-term or high-frequency reverse heating defrosting will cause insufficient heating of the water tank or the indoor side, reducing user comfort and increasing operating costs. Defrosting control is mostly based on single-point temperature detection or fixed-time triggering, and it is difficult to accurately judge the development trend of the frost layer. There may be a phenomenon that the defrosting mode is switched back to the heating mode before the frost layer is completely cleared, or the defrosting continues at a high frequency even though the frost layer has been cleared. And when external conditions such as temperature, humidity, and wind speed fluctuate greatly, fixed defrosting strategies cannot fully cope with the changing climate environment, and the overall operating efficiency of the system is limited.

[0008] With the continuous development of intelligent control technology, existing research has attempted to apply algorithms based on multi-parameter monitoring of temperature, pressure, and humidity to the defrost control of air-source heat pumps; some solutions also use machine learning models to predict the formation and elimination trends of frost layers to achieve more accurate defrost control. However, there is still much room for improvement in existing solutions in aspects such as real-time data processing, adaptive parameter optimization, and energy recovery and utilization. In addition, although multi-point temperature monitoring and big data-driven predictive control technologies have been explored at the theoretical level, no mature commercial solutions have been formed in practical engineering applications, and the industry generally lacks a comprehensive control system that takes into account efficient defrosting, energy conservation and consumption reduction, system stability, and user comfort. Summary of the Invention

[0009] In view of the above deficiencies, the present invention proposes a defrost control method and system for a low-temperature variable-frequency air-source heat pump based on multi-sensor monitoring, machine learning prediction, and energy recovery optimization. By adaptively adjusting the defrost frequency of the compressor and combining an energy recovery mechanism, the defrost efficiency is greatly improved, and the system energy consumption is reduced, providing a practical solution for achieving efficient, energy-saving, and intelligent defrosting in a low-temperature environment.

[0010] To achieve the above objectives, the present invention adopts the following technical solutions:

[0011] A defrost control method for a low-temperature variable-frequency air-source heat pump, comprising the following steps:

[0012] 1) Data acquisition: Real-time monitor the temperature changes at multiple key positions of the defrost coil of the fin heat exchanger through multiple temperature sensors, and collect external environmental parameters, including external temperature and humidity;

[0013] 2) Data preprocessing: Clean, synchronize, and normalize the temperature data and environmental parameters to improve data quality;

[0014] 3) Feature extraction: Extract the temperature change rate (ΔT / Δt), temperature distribution characteristics, average temperature, temperature standard deviation, and environmental impact characteristics from the preprocessed data;

[0015] 4) Application of the prediction model: Use a machine learning-based prediction model to analyze the feature data, predict the formation and elimination trends of the frost layer, and determine the defrosting requirements for a future period of time;

[0016] 5) Adjustment of the defrost frequency: Dynamically and adaptively adjust the operating frequency of the compressor according to the prediction results. When it is predicted that the frost layer will form rapidly, increase the compressor to high frequency F 1 operation; when it is predicted that the frost layer is approaching being removed, reduce the compressor to low frequency F 2 operation;

[0017] 6) Mode switching: When it is predicted that the frost layer is about to be completely removed, that is, the defrosting coil temperature reaches or exceeds the set temperature T 2 At this time, switch the four-way valve back to the heating state to end the defrosting process;

[0018] 7) Feedback and optimization: Compare the actual defrosting effect with the predicted result, and re-train the prediction model regularly based on the feedback data to optimize the defrosting control strategy and achieve closed-loop control.

[0019] Preferably, the high frequency F 1 and the low frequency F 2 Satisfy F 1 >F 2 And dynamically adjust the specific values of F 1 and F 2 by an intelligent algorithm to adapt to different environmental conditions; and / or, the set temperature T 1 and T 2 Are adaptively adjusted according to environmental conditions and continuously optimized through a machine learning model.

[0020] Preferably, the temperature sensors are installed at multiple key positions of the defrosting coil to monitor the temperature distribution changes in real time and improve the accuracy of temperature monitoring through data fusion technology.

[0021] Preferably, the prediction model adopts an adaptive frost layer prediction model based on a multi-layer long short-term memory network (LSTM) and an attention mechanism.

