Defrosting control method and system of energy storage heat pump air conditioning system

Abstract

The invention discloses a defrosting control method and system for an energy storage heat pump air conditioning system, belongs to the technical field of defrosting control, and aims to solve the problems that a traditional defrosting method is low in energy efficiency, large in energy consumption and large in influence on system loads. The method comprises the following steps: acquiring an outdoor heat exchanger surface image in real time through an image acquisition device, processing image data to calculate a frost coverage rate, and inputting the image into a pre-trained neural network model to perform frosting grade classification when a system is in a heating mode and meets continuous operation time and environment temperature conditions, so as to obtain a frosting grade classification result; and triggering a weak frost mode or an enhanced frost mode based on the classification result, and continuously monitoring the frost coverage rate and the surface temperature uniformity in the defrosting process to control the exit opportunity. The defrosting precision is improved, invalid operation is reduced, energy consumption is reduced, and system stability and economical efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of defrosting control technology, and particularly relates to a defrosting control method and system for an energy storage heat pump air conditioning system. Background Technology

[0002] Against the backdrop of dual-carbon goals and the development of energy storage technologies, carbon dioxide heat pump air conditioning systems with energy storage have become an important development direction due to their environmental protection and energy-saving characteristics. This system relies on the phase change cycle of carbon dioxide refrigerant for heating, offering advantages in low energy consumption and cost. However, in actual operating conditions of low temperature and high humidity, the surface temperature of the outdoor heat exchanger easily falls below the air dew point and begins to frost. The frost layer gradually blocks airflow channels and heat transfer surfaces, leading to a reduction in effective heat transfer area and airflow, which in turn causes a decrease in system heating capacity, an increase in compressor pressure ratio, and an increase in exhaust temperature, ultimately severely damaging the overall system performance and operational reliability. Therefore, efficient defrosting of the outdoor heat exchanger is a key challenge to ensure the stable and efficient operation of carbon dioxide heat pump systems under a wide range of climatic conditions.

[0003] To improve system efficiency, existing technologies often employ reverse circulation defrosting or electric heating defrosting. While these methods can remove frost, they have significant drawbacks in practical applications. Reverse circulation defrosting requires the system to switch from heating to cooling mode, a process that not only interrupts indoor heating and affects comfort but also causes significant fluctuations in system pressure and flow, impacting critical components. Electric heating defrosting directly consumes high-grade electrical energy, increasing system energy consumption and contradicting the initial goal of energy conservation. More fundamentally, traditional defrosting control strategies rely on fixed time periods or limited temperature parameters for judgment, failing to accurately perceive the true thickness and distribution of frost on the evaporator surface in real time. This indirect criterion-based control method easily leads to two types of problems in complex operating environments: first, initiating defrosting when there is no frost or only light frost, resulting in energy waste and ineffective operation; second, waiting until the frost layer is too thick and system performance has severely degraded before defrosting, leading to prolonged defrosting time, increased energy consumption, and exacerbated system performance fluctuations and equipment wear. Therefore, developing a control method that can accurately sense the frost layer status, make intelligent decisions, and execute efficient defrosting is of great significance for improving the practical application value and market competitiveness of energy storage heat pump air conditioning systems. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a defrosting control method and system for an energy storage heat pump air conditioning system, which improves defrosting accuracy, reduces ineffective operations, lowers energy consumption, and enhances system stability and economy.

[0005] To achieve the above objectives, this invention provides a defrosting control method for an energy storage heat pump air conditioning system, comprising:

[0006] S1. Real-time image data of the outdoor heat exchanger surface is acquired using an image acquisition device;

[0007] S2. Process the image data to calculate the frost coverage rate;

[0008] S3. When the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, the image data is input into the pre-trained neural network model for frost level classification to obtain the frost level classification result.

[0009] S4. Based on the frost level classification result, trigger the corresponding defrosting mode, wherein the defrosting mode includes a weakened frost mode and a strengthened frost mode;

[0010] S5. During the defrosting mode, continuously monitor the frost coverage rate. When the frost coverage rate is lower than the fourth preset threshold, control the system to exit the defrosting mode and resume the heating mode.

