CN-122015402-A - Defrosting control method for air source heat pump and air source heat pump system

CN122015402ACN 122015402 ACN122015402 ACN 122015402ACN-122015402-A

Abstract

The application provides an air source heat pump defrosting control method and an air source heat pump system, wherein the method comprises the steps of obtaining current sampling data of an outdoor fan of an air source heat pump to obtain a current time sequence segment; acquiring outdoor environment parameters and air source heat pump operation state parameters which are synchronously acquired with current sampling data, converting all the outdoor environment parameters and the air source heat pump operation state parameters into environment state vectors, converting a current time sequence segment into waveform feature vectors, carrying out data fusion on the waveform feature vectors and the environment state to obtain multidimensional input vectors, inputting the multidimensional input vectors into a one-dimensional residual network model to obtain a dynamic current feature threshold, and controlling the air source heat pump to defrost under the condition that a current feature index is larger than the dynamic current feature threshold, thereby solving the problem of poor accuracy of defrosting control.

Inventors

ZHANG HAO

Assignees

广东万和新电气股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260331

Claims (15)

1. The defrosting control method for the air source heat pump is characterized by comprising the following steps of: acquiring current sampling data of an outdoor fan of an air source heat pump to obtain a current time sequence segment; Acquiring outdoor environment parameters and air source heat pump operation state parameters which are synchronously acquired with the current sampling data, converting all the outdoor environment parameters and the air source heat pump operation state parameters into environment state vectors, converting the current time sequence segments into waveform feature vectors, and carrying out data fusion on the waveform feature vectors and the environment states to obtain multidimensional input vectors, wherein the air source heat pump operation state parameters are parameters of the air source heat pump except for the current of the outdoor fan; inputting the multi-dimensional input vector into a one-dimensional residual error network model to obtain a dynamic current characteristic threshold, wherein the one-dimensional residual error network model is obtained by training a historical multi-dimensional input vector and a corresponding historical dynamic current characteristic threshold, and the dynamic current characteristic threshold is the maximum value of a current characteristic index formed by nonlinear texture characteristics which do not trigger defrosting; and controlling the air source heat pump to defrost under the condition that the current characteristic index is larger than the dynamic current characteristic threshold, wherein the current characteristic index is one or more of roughness, aperiodic dithering and envelope distortion of the current of the outdoor fan in a preset period.
2. The method of claim 1, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: Acquiring the current characteristic indexes and the corresponding dynamic current characteristic thresholds of a plurality of preset periods of a continuous verification window; and controlling the air source heat pump to defrost under the condition that the number of the current characteristic indexes which are larger than the corresponding dynamic current characteristic threshold is larger than or equal to a preset number.
3. The method of claim 1, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: And controlling the air source heat pump to defrost under the condition that the classification label in the frosting stage is the moderate frosting or the heavy frosting and the current characteristic index is larger than the dynamic current characteristic threshold value.
4. A method according to claim 3, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold, comprising: Acquiring the current characteristic indexes and the corresponding dynamic current characteristic thresholds of a plurality of preset periods of a continuous verification window; and controlling the air source heat pump to defrost under the condition that the classification label in the frosting stage is the moderate frosting or the heavy frosting and the quantity of the current characteristic indexes which are larger than the corresponding dynamic current characteristic threshold value is larger than or equal to the preset quantity.
5. The method of claim 1, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: And controlling the air source heat pump to defrost under the conditions that the continuous heating time length is greater than or equal to a time length threshold, the heat exchange temperature difference between the outdoor air and the heat exchange coil of the outdoor heat exchanger of the air source heat pump is greater than or equal to a temperature difference threshold and the current characteristic index is greater than the dynamic current characteristic threshold.