[0022] Preferably, the method further includes an energy recovery step. During the defrosting process, the waste heat generated by the compressor is recovered through an energy recovery device and used to preheat the intake air to further reduce the system energy consumption.

[0023] Furthermore, the present invention also discloses a defrosting system for a low-temperature variable-frequency air source heat pump. This system implements the method described in any one of claims 1-5 and includes:

[0024] a. Multiple temperature sensors for monitoring the temperatures at multiple key positions of the fin heat exchanger defrosting coil;

[0025] b. An environmental sensor for collecting external environmental parameters, including external temperature and humidity;

[0026] c. A controller connected to the multiple temperature sensors and the environmental sensor, configured to adaptively adjust the operating frequency of the compressor according to the temperature change rate and environmental parameters by using an intelligent algorithm;

[0027] d. A compressor connected to the controller for operating at a high frequency F 1 or a low frequency F2 Run;

[0028] e. A four-way valve, connected to the controller, for switching between the defrost mode and the heating mode;

[0029] f. A prediction model module, configured to analyze the temperature change rate and environmental parameters based on a machine learning algorithm, predict the formation and elimination trend of the frost layer, and generate control instructions for adjusting the defrost frequency of the compressor;

[0030] g. A feedback module, for comparing the actual defrost effect with the prediction result, and retraining the prediction model based on the feedback data to optimize the defrost control strategy.

[0031] Preferably, the controller is configured to increase the operating frequency of the compressor to the high frequency F 1 when predicting the rapid formation of the frost layer, and reduce the operating frequency to the low frequency F 2 when predicting that the frost layer is nearly removed; and / or, the controller is further configured to switch the four-way valve back to the heating state and end the defrost process when the defrost coil temperature reaches or exceeds the set temperature T 2 .

[0032] Preferably, the system further includes an energy recovery device, which is connected to the compressor and is used to recover and reuse the waste heat generated during the defrost process; the energy recovery device includes a heat exchanger and an energy storage system, which are used to recover the waste heat generated during the operation of the compressor and use it to preheat the intake air, thereby further reducing the overall energy consumption of the system.

[0033] Furthermore, the present invention also discloses a computer-readable storage medium, on which a computer program or instruction is stored, and when the computer program or instruction is executed by a processor, the method is implemented.

[0034] Furthermore, the present invention also discloses a computer program product, including a computer program or instruction, and when the computer program or instruction is executed by a processor, the method is implemented.

[0035] Due to the adoption of the above technical solutions, the low-temperature variable-frequency air-source heat pump defrost control method and system of the present invention, through the combination of multi-point temperature monitoring and external environmental parameters, and the use of a predictive control algorithm and an energy recovery device, can significantly improve the accuracy of defrosting and the overall energy efficiency of the system. The specific manifestations are as follows:

[0036] 1. Precise defrosting and energy consumption reduction: The present invention introduces a predictive control algorithm based on the temperature change rate and environmental parameters (such as external temperature, humidity), and dynamically and adaptively adjusts the high frequency F 1 or the low frequency F 2Compared with the traditional defrosting methods triggered by fixed time or single temperature, it can more accurately master the formation and elimination process of the frost layer, avoiding the energy consumption waste caused by frequent or excessive defrosting. At the same time, the setting of the energy recovery device can recover the waste heat during the defrosting process and use it to preheat the intake air or the water tank, thereby further reducing the overall energy consumption of the system and improving the energy utilization efficiency.

[0037] 2. Shorten the defrosting time and improve the system efficiency: Since it can predict the frost layer trend in real time and switch the defrosting frequency in a timely manner, when the frost layer has been basically removed, it can quickly switch back to the heating mode without maintaining a high-frequency defrosting for a long time, reducing the occupation of the defrosting cycle on the system heating. After the defrosting cycle is significantly shortened, the heat pump can return to the normal heating mode more quickly, improving the availability of hot water or heating, ensuring the heating demand at the user end, and enhancing the overall operating efficiency and comfort of the system.