[0011] Optionally, in S1, acquiring image data of the outdoor heat exchanger surface includes: simultaneously acquiring image data of the outdoor heat exchanger surface using an infrared thermal imager and a visible light camera; the infrared thermal imager operates at a sampling frequency not lower than a preset frequency; the image acquisition device is positioned at a preset distance from the outdoor heat exchanger; the image data includes infrared images and visible light images, and is used for subsequent processing.

[0012] Optionally, in S2, processing the image data to calculate the frost coverage rate includes: converting the image data into a grayscale image; performing binarization processing on the grayscale image using the Otsu threshold segmentation algorithm to distinguish between the frost layer region and the background region; calculating the ratio of the number of frost layer pixels to the total number of pixels in the binarized image, and using this ratio as the frost coverage rate; wherein the Otsu algorithm automatically determines the segmentation threshold.

[0013] Optionally, in S3, the first preset condition is that the continuous running time of the system exceeds a preset time threshold; the second preset condition is that the surface temperature uniformity of the outdoor heat exchanger is lower than a preset temperature threshold, and the surface temperature distribution of the outdoor heat exchanger is monitored by a temperature sensor; the third preset condition is that the ambient temperature is within a preset temperature range; the neural network model is a convolutional neural network based on residual learning and has an embedded attention mechanism; the neural network is trained with image data and is used to classify the frosting level.

[0014] Optionally, in S4, triggering the corresponding defrosting mode includes: triggering a weakened defrosting mode when the frost level is classified as the first level; and triggering a strengthened defrosting mode when the frost level is classified as the second level. The weakened defrosting mode and the strengthened defrosting mode are achieved by controlling the opening and closing of multiple valves to change the refrigerant flow path. The valve control is based on the controller output command.

[0015] Optionally, in S5, continuous monitoring includes: during the execution of defrosting mode, continuing to acquire image data and calculate frost coverage rate through the image acquisition device; the fourth preset threshold is a preset frost coverage rate threshold; when the frost coverage rate is higher than the fourth preset threshold, the system switches back to heating mode.

[0016] Optionally, the method further includes, before S4, calculating the system's heating capacity and coefficient of performance (COP); the heating capacity is calculated based on the compressor power and the outdoor heat exchanger's heat exchange capacity; the COP is calculated based on the heating capacity and the compressor power; the calculation process uses a preset formula.

[0017] On the other hand, to achieve the above objectives, the present invention provides a defrosting control system for an energy storage heat pump air conditioning system, comprising:

[0018] The image acquisition module is used to acquire image data of the outdoor heat exchanger surface in real time through an image acquisition device;

[0019] The frost coverage calculation module is used to process the image data to calculate the frost coverage.

[0020] The intelligent classification module is used to input the image data into a pre-trained neural network model to classify the frost level when the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, so as to obtain the frost level classification result.

[0021] The defrost mode triggering module is used to trigger the corresponding defrost mode based on the frost level classification result, wherein the defrost mode includes a weakened frost mode and a strengthened frost mode.

[0022] The exit control module is used to continuously monitor the frost coverage rate during the execution of the defrost mode. When the frost coverage rate is lower than the fourth preset threshold, the module controls the system to exit the defrost mode and resume the heating mode.

[0023] Technical advantages of this invention: This invention discloses a defrosting control method and system for an energy storage heat pump air conditioning system. By combining image recognition and intelligent control, it achieves accurate perception of the frost status of the outdoor heat exchanger and on-demand defrosting. Image processing technology is used to calculate the frost coverage rate, and a neural network is used to intelligently classify the frost level, effectively avoiding the ineffective operation and energy waste of traditional defrosting methods. By triggering differentiated defrosting modes based on the frost level, the system can minimize interference with the system's heating operation while ensuring defrosting effectiveness. This system effectively improves the accuracy and timeliness of defrosting control, significantly reduces system energy consumption, enhances operational stability and reliability, and reduces equipment maintenance requirements. Attached Figure Description

[0024] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0025] Figure 1 The following are experimental measurement photos and structural schematic diagrams of the outdoor heat exchanger, as well as the actual product and structure of the embodiment of the present invention.