6. The method of claim 5, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: Acquiring the current characteristic indexes and the corresponding dynamic current characteristic thresholds of a plurality of preset periods of a continuous verification window; And controlling the air source heat pump to defrost under the condition that the continuous heating time length is greater than or equal to the time length threshold, the heat exchange temperature difference between the outdoor air and the heat exchange tube of the outdoor heat exchanger of the air source heat pump is greater than or equal to the temperature difference threshold and the quantity of the current characteristic indexes which are greater than the corresponding dynamic current characteristic threshold is greater than or equal to the preset quantity.
7. The method of claim 1, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: And controlling the air source heat pump to defrost under the conditions that the continuous heating time length is greater than or equal to a time length threshold, the heat exchange temperature difference between the outdoor air and the heat exchange coil of the outdoor heat exchanger of the air source heat pump is greater than or equal to a temperature difference threshold, the classification label of the frosting stage is the medium frosting or the heavy frosting, and the current characteristic index is greater than the dynamic current characteristic threshold.
8. The method of claim 7, wherein the output of the one-dimensional residual network model further comprises a frost phase classification tag comprising no frost or incipient frost, moderate frost, and heavy frost, wherein controlling the air source heat pump to defrost if a current signature indicator is greater than the dynamic current signature threshold comprises: Acquiring the current characteristic indexes and the corresponding dynamic current characteristic thresholds of a plurality of preset periods of a continuous verification window; And controlling the air source heat pump to defrost under the conditions that the continuous heating time is longer than or equal to the time threshold, the heat exchange temperature difference between the outdoor air and the heat exchange tube of the outdoor heat exchanger of the air source heat pump is greater than or equal to the temperature difference threshold, the classification label of the frosting stage is the medium frosting or the heavy frosting, and the quantity of the current characteristic indexes which are greater than the corresponding dynamic current characteristic threshold is greater than or equal to the preset quantity.
9. The method according to any one of claims 2, 4, 6 and 8, wherein after obtaining the current signature indicators and the corresponding dynamic current signature thresholds for a plurality of predetermined periods of a continuous verification window, the method further comprises: And under the condition that the number of the current characteristic indexes larger than the corresponding dynamic current characteristic threshold value is smaller than the preset number, entering the next continuous check window to judge whether defrosting is carried out again.
10. The method of claim 1, wherein inputting the multi-dimensional input vector into a one-dimensional residual network model results in a dynamic current signature threshold, comprising: inputting the multidimensional input vector into a feature extraction trunk of the one-dimensional residual error network model to obtain nonlinear texture features of a current waveform; And inputting the nonlinear texture features into a normalization layer of a classification branch of the one-dimensional residual network model to obtain a frosting stage classification label, and inputting the nonlinear texture features into a full-connection layer of a regression branch of the one-dimensional residual network model to obtain the dynamic current feature threshold.
11. The method of claim 10, wherein inputting the multi-dimensional input vector into the feature extraction backbone of the one-dimensional residual network model yields nonlinear texture features of a current waveform, comprising: And inputting the multidimensional input vector into a plurality of stacked residual blocks of the feature extraction trunk, extracting to obtain the nonlinear texture features, wherein each residual block comprises a one-dimensional convolution layer, a batch normalization layer and an activation function layer with convolution kernel size adapted to the waveform texture time scale and is provided with jump connection, and the nonlinear texture features are used for representing the roughness, jitter frequency distribution and envelope distortion degree of a current waveform.
12. The method of claim 1, wherein obtaining current sampling data of an outdoor fan of an air source heat pump to obtain a current timing segment comprises: acquiring the current sampling data of the outdoor fan acquired according to a preset sampling rate; Denoising pretreatment is carried out on the current sampling data to obtain denoising current sampling data, wherein the denoising pretreatment at least comprises filtering of power frequency fundamental waves, harmonic components and fan start-stop surge components; and intercepting the denoising current sampling data with preset time length to obtain the current time sequence segment.
13. The method of claim 12, wherein the predetermined sampling rate has a value in the range of 50Hz to 3000Hz.
14. The method according to any one of claims 5 to 8, wherein the time duration threshold is in the range of 35min to 60min and the temperature difference threshold is in the range of 6 ℃ to 9 ℃.
15. An air source heat pump system comprising an air source heat pump and a control system, the control system comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for performing the method of any of claims 1-14.