[0038] 3. Improve the system stability and reliability: The comprehensive monitoring of the multi-point temperature sensors and environmental sensors enables the controller to accurately obtain the temperature changes in each area of the heat exchanger and the dynamic changes of the external climate conditions. The predictive control algorithm can start the corresponding defrosting strategy in advance before detecting the frost layer growth trend; combined with the feedback mechanism to adaptively optimize the control parameters, ensuring the real-time and efficient operation of the entire system and avoiding system fluctuations and abnormal shutdowns caused by untimely defrosting or excessive defrosting.

[0039] In summary, the present invention realizes high-precision monitoring and control of the defrosting process in a low-temperature environment, effectively reduces energy consumption, shortens the defrosting duration, improves the heating efficiency, and ensures the stability and reliability of the system operation, significantly enhancing the user experience and economic benefits, and having obvious technological progress and innovation value. Brief Description of the Drawings

[0040] Figure 1 It is a system block diagram of the present invention. Detailed Embodiments

[0041] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0042] I. Overall System Structure and Hardware Configuration

[0043] As Figure 1 shown, a defrosting system for a low-temperature variable-frequency air-source heat pump includes:

[0044] a. Multiple temperature sensors for monitoring the temperatures at multiple key positions of the fin heat exchanger defrosting coil;

[0045] b. An environmental sensor for collecting external environmental parameters, including external temperature and humidity;

[0046] c. A controller connected to the multiple temperature sensors and the environmental sensor, configured to adaptively adjust the operating frequency of the compressor according to the temperature change rate and environmental parameters using an intelligent algorithm;

[0047] d. A compressor connected to the controller for operating at a high frequency F 1 or a low frequency F 2 ;

[0048] e. A four-way valve connected to the controller for switching between the defrost mode and the heating mode;

[0049] f. An energy recovery device connected to the compressor for recovering and reusing the waste heat generated during the defrosting process;

[0050] g. A prediction model module configured to analyze the temperature change rate and environmental parameters based on a machine learning algorithm, predict the formation and elimination trends of the frost layer, and generate control instructions for adjusting the defrost frequency of the compressor;

[0051] h. A feedback module for comparing the actual defrost effect with the prediction result and retraining the prediction model based on the feedback data to optimize the defrost control strategy.

[0052] 1. Fin heat exchanger

[0053] Adopt a combination of aluminum foil fins and copper tubes or other anti-corrosion and low-temperature-resistant materials suitable for low-temperature environments. The fin surface can be treated with a hydrophilic / hydrophobic coating to further reduce frosting or facilitate defrosting (selected according to specific requirements).

[0054] 2. Multiple temperature sensors

[0055] Type selection: Common types such as NTC thermistors, PT100, PT1000, or semiconductor digital temperature sensors (such as DS18B20) can be used, and the accuracy is preferably selected as ±0.5°C or higher.

[0056] Installation location:

[0057] On the inlet side of the heat exchanger, mainly for monitoring the external environmental temperature and the initial frosting condition. In the middle section of the heat exchanger, at least one temperature sensor is configured at the corresponding position of each heat exchange tube for collecting the temperature in the concentrated frost growth area. On the outlet side of the heat exchanger, for monitoring the exchange effect and making a difference judgment with the inlet temperature.

[0058] Sampling frequency: It is recommended to be between 1Hz and 5Hz, which can not only meet the real-time control requirements but also avoid excessive data redundancy.

[0059] 3. Environmental sensors

[0060] Ambient temperature: The same as that of NTC / PT100 / PT1000 / Digital Sensor, etc., can be installed outdoors with a windproof and waterproof protective shell.

[0061] Humidity sensor: Such as SHT series, DHT series or high-precision capacitive humidity sensors, and the installation location should avoid direct blowing and the influence of rain and snow.

[0062] Wind speed sensor (optional): Used for accurate judgment under extreme conditions. If the wind speed is too high, the frosting speed will change rapidly.

[0063] 4. Compressor and inverter drive circuit

[0064] Frequency conversion controller: It includes IGBT or MOSFET inverter modules and drive circuits, DC bus capacitors, etc., and can be selected according to the compressor power specifications.

[0065] Compressor: Select the liquid injection enhanced enthalpy (EVI) or two-stage compression technology suitable for low-temperature working conditions to improve the heating performance at ultra-low temperatures.

[0066] Frequency range: Typically between 30Hz and 120Hz, and the high frequency F 1 (such as 80Hz - 90Hz) and low frequency F 2 (such as 40Hz - 50Hz) should be set according to the actual model.