[0026] Figure 2 This is a schematic diagram of the heat pump mode in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram illustrating the principle of the enhanced frost mode in an embodiment of the present invention.

[0028] Figure 4 This is a schematic diagram illustrating the principle of the weakened frosting mode in an embodiment of the present invention;

[0029] Figure 5 This is a flowchart illustrating the application of a convolutional neural network algorithm in an embodiment of the present invention.

[0030] Figure 6 This is a schematic flowchart of a defrosting control method for an energy storage heat pump air conditioning system according to an embodiment of the present invention;

[0031] Attached reference numerals: 1-Condenser fan, 2-Humidifier, 3-Cooling coil, 4-Sensor, 5-Nozzle, 6-Outdoor heat exchanger, 7-Infrared thermal imager, 8-Visible light camera, 9-Compressor, 10-Gas-liquid separator, 11-Electronic expansion valve, 12-Crew compartment, 13-Indoor air cooler, 14-Electric heater, 15-Indoor evaporator, 16-Pressure sensor, 17-Temperature sensor, Valves 1-6-Control switch. Detailed Implementation

[0032] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0033] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0034] like Figure 6 As shown, this embodiment provides a defrosting control method for an energy storage heat pump air conditioning system, including: S1, acquiring image data of the outdoor heat exchanger surface in real time through an image acquisition device;

[0035] S2. Process the image data to calculate the frost coverage rate;

[0036] S3. When the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, the image data is input into the pre-trained neural network model for frost level classification to obtain the frost level classification result.

[0037] S4. Based on the frost level classification result, trigger the corresponding defrosting mode, wherein the defrosting mode includes a weakened frost mode and a strengthened frost mode;

[0038] S5. During the defrosting mode, continuously monitor the frost coverage rate. When the frost coverage rate is lower than the fourth preset threshold, control the system to exit the defrosting mode and resume the heating mode.

[0039] Furthermore, in S1, acquiring image data of the outdoor heat exchanger surface includes: simultaneously acquiring image data of the outdoor heat exchanger surface using an infrared thermal imager 7 and a visible light camera 8; the infrared thermal imager 7 operates at a sampling frequency no lower than a preset frequency; the image acquisition device is positioned at a preset distance from the outdoor heat exchanger; the image data includes infrared images and visible light images, and is used for subsequent processing.

[0040] Specifically, the implementation process of this embodiment includes:

[0041] like Figure 1 As shown, the left side is a physical picture of the experimental measurement of the present invention. The visible light camera 8 and the infrared thermal imager 7 are placed 0.5m in front of the heat exchanger to collect images. The frost process of the heat exchanger is recorded by the ordinary camera and the infrared camera at the same time and used for subsequent image recognition. The right side is a physical and structural schematic diagram of the outdoor heat exchanger, which is a microchannel heat exchanger.

[0042] Furthermore, the defrosting prediction mechanism in this embodiment mainly includes two core judgment conditions: firstly, the system must be in heating operation mode, and the continuous operation time must exceed 11 minutes. (≥660s). This time threshold is set to avoid frequent system actions under frosting conditions. At the same time, considering that the necessity of defrosting is low during short-distance driving, the system will not perform defrosting actions.

[0043] like Figure 2 As shown, humidifier 2 and nozzle 5 are used to control outdoor air humidity to simulate the outdoor humidity conditions in winter. In heating mode, valves 1, 3, and 4 need to be opened, valves 2, 5, and 6 need to be closed, and electronic expansion valve 11 needs to be opened. The flow direction of refrigerant is shown by the arrow. After being compressed by compressor 9, the refrigerant becomes a high-temperature and high-pressure gaseous refrigerant working fluid. It enters the two series-connected indoor air coolers 13 and indoor evaporators 15 in the air conditioning unit and exchanges heat with the air brought by the fan. After the air absorbs heat and rises in temperature, it is blown into the passenger compartment 12. In order to accelerate the heating speed, an electric heater 14 is used to raise the temperature of the passenger compartment 12, while the refrigerant working fluid releases heat and condenses into a medium-temperature and high-pressure subcooled liquid. After being throttled by the electronic expansion valve 11 (the outlet pressure of the electronic expansion valve 11 is measured by the pressure sensor 16 to obtain system data), the refrigerant becomes a low-temperature, low-pressure mist. After flowing through the gas-liquid separator 10, it enters the outdoor heat exchanger 6 (the inlet and outlet temperatures and pressures of the outdoor heat exchanger 6 are observed by the sensor 4 to determine the stability of frost accumulation. The condenser fan 1 helps the outdoor heat exchanger 6 to exchange heat efficiently with the outdoor air, and the cooling coil 3 realizes the heat transfer process of the refrigerant to the indoor air). It absorbs heat from the ambient air through the flat tube wall and fins and gradually evaporates into a low-temperature, low-pressure superheated gas. After flowing through the gas-liquid separator 10 again, it evaporates and absorbs heat, completing the work of bringing ambient heat into the system. Finally, under the suction of the compressor 9, it returns to the compressor 9 and is compressed, continuing the cycle.

