An intelligent detection method for cold storage evaporator frosting based on visual recognition

Abstract

This invention discloses a visual recognition-based intelligent frost detection method for cold storage evaporators. The method includes: first, real-time monitoring of the evaporator's ambient temperature, relative humidity, surface temperature, and air pressure difference before and after air passage through the evaporator using data sensors installed on the evaporator; then, comparing the collected evaporator surface temperature with the dew point temperature, and activating an image acquisition device to monitor frost layers in multiple image acquisition areas distributed along the evaporator fins when a certain condition is met; next, processing the frost layer images uploaded by the image acquisition device using a visual frost thickness recognition module; finally, during defrosting, the image acquisition device monitors the frost layer status in real-time, and defrosting ends and normal refrigeration mode is restored when certain conditions are met. This invention can intelligently detect the frost condition of cold storage evaporators, achieving precise defrosting control.

Description

Technical Field

[0001] This invention relates to the field of evaporator frost detection technology, specifically to a visual recognition-based intelligent detection system and method for evaporator frost in cold storage. Background Technology

[0002] When the evaporator temperature of a refrigeration system is below the air dew point and below 0°C, frost will inevitably form on the evaporator. In the initial stage of frost formation, the frost layer indirectly increases the heat exchange area of ​​the evaporator, enhancing heat exchange, which is beneficial to the system. However, as the frost layer thickens, it reduces the fin spacing, blocks airflow channels, increases thermal resistance, and degrades the evaporator's heat exchange performance. If the evaporator is not defrosted in time, the frost layer will continue to thicken, reducing the efficiency of the refrigeration system and increasing overall energy consumption. Therefore, to ensure stable operation of refrigeration equipment and reduce energy consumption, it is necessary to defrost the evaporator.

[0003] Currently, defrosting can be broadly categorized into two control methods based on the degree of frost formation: indirect measurement control and direct measurement control. Indirect measurement control includes temperature-time control, time-pressure control, temperature difference-time control, and unit performance-based control. These methods are characterized by simple control devices and convenient operation; however, due to the complexity of frost formation and varying actual operating conditions of the unit, these methods inevitably experience "false defrosting" and have relatively stringent applicable conditions. Direct measurement control, on the other hand, disregards external factors affecting frost formation and directly measures the thickness of the frost layer on the system using equipment. This method is the most direct and effective, representing the most efficient control method under ideal conditions. However, since current frost thickness measurements mostly rely on microscopic imaging, laser diffraction, micrometer technology, photoelectric coupling technology, etc., these methods are easily affected by the objective environment and have high economic costs. Summary of the Invention

[0004] The purpose of this invention is to provide a visual recognition-based intelligent detection method for frost formation on cold storage evaporators. By utilizing artificial intelligence visual recognition technology, the frost formation on cold storage evaporators can be intelligently detected, thereby achieving precise control of defrosting.

[0005] To achieve the above objectives, the present invention provides a visual recognition-based intelligent detection method for frost formation on cold storage evaporators, comprising the following steps:

[0006] Step S1: Monitor the ambient temperature t, relative humidity RH, and surface temperature t of the evaporator in real time using data sensors installed on the evaporator. surface And the pressure difference ΔP between the air before and after passing through the evaporator;

[0007] Step S2: The surface temperature t of the evaporator is collected. surface With dew point temperature t f Compare; when t surface ≤t f When the temperature reaches +1℃, frost begins to gradually form on the evaporator fin side. Then, the image acquisition device is activated to monitor the frost layer in multiple image acquisition areas distributed on the evaporator fin side.

[0008] Step S3: Process the frost layer photos uploaded by the image acquisition device using a visual recognition frost layer thickness module, by analyzing the dimensionless gaps. When conducting analysis, When this happens, the system issues an early warning and continues monitoring; when At that time, defrosting should begin;

[0009] Step S4: During defrosting, the image acquisition device monitors the frost layer status in real time. and At this time, defrosting ends and normal cooling mode resumes.

[0010] Furthermore, the dew point temperature is calculated using the ambient temperature t and relative humidity RH, as shown in the formula:

[0011]

[0012]

[0013] Among them, ambient temperature t and dew point temperature t f The unit is degrees Celsius, the relative humidity RH is a percentage, ln is the natural logarithm, and the constants a and b are a = 17.27 and b = 237.7℃, respectively.