[0067] 5. Four-way valve

[0068] Adopt a common refrigerant cycle switching valve, which can complete the switching from heating to defrosting (thawing) mode within a few seconds. The inlet / outlet positions should be marked on the valve body to ensure correct wiring connection.

[0069] 6. Controller

[0070] Main control CPU / MCU: The ARM Cortex-M series, STM32, ESP32, or higher-order embedded processors (such as NXP i.MX RT, etc.) can be used to meet the real-time operation requirements.

[0071] Memory and data storage: It is recommended to reserve enough Flash / EEPROM for storing programs and model parameters, and the RAM capacity should meet the real-time algorithm requirements (at least dozens of KB to hundreds of KB).

[0072] Communication interfaces: I²C / SPI / USART, etc., for connecting temperature sensors and environmental sensors; at the same time, RS485, CAN or Ethernet interfaces are reserved for communication with the host computer or the cloud.

[0073] Analog interfaces: Monitor the operating current, voltage, etc. of the compressor, which can provide a basis for energy consumption assessment and protection strategies.

[0074] Prediction model module: Can be integrated in the form of firmware or dynamic libraries within the embedded operating system.

[0075] Feedback optimization module: Real-time collect defrosting effects and energy consumption indicators, and update model parameters during the idle period of the system.

[0076] 7. Energy recovery device

[0077] Heat exchanger: Can adopt a concentric tube type, plate type or shell and tube type structure, depending on the installation space and efficiency requirements.

[0078] Energy storage system: If the water tank is used as the energy storage medium, it is necessary to ensure that the water tank has a good thermal insulation layer (such as polyurethane foam), and is equipped with a water pump and a heat exchange coil for introducing the waste heat into the water tank during the high-frequency defrosting stage.

[0079] Automatic valve or three-way valve: Control the flow direction of the waste heat in different modes, and cooperate with the defrosting process to maintain a good heat exchange effect.

[0080] II. Software and control flow

[0081] 1. Data acquisition and synchronous processing

[0082] 1) Acquisition scheduling

[0083] The controller periodically triggers the sampling of multiple temperature sensors and environmental sensors, and the recommended frequency is 1-5 times per second.

[0084] Key data includes:

[0085] Temperatures T at each point i (i = 1, 2,..., n);

[0086] Ambient temperature T out ;

[0087] Humidity H out ;

[0088] Wind speed W out (optional);

[0089] Either parallel acquisition or polling acquisition method can be used, configured according to the characteristics of the sensors and the bus.

[0090] 2) Data cleaning and filtering

[0091] Perform validity checks (such as out-of-bounds, no response, etc.) on each piece of collected data and perform average filtering or Kalman filtering.

[0092] If a sensor data is lost or significantly abnormal, trigger an alarm or use interpolation of adjacent sensors for replacement.

[0093] 3) Timestamp alignment

[0094] Package the samples of the same batch into a data packet with a system timestamp for input to the subsequent prediction model and retrospective analysis.

[0095] 2. Feature extraction and model input

[0096] 1) Temperature gradient and rate

[0097] Calculate the temperature change rate ΔT / Δt and the temperature differences at each key position. The formula is as follows:

[0098] ΔT i = T i (t) - T i (t - Δt);

[0099] Combine the ambient temperature, humidity, etc. to form a set of feature vectors:

[0100] X t = [T 1 , T 2 ,..., T n , ΔT 1 , ΔT 2 ,..., ΔT n , T out , H out , W out

[0101] 2) Data normalization / standardization

[0102] Use μ, σ standardization or Min - Max normalization method to process X so that the input data is mapped to the range of [0, 1] or the standard normal distribution.

[0103] 3) Feature buffer

[0104] Perform sliding window processing on the data of the previous few seconds or dozens of seconds (window length 5 - 10 seconds) to obtain more time - series related features and improve the prediction accuracy.

[0105] 3. Predictive algorithm and control decision

[0106] 1) Prediction model selection

[0107] ​An adaptive frost layer prediction model based on a multi-layer long short-term memory network (LSTM) and an attention mechanism can be used to dynamically adjust the defrosting strategy of a low-temperature variable-frequency air source heat pump.