[0044] Furthermore, in S2, processing the image data to calculate the frost coverage rate includes: converting the image data into a grayscale image; performing binarization processing on the grayscale image using the Otsu threshold segmentation algorithm to distinguish between the frost layer region and the background region; calculating the ratio of the number of frost layer pixels to the total number of pixels in the binarized image, and using this ratio as the frost coverage rate; wherein the Otsu algorithm automatically determines the segmentation threshold.

[0045] Specifically, the implementation process of this embodiment includes:

[0046] During system operation, the heating capacity is one of the important indicators of system performance and is crucial for evaluating system operation. The heating capacity of the system is calculated using formulas (1) and (2).

[0047] (1);

[0048] (2);

[0049] in, It is the mass flow rate of air. It is the enthalpy value at the outlet of the outdoor heat exchanger. It is the enthalpy value at the inlet of the outdoor heat exchanger. It refers to the compressor power.

[0050] The coefficient of performance (COP) of a heat pump system is calculated using formula (3).

[0051] (3);

[0052] In addition, this embodiment introduces the concept of frost coverage ratio (FCR) to characterize the frosting characteristics of the outdoor heat exchanger under different operating conditions and the changes in system performance under different frost coverage ratios, as calculated below:

[0053] (4);

[0054] Afrost represents the frost area of ​​the heat exchanger. The frost coverage ratio (FCR) is calculated by converting the RGB image to a grayscale image using MATLAB programming. The Otsu algorithm, because it does not require manual threshold setting and is well-adapted to different lighting conditions and frost density, and based on maximum inter-class variance, ensures optimal segmentation results. Therefore, the Otsu thresholding method is used to calculate the global threshold for each process, and based on this global threshold, the frost image is converted into a binary image. Finally, the pixels of the binary image are scanned, and the pixel percentage is calculated to obtain the frost coverage ratio.

[0055] Otsu thresholding works by automatically finding the optimal threshold that maximizes the inter-class variance between the foreground (frost layer) and the background. Let the image grayscale level be [0, L-1], and the number of pixels with grayscale value i be n_i. The total number of pixels and the probability of grayscale value i are calculated using formulas (5) and (6):

[0056] (5);

[0057] (6);

[0058] The formula for calculating the variance between classes is (7):

[0059] (7);

[0060] In the formula, It is the foreground pixel ratio. The background pixel ratio, Foreground average gray level, The average gray level of the background.

[0061] The complete process of calculating frost coverage based on Otsu threshold segmentation is attached below. Frost coverage is calculated using formula (8). This complete process ensures the accuracy and repeatability of frost coverage calculation and is particularly suitable for monitoring frost growth under different experimental conditions.

[0062] (8).

[0063] Furthermore, in S3, the first preset condition is that the continuous operating time of the system exceeds a preset time threshold; the second preset condition is that the surface temperature uniformity of the outdoor heat exchanger is lower than a preset temperature threshold, and the surface temperature distribution of the heat exchanger is monitored by temperature sensor 17; the third preset condition is that the ambient temperature is within a preset temperature range; the neural network model is a convolutional neural network based on residual learning and has an embedded attention mechanism; the neural network is trained with image data and is used to classify the frosting level.