[0014] Furthermore, the image acquisition area is set to four, which are distributed on the evaporator fin side. The positional relationship between the four image acquisition areas and the upper, lower, left, and right sides of the evaporator fin side is expressed by the following formula:

[0015]

[0016]

[0017] Where L and H are the length and width of the evaporator fin side, respectively, and L1, L2, L3, L4, L5, H1, H2, H3, and H4 are the distances of the four image acquisition areas relative to the top, bottom, left, and right sides of the evaporator.

[0018] Furthermore, to better describe the state of the frost layer, a method for obtaining the dimensionless gaps in the frost layer is as follows: Figure 3As shown, the frost layer thickness recognition module identifies the frost layer boundary. After determining the boundary, the spacing Δ between the upper and lower boundaries of each fin is determined based on the pixels and magnification. Then, the dimensionless gap δ of the frost layer between the fins can be obtained.

[0019]

[0020] Where l is the distance between two adjacent fins, the average of the dimensionless gaps of the frost layer between all fins in the four image acquisition areas is used as the parameter for judging defrosting.

[0021]

[0022] Where N is the total number of fin spacings, n is the number of fin spacings in the i-th region, and k is the number of regions.

[0023] Furthermore, the visual frost thickness recognition module identifies the frost layer boundary through grayscale image binarization; after grayscale processing, the threshold value of each region of the frost layer image is obtained; the maximum grayscale value of the i-th region is defined as g. i,max The minimum gray level is g i,min The median value of this region is g. i,mid .

[0024] Furthermore, to narrow down the region where the threshold is determined, the median value is used as the boundary to obtain two gray-level intervals for the i-th region, which are respectively

[0025] I i,1 =(g i,min g i,mid )

[0026] I i,2 =(g i,mid g i,max )

[0027] Since the grayscale ranges of different regions are not uniform, the grayscale ranges of all regions are combined, and region I is then analyzed. i,1 Calculate the expected value and use it as the threshold for binarization, i.e.

[0028]

[0029] Where p j grayscale value g j The probability of occurrence is calculated; thus, a binarized map of each region is obtained. For each region's binarized map, the positions in the matrix that are all 0s and all 255s are selected, and the position in between is taken as the frost layer boundary position, i.e.:

[0030] L 边 = mid(L0,L 255 )

[0031] The beneficial effects of this invention are:

[0032] (1) The present invention sets four image acquisition points at different positions on the surface of a single evaporator, and simultaneously acquires images and data of four regions, so that the data obtained is more comprehensive and reliable.

[0033] (2) This invention performs grayscale binarization analysis on the image frost layer boundary, making the frost layer boundary more accurate and accurately judging the dimensionless gap of the frost layer. The dimensionless gap is the ratio of the gap between adjacent frost layers to the distance between adjacent fins. Compared with the traditional direct measurement of frost layer thickness, the dimensionless gap links the frost layer thickness with the fin distance, thereby enabling precise control and adjustment of the defrosting equipment.

[0034] (3) The present invention further improves the accuracy of defrosting by making dual judgments on the dimensionless gap of the frost layer and the rate of change of pressure difference before and after the evaporator.

[0035] (4) The present invention uses artificial intelligence visual recognition to accurately control the defrosting equipment of the evaporator, avoids false judgment of defrosting, reduces energy consumption and improves the efficiency of the refrigeration system. Attached Figure Description

[0036] Figure 1 This is a flowchart of the intelligent detection method for frost formation on cold storage evaporators based on visual recognition, as described in this invention.