[0108] Model function: Predict the frost layer state after the next k seconds or minutes and output a defrosting demand intensity score (or directly output the decision of "high-frequency defrosting / low-frequency defrosting / no defrosting").

[0109] Model structure:

[0110] Input features: X t ;

[0111] One-dimensional convolutional feature extraction layer (Conv1D): 32 convolutional kernels, kernel size of 3, activation function ReLU;

[0112] Multi-layer LSTM layer: 2 layers, 128 hidden state units in each layer, Dropout = 0.2;

[0113] Attention mechanism layer: Weighted sum of the LSTM output features to strengthen key time steps;

[0114] Fully connected output layer (Dense): 32 units, activation function ReLU, 1 output layer unit (linear activation).

[0115] Specific formula of the attention mechanism:

[0116]

[0117] where h t is the output of the LSTM at time step t, W a , b a are trainable parameters, and α t is the attention weight.

[0118] Model training and online update method:

[0119] Loss function:

[0120] Regression prediction (such as frost layer growth rate): Use the mean squared error (MSE) loss function.

[0121]

[0122] Classification prediction (defrosting demand probability): Use binary cross entropy:

[0123]

[0124] Optimization algorithm:

[0125] Use the Adam optimizer with an initial learning rate of 0.001 to 0.0005.

[0126] Training batch size: 32 to 64.

[0127] Number of training epochs: 50 to 100. Use Early Stopping to prevent overfitting.

[0128] Validation strategy:

[0129] Divide the dataset into a training set (70%), a validation set (15%), and a test set (15%).

[0130] Use the validation set to select the optimal model and evaluate the performance on the test set.

[0131] Online update: After each defrost cycle, record the actual defrost effect and compare it with the prediction. Use incremental learning to fine-tune the model parameters.

[0132] 2) Model operation process

[0133] Real-time input: Input the feature vector Xt into the prediction model.

[0134] Output result: The model may output a classification label (such as 0, 10, 10, 1 corresponding to low frequency or high frequency) or a regression value, which is then thresholded by the controller (6).

[0135] Threshold determination: If the output > θ high , then enter the high-frequency defrost mode; if the output is near θ low , it can enter the low-frequency defrost or maintain heating.

[0136] 3) Control logic

[0137] High-frequency F 1 Trigger: When the prediction result shows that the frost layer grows rapidly in a short time (such as the temperature drop rate is greater than the set threshold or the estimated frost layer thickness > a certain threshold), the controller executes:

[0138] ① Switch the four-way valve to the defrost circuit;

[0139] ② Adjust the frequency of the compressor (4) to F1 (for example, 80 Hz to 90 Hz).

[0140] Low-frequency F 2 Maintain: If the model detects that the frost layer is gradually subsiding and the temperature change rate slows down, the frequency of the compressor (4) is reduced to F2 (40 Hz to 50 Hz).

[0141] Defrost Exit: When the coil temperature is monitored to reach or exceed the set temperature T2 (12°C) and remains stable without rising for a certain period (3 - 5 seconds), the system determines that the frost layer has been basically removed. Subsequently:

[0142] Switch the four-way valve back to the heating mode;

[0143] Set the compressor (4) to return to the frequency required for heating (in the conventional range of 30Hz - 60Hz).

[0144] 4) Adaptive adjustment

[0145] Dynamic threshold: θ high ,θ low ,T 1 ,T 2 Key thresholds such as θ, T, etc. can be adjusted according to changes in the external temperature / humidity. For example, the upper limit of the high frequency can be moderately increased in extremely cold weather, or the frequency can be decreased in warm weather.

[0146] Machine learning online update: After each defrost cycle, the system records the actual defrost effect (including indicators such as defrost time, energy consumption, and water tank temperature drop) and compares it with the predicted results to adjust the model parameters.

[0147] 4. Energy recovery mechanism

[0148] 1) High-frequency waste heat recovery during defrost

[0149] When the compressor operates at high frequency, the temperature / pressure on the refrigerant side is relatively high, and the excess heat can be introduced into the water tank or the air preheating chamber through the energy recovery device (7).