[0064] Specifically, the implementation process of this embodiment includes:

[0065] In the frosting severity determination stage, based on the key system performance parameters (heat output, COP) and the two stages of slow and rapid reduction, the frosting images are classified into categories 1 and 2. The classified images are then fed into a convolutional neural network (CNN) in a ratio of 1:1:8 for training, validation, and testing. The ResNet-50 CNN used in this embodiment is an important variant of deep residual networks. Its core innovation lies in its residual learning framework, which successfully solves the gradient vanishing and degradation problems in extremely deep neural networks. Traditional deep networks directly learn the target mapping H(x), while ResNet learns the residual mapping instead. The final output is obtained by adding the original input x with the learned residual F(x) through a shortcut connection. The profound insight of this design is that if multiple nonlinear layers can asymptotically approximate a complex objective function, then the equivalent residual function should also be easy to optimize. When the ideal mapping is close to the identity mapping, it is much easier to push the residual to zero than to fit the identity mapping with nonlinear layers.

[0066] The ResNet-50 architecture comprises 50 weighted layers, employing a bottleneck structure (containing three convolutional layers) to balance computational efficiency and representational power. The network begins with a large 7×7 convolutional kernel and a 3×3 max-pooling layer to quickly reduce spatial resolution. This is followed by four convolutional stages (convolutional layers 2 to 5), containing 3, 4, 6, and 3 residual blocks respectively. Each residual block uses a 1×1-3×3-1×1 convolutional sequence, forming a bottleneck design of dimensionality reduction-convolution-upgrading. The first 1×1 convolution reduces the number of channels to lower computational cost, the 3×3 convolution extracts effective features in the dimensionality-reduced space, and the second 1×1 convolution restores the number of channels. When the input and output dimensions do not match, dimensionality projection is performed through shortcut connections of 1×1 convolutions. At the network's end, global average pooling replaces traditional fully connected layers, significantly reducing the number of parameters while enhancing translation invariance.

[0067] Furthermore, in S4, triggering the corresponding defrosting mode includes: triggering a weakened defrosting mode when the frost level is classified as the first level; and triggering a strengthened defrosting mode when the frost level is classified as the second level. The weakened defrosting mode and the strengthened defrosting mode are achieved by controlling the opening and closing of multiple valves to change the refrigerant flow path. The valve control is based on the controller output command.

[0068] Specifically, the implementation process of this embodiment includes:

[0069] Specific methods for applying convolutional neural networks include Figure 5 As shown, the class loads image data from a directory structure, assigns a label to each image, and counts the occurrences of each label, through initialization, data preparation, label acquisition, dataset size retrieval, data item retrieval, and label statistics printing.

[0070] This section describes the process of loading and adjusting a ResNet-50 model using a custom dataset for deep learning. The specific steps are as follows: 1. Data loading and processing. 2. Training: This involves feeding data into the model via a data loader. 3. Validation: After each iteration, the model is evaluated using the validation set, calculating loss and accuracy. 4. Saving the best model: When the validation set accuracy exceeds the current best accuracy, the current model is saved. 5. Testing: After training and validation, the final model's performance is evaluated using the test set. The test results not only show the model's accuracy but also include a confusion matrix, providing a better understanding of the model's classification performance across different categories. 6. Plotting loss and accuracy: By recording the loss and accuracy for each iteration, the `plot_metrics` function is used to plot the changes during training, helping to monitor the model's learning process.

[0071] An attention mechanism is embedded in the model architecture, using a convolutional neural network to allow the model to adjust the weights of different parts of the input data based on their "importance," thus enabling the model to focus more on key features and ignore unimportant parts. Furthermore, transformations such as rotation, cropping, scaling, and flipping of the original image are applied to increase data diversity, thereby helping to improve the model's generalization ability.

[0072] Furthermore, in S5, continuous monitoring includes: during the execution of defrosting mode, continuing to acquire image data and calculate frost coverage rate through the image acquisition device, while monitoring the surface temperature distribution of the heat exchanger through the temperature sensor 17; the third preset threshold is a preset frost coverage rate threshold; the fourth preset threshold is a preset temperature uniformity error threshold; when the conditions are met, the system switches back to heating mode.