Description

Defrosting control method for air source heat pump and air source heat pump system Technical Field The application relates to the technical field of air source heat pumps, in particular to an air source heat pump defrosting control method and an air source heat pump system. Background The Air Source Heat Pump (ASHP) is used as a high-efficiency energy-saving heating device and is widely applied to cold, summer hot and winter cold areas. However, when the outdoor ambient temperature is low and the relative humidity is high, frosting of the surface of the outdoor heat exchanger (evaporator) inevitably occurs. The accumulation of the frost layer can gradually block the air flow channels among the heat exchange fins, so that the heat resistance is increased, the air quantity is attenuated, and even the liquid return or shutdown protection of the compressor can be caused when the air quantity is serious. Therefore, switching on the defrost mode at the best timing is critical to maintaining the air source heat pump energy efficiency ratio (COP) and heating comfort. The existing defrosting control technology is mainly divided into a direct measurement method and an indirect push algorithm. Direct measurement relies on infrared, photoelectric or pressure sensors, although the accuracy is high, the hardware cost is high and is easily interfered by environmental factors such as dust accumulation, vibration and the like, and the method is difficult to popularize in a commercial air source heat pump on a large scale. The indirect calculation method estimates the frosting degree by monitoring system operation parameters (such as heating time, coil temperature, current change and the like). The detection method based on the outdoor fan current is a main stream direction of the current engineering application because the existing current sensor of the air source heat pump is utilized, and additional hardware transformation is not needed. The existing fan current defrosting control technology is generally based on the physical assumption that thickening of a frost layer causes narrowing of flow channels among fins, increases of air flow resistance, further causes change of load torque of a fan motor, and finally shows macroscopic increase of current effective value (RMS) or average amplitude. Therefore, the prior art mostly uses the average increase rate, the amplitude increment or the slope of the calculated current as the core decision index, and sets a linear fixed threshold to trigger defrosting. But the ambient wind speed and voltage fluctuation interfere to cause misjudgment, and the defrosting control has poor accuracy. Disclosure of Invention The first technical problem to be solved by the present invention is to provide a defrosting control method for an air source heat pump, which effectively improves the accuracy of defrosting control. A second technical problem to be solved by the present invention is to provide an air source heat pump system, which effectively improves the accuracy of defrosting control. The first technical problem is solved by the following technical scheme: A defrosting control method of an air source heat pump comprises the steps of obtaining current sampling data of an outdoor fan of the air source heat pump to obtain a current time sequence segment, obtaining outdoor environment parameters and air source heat pump operation state parameters which are synchronously collected with the current sampling data, converting all the outdoor environment parameters and the air source heat pump operation state parameters into environment state vectors, converting the current time sequence segment into waveform characteristic vectors, carrying out data fusion on the waveform characteristic vectors and the environment state to obtain a multidimensional input vector, wherein the air source heat pump operation state parameters are parameters of the air source heat pump except for current of the outdoor fan, inputting the multidimensional input vector into a one-dimensional residual network model to obtain a dynamic current characteristic threshold, wherein the one-dimensional residual network model is obtained by training the historical multidimensional input vector and a corresponding historical dynamic current characteristic threshold, the dynamic current characteristic threshold is a maximum value of a current characteristic index formed by non-triggering defrosting, and the current characteristic index is controlled to be one or more than one of roughness, distortion and vibration of the outdoor fan in a preset period under the condition that the current characteristic index is larger than the dynamic current characteristic threshold. Compared with the background technology, the defrosting control method of the air source heat pump has the following beneficial effects: According to the technical scheme, in the air source heat pump defrosting control method, by capturing a current time sequence wavefo