[0150] The three-way valve / solenoid valve can switch the circuit under the command of the controller (6), giving priority to flowing the waste heat into the heat exchanger and then storing it through the energy storage system.

[0151] 2) Secondary utilization

[0152] In the water tank, the temperature sensor can monitor the water temperature. When the water temperature reaches the target value, the valve automatically closes or switches to avoid overheating.

[0153] If there is a need for air preheating (such as preheating the ventilation duct), the recovered waste heat is directly transported to the air duct to increase the intake air temperature.

[0154] 3) Energy consumption reduction

[0155] By reasonably scheduling the waste heat recovery path, the impact on the water tank or indoor heat dissipation during defrost can be effectively reduced, realizing the secondary utilization of primary energy, and the overall energy consumption can be reduced by 10% - 15% (depending on the system configuration and the measured environment).

[0156] III. Example illustration

[0157] Example 1: -15°C Low Temperature Environment Test

[0158] 1. Test Environment

[0159] Simulate the outdoor temperature to be maintained at -15°C, relative humidity about 70%, and wind speed 2 m / s.

[0160] Set the output hot water temperature of the heat pump to 40°C.

[0161] 2. System Settings

[0162] High frequency F 1 = 80 Hz; Low frequency F 2 = 45 Hz; Set the defrost exit temperature T 2 = 12°C;

[0163] The prediction model uses random forest and is trained offline with 30 groups of low temperature working condition data.

[0164] 3. Test Process

[0165] After about 1 hour of continuous heating operation starts, frost begins to form on the surface of the heat exchanger. The sensor detects an accelerated frosting rate, and the prediction model outputs a "high frequency defrost" command; the controller commands the four-way valve to switch to the defrost mode, the compressor frequency is increased to 80 Hz, and the energy recovery (7) is synchronously turned on to inject the waste heat into the water tank.

[0166] After about 3 minutes, the heat exchanger temperature rises to 8 - 9°C, the sensor feedback rate slows down, and the prediction model further outputs a "frequency reducible defrost" command; the controller reduces the compressor to 45 Hz to maintain low-load defrosting.

[0167] When the temperature of the heat exchanger coil stabilizes and rises to 12°C and the frost layer is significantly thinner, the controller switches the four-way valve back to the heating mode and allows the compressor to heat in the 60 Hz range.

[0168] The entire defrosting process takes about 4 minutes and 20 seconds, which is about 25% shorter than the traditional timing method, and the water temperature in the water tank only drops by about 1.5°C.

[0169] 4. Energy Efficiency Results

[0170] Compared with the defrosting method without energy recovery, this test saves about 12% of the electricity; the hot water usage temperature at the user end changes smoothly, meeting the comfort requirements.

[0171] Example 2: -20°C Extreme Cold Environment Test

[0172] 1. Test Environment

[0173] Set the outdoor environmental temperature to -20°C, relative humidity 50%, and wind speed 4 m / s.

[0174] The day-night temperature difference during the test period fluctuates within ±2°C, continuously testing the adaptability of the system.

[0175] 2. System Adaptive Strategy

[0176] Initial setting: T 2 = 12°C, high frequency F 1 = 85Hz, low frequency F 2 = 50Hz. As the outdoor temperature further drops to -22°C, the prediction model finds that the frost layer growth rate increases. The system automatically raises the high-frequency upper limit to 88Hz and raises T 2 correspondingly to 14°C to ensure complete defrosting.

[0177] 3. Result Comparison

[0178] The traditional "temperature threshold + fixed high frequency" defrosting method defrosts 6 times during 4 hours of operation, with a total defrosting duration of 40 minutes. The adaptive algorithm of the present invention only defrosts 5 times, with a total defrosting duration of 29 minutes, saving about 15% of the defrosting energy consumption cumulatively. Even under such low temperature and strong wind speed conditions, the water tank temperature (in the range of 35°C to 40°C) can always be maintained at a relatively stable level.

[0179] The above is the description of the embodiments of the present invention. Through the above description of the disclosed embodiments, those skilled in the art can implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel points disclosed herein.