[0073] Specifically, the implementation process of this embodiment includes:

[0074] The defrosting mode switching process adopts a step-by-step execution strategy. The compressor 9 stops and performs system pressure balancing for about 10 seconds. Based on the algorithm's identification result of 1 or 2, the system is switched from heating mode to the corresponding 1 for weak defrosting mode and 2 for strong defrosting mode.

[0075] like Figure 4 In the weakened defrosting mode shown, valves 1, 2, and 5 are open, while valves 3, 4, and 6 are closed. The electronic expansion valve 11 is open, and the refrigerant flow direction is as shown by the arrow. After being compressed by the compressor 9, the refrigerant becomes a high-temperature, high-pressure gaseous refrigerant working fluid. It first enters the indoor air cooler (IGC) 13, and then enters the outdoor heat exchanger 6, where it exchanges heat with the air brought by the fan. The refrigerant working fluid releases heat and condenses into a medium-temperature, high-pressure subcooled liquid. After flowing through the gas-liquid separator 10, the refrigerant is throttled by the electronic expansion valve 11 and becomes a low-temperature, low-pressure mist working fluid. After entering the indoor evaporator 15, it exchanges heat with the air brought by the fan. After flowing through the gas-liquid separator 10 again, it evaporates and absorbs heat, completing the work of bringing ambient heat into the system. Finally, under the suction action of the compressor 9, it returns to the compressor 9 and is compressed again, continuing the cycle.

[0076] like Figure 3In the enhanced defrosting mode shown, valves 2 and 6 are open, while valves 1, 3, 4, and 5 are closed. The electronic expansion valve 11 is open, and the refrigerant flow direction is as shown by the arrow. After being compressed by the compressor 9, the refrigerant becomes a high-temperature, high-pressure gaseous refrigerant working fluid. It then enters the outdoor heat exchanger 6, where it exchanges heat with the air brought in by the fan. The refrigerant working fluid releases heat and condenses into a medium-temperature, high-pressure subcooled liquid. After flowing through the gas-liquid separator 10, the refrigerant is throttled by the electronic expansion valve 11 and becomes a low-temperature, low-pressure mist working fluid. After entering the indoor evaporator 15, it exchanges heat with the air brought in by the fan. It then flows through the gas-liquid separator 10 again, evaporating and absorbing heat, thus completing the work of bringing ambient heat into the system. Finally, under the suction action of the compressor 9, it returns to the compressor 9 and is compressed again, continuing the cycle.

[0077] The frost coverage ratio (FCR) on the system heat exchanger is read at this time. Defrosting ends when FCR < 21%. The defrosting strategy proposed in this embodiment integrates traditional control methods and finally uses a combination of multiple parameters for joint control.

[0078] This embodiment also provides a defrosting control system for an energy storage heat pump air conditioning system, including:

[0079] The image acquisition module is used to acquire image data of the outdoor heat exchanger surface in real time through an image acquisition device;

[0080] The frost coverage calculation module is used to process the image data to calculate the frost coverage.

[0081] The intelligent classification module is used to input the image data into a pre-trained neural network model to classify the frost level when the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, so as to obtain the frost level classification result.

[0082] The defrost mode triggering module is used to trigger the corresponding defrost mode based on the frost level classification result, wherein the defrost mode includes a weakened frost mode and a strengthened frost mode.

[0083] The exit control module is used to continuously monitor the frost coverage rate during the execution of the defrost mode. When the frost coverage rate is lower than the fourth preset threshold, the module controls the system to exit the defrost mode and resume the heating mode.

[0084] This invention discloses a defrosting control method and system for an energy storage heat pump air conditioning system. By combining image recognition and intelligent control, it achieves precise perception of the frost status of the outdoor heat exchanger and on-demand defrosting. Image processing technology is used to calculate the frost coverage rate, and a neural network is used to intelligently classify the frost level, effectively avoiding the ineffective operations and energy waste of traditional defrosting methods. By triggering differentiated defrosting modes based on the frost level, the system can minimize interference with the system's heating operation while ensuring defrosting effectiveness. This system effectively improves the accuracy and timeliness of defrosting control, significantly reduces system energy consumption, enhances operational stability and reliability, and reduces equipment maintenance requirements.