[0037] Figure 2 This is a structural diagram of the intelligent frost detection system upon which this detection method is based;

[0038] Figure 3 This is a schematic diagram showing the frost formation between adjacent fins of the evaporator;

[0039] In the figure, 1-evaporator fin side, 2-image acquisition area, 3-image acquisition device, 4-circulation guide rail device, 5-fin, 6-frost layer boundary between upper and lower fins. Detailed Implementation

[0040] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] This invention provides a visual recognition-based intelligent detection method for frost formation on cold storage evaporators, and a cold storage evaporator frost detection system, such as... Figure 1As shown. This detection system includes data sensors, a circulating guide rail device 4, an image acquisition device 3, and a visual recognition module for frost thickness. The data sensors include temperature and humidity sensors and pressure sensors, installed on the fin side of the evaporator. The image acquisition device 3 includes a camera assembly and an electric actuator. The circulating guide rail device 4 is fixed to the four sides (top, bottom, left, and right) of the evaporator fin side 1 by a bracket. The image acquisition device 3 can circulate on the circulating guide rail device 4, sequentially acquiring data from multiple image acquisition areas 2 and uploading the data. Four image acquisition areas 2 are set on the evaporator fin side 1. The evaporator fins 5 are mainly flat fins, corrugated fins, slotted fins, or a combination of both (corrugated and slotted fins). This invention uses flat fins as an example for explanation.

[0042] The visual frost thickness recognition module is implemented using C++ programming. By analyzing the images acquired by the image acquisition device 3, it accurately determines the frost thickness, thereby controlling the start and stop of the defrosting device, thus achieving intelligent and precise control of defrosting.

[0043] like Figure 2 The diagram shows a flowchart of the intelligent detection method for frost formation on a cold storage evaporator based on visual recognition according to the present invention. The method includes the following steps:

[0044] Step S1: Monitor the ambient temperature t, relative humidity RH, and surface temperature t of the evaporator in real time using data sensors installed on the fin side 1 of the evaporator. surface And the pressure difference ΔP between the air before and after passing through the evaporator;

[0045] Step S2: The surface temperature t of the evaporator is collected. surface With dew point temperature t f For comparison, when t surface ≤t f At +1℃, the image acquisition device is activated to monitor frost layers in multiple image acquisition areas 2 distributed on the evaporator fins. Dew point temperature refers to the temperature at which air reaches saturation when cooled, provided that the water vapor content and air pressure remain unchanged. In simpler terms, it's the temperature at which water vapor in the air turns into dew. That is, when the temperature reaches the dew point temperature, water droplets begin to condense on the evaporator surface, and the conditions for frost formation are met.

[0046] In this embodiment, four image acquisition areas 2 are set, and the positions of the four areas are as follows: Figure 2 As shown, L and H are the length and width of the evaporator fins, respectively, and L1, L2, L3, L4, L5, H1, H2, H3, and H4 are the distances of four image acquisition areas relative to the top, bottom, left, and right sides of the evaporator, respectively.

[0047]

[0048]

[0049] Dew point temperature t f The formula is derived from the ambient temperature t and relative humidity RH:

[0050]

[0051]

[0052] Among them, ambient temperature t and dew point temperature t f The unit is degrees Celsius, the relative humidity RH is a percentage, ln is the natural logarithm, and the constants a and b are a = 17.27 and b = 237.7℃, respectively.

[0053] Step S3: Process the frost layer photos uploaded by the image acquisition device using a visual recognition frost layer thickness module, by analyzing the dimensionless gaps. When conducting analysis, When this happens, the system issues an early warning and continues monitoring; when At that time, defrosting treatment begins. This is achieved by determining the dimensionless gap between adjacent fins 5. If the dimensionless gap is greater than 0.8, the system continues to acquire images of the four regions and continues detection. Otherwise, the system issues a warning and determines whether the dimensionless gap is less than or equal to 0.4. If the dimensionless gap is greater than 0.4, the system continues detection. If the dimensionless gap is less than or equal to 0.4, the system starts the defrosting device and begins defrosting.

[0054] The working principle of the Baidu visual recognition frost layer thickness module is as follows: First, the module converts the acquired image to grayscale. The color value of each pixel in a grayscale image is called grayscale, which refers to the color depth of a point in a black and white image. The range is generally from 0 to 255, with white being 255 and black being 0. Grayscale value refers to the intensity of a color. Image grayscale conversion can serve as a preprocessing step for image processing, preparing for subsequent higher-level operations such as image segmentation, image recognition, and image analysis.

[0055] Then, after the acquired image is converted to grayscale, it is binarized by a program. The fundamental principle is to use a set threshold to determine whether an image pixel is 0 or 255, so the threshold setting