Claims

1. A defrosting control method for a low-temperature variable frequency air source heat pump, characterized in that: The following steps are involved: 1) Data collection: Multiple temperature sensors are used to monitor the temperature changes at multiple key locations of the fin heat exchanger defrost coil in real time, and collect external environmental parameters, including external temperature and humidity; 2) Data preprocessing: cleaning, synchronizing and normalizing the temperature data and environmental parameters to improve data quality; 3) Feature extraction: extract the temperature change rate (ΔT / Δt), temperature distribution characteristics, average temperature, temperature standard deviation and environmental impact characteristics from the preprocessed data; 4) Application of prediction model: using a prediction model based on machine learning to analyze the characteristic data, predict the formation and elimination trend of the frost layer, and determine the defrosting demand in the future; 5) Defrosting frequency adjustment: According to the prediction results, the operating frequency of the compressor is dynamically and adaptively adjusted. When it is predicted that the frost layer will form quickly, the compressor is increased to operate at a high frequency F1; when it is predicted that the frost layer is close to being completely defrosted, the compressor is reduced to operate at a low frequency F2; 6) Mode switching: When it is predicted that the frost layer will be completely removed, that is, when the defrost coil temperature reaches or exceeds the set temperature T2, the four-way valve is switched back to the heating state, ending the defrosting process; 7) Feedback and optimization: Compare the actual defrost effect with the predicted result, regularly retrain the prediction model based on the feedback data, optimize the defrost control strategy, and achieve closed-loop control.

2. The defrost control method according to claim 1, characterized in that: The high frequency F1 and the low frequency F2 satisfy F1>F2, and the specific values ​​of F1 and F2 are dynamically adjusted through an intelligent algorithm to adapt to different environmental conditions; and / or, the set temperatures T1 and T2 are adaptively adjusted according to environmental conditions, and the set values ​​are continuously optimized through a machine learning model.

3. The defrosting control method according to claim 1, characterized in that: The temperature sensors are installed at multiple key locations of the defrost coil to monitor temperature distribution changes in real time and improve the accuracy of temperature monitoring through data fusion technology.

4. The defrosting control method according to claim 1, characterized in that: The prediction model adopts an adaptive frost layer prediction model based on a multi-layer long short-term memory network (LSTM) and an attention mechanism.

5. The defrosting control method according to claim 1, characterized in that: The method also includes an energy recovery step. During the defrosting process, waste heat generated by the compressor is recovered through an energy recovery device and used to preheat intake air to further reduce system energy consumption.

6. A defrosting system for a low-temperature variable frequency air source heat pump, characterized in that: The system implements the method described in any one of claims 1 to 5, including: a. Multiple temperature sensors for monitoring the temperature of multiple key locations of the fin heat exchanger defrost coil; b. Environmental sensors, used to collect external environmental parameters, including external temperature and humidity; c. a controller connected to the plurality of temperature sensors and environmental sensors, configured to adaptively adjust the operating frequency of the compressor using an intelligent algorithm according to the temperature change rate and environmental parameters; d. a compressor connected to the controller for operating at a high frequency F1 or a low frequency F2 according to an instruction of the controller; e. a four-way valve connected to the controller for switching between a defrost mode and a heating mode; f. A prediction model module, configured to analyze the temperature change rate and environmental parameters based on a machine learning algorithm, predict the formation and elimination trend of the frost layer, and generate a control instruction for adjusting the defrosting frequency of the compressor; g. Feedback module, which is used to compare the actual defrosting effect with the predicted result, and retrain the prediction model based on the feedback data to optimize the defrosting control strategy.

7. The defrosting system according to claim 7, characterized in that: The controller is configured to increase the operating frequency of the compressor to a high frequency F1 when it is predicted that the frost layer will form rapidly, and to reduce the operating frequency to a low frequency F2 when it is predicted that the frost layer is close to being completely removed; and / or, the controller is further configured to switch the four-way valve back to the heating state to end the defrost process when the defrost coil temperature reaches or exceeds the set temperature T2.

8. The defrosting system according to claim 7, characterized in that: The system also includes an energy recovery device, which is connected to the compressor and is used to recover and reuse the waste heat generated during the defrosting process; the energy recovery device includes a heat exchanger and an energy storage system, which is used to recover the waste heat generated during the operation of the compressor and use it to preheat the intake air, thereby further reducing the overall energy consumption of the system.

9. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images