[0085] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A defrosting control method for an energy storage heat pump air conditioning system, characterized in that, include: S1. Real-time image data of the outdoor heat exchanger surface is acquired using an image acquisition device; S2. Process the image data to calculate the frost coverage rate; S3. When the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, the image data is input into the pre-trained neural network model for frost level classification to obtain the frost level classification result. S4. Based on the frost level classification result, trigger the corresponding defrosting mode, wherein the defrosting mode includes a weakened frost mode and a strengthened frost mode; S5. During the defrosting mode, continuously monitor the frost coverage rate. When the frost coverage rate is lower than the fourth preset threshold, control the system to exit the defrosting mode and resume the heating mode.

2. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, In S1, acquiring image data of the outdoor heat exchanger surface includes: simultaneously acquiring image data of the outdoor heat exchanger surface using an infrared thermal imager and a visible light camera; the infrared thermal imager operates at a sampling frequency no lower than a preset frequency; the image acquisition device is positioned at a preset distance from the outdoor heat exchanger; the image data includes infrared images and visible light images, and is used for subsequent processing.

3. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, In S2, processing the image data to calculate the frost coverage rate includes: converting the image data into a grayscale image; performing binarization processing on the grayscale image using the Otsu threshold segmentation algorithm to distinguish between the frost layer region and the background region; calculating the ratio of the number of frost layer pixels to the total number of pixels in the binarized image, and using this ratio as the frost coverage rate; wherein the Otsu algorithm automatically determines the segmentation threshold.

4. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, In S3, the first preset condition is that the continuous running time of the system exceeds a preset time threshold; the second preset condition is that the surface temperature uniformity of the outdoor heat exchanger is lower than a preset temperature threshold, and the surface temperature distribution of the outdoor heat exchanger is monitored by a temperature sensor; the third preset condition is that the ambient temperature is within a preset temperature range; the neural network model is a convolutional neural network based on residual learning and has an embedded attention mechanism; the neural network is trained with image data and is used to classify the frosting level.

5. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, In S4, the corresponding defrosting modes are triggered as follows: when the frost level is classified as the first level, a weakened defrosting mode is triggered; when the frost level is classified as the second level, a strengthened defrosting mode is triggered. The weakened defrosting mode and the strengthened defrosting mode are achieved by controlling the opening and closing of multiple valves to change the refrigerant flow path; wherein the valve control is based on the controller output command.

6. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, In S5, continuous monitoring includes: during the defrosting mode, continuing to acquire image data and calculate the frost coverage rate through the image acquisition device; the fourth preset threshold is a preset frost coverage rate threshold; when the frost coverage rate is higher than the fourth preset threshold, the system switches back to the heating mode.

7. The defrosting control method for an energy storage heat pump air conditioning system as described in claim 1, characterized in that, The method also includes, prior to S4, calculating the system's heating capacity and coefficient of performance (COP); the heating capacity is calculated based on the compressor power and the outdoor heat exchanger's heat exchange capacity; the COP is calculated based on the heating capacity and the compressor power; the calculation process uses preset formulas.

8. A defrosting control system for an energy storage heat pump air conditioning system, characterized in that, A defrosting control method for implementing an energy storage heat pump air conditioning system as described in any one of claims 1-7, the system comprising: The image acquisition module is used to acquire image data of the outdoor heat exchanger surface in real time through an image acquisition device; The frost coverage calculation module is used to process the image data to calculate the frost coverage. The intelligent classification module is used to input the image data into a pre-trained neural network model to classify the frost level when the system is in heating mode and the continuous running time meets the first preset condition, the surface temperature uniformity of the outdoor heat exchanger meets the second preset condition, and the ambient temperature meets the third preset condition, so as to obtain the frost level classification result. The defrost mode triggering module is used to trigger the corresponding defrost mode based on the frost level classification result, wherein the defrost mode includes a weakened frost mode and a strengthened frost mode; The exit control module is used to continuously monitor the frost coverage rate during the execution of the defrost mode. When the frost coverage rate is lower than the fourth preset threshold, the module controls the system to exit the defrost mode and resume the heating mode.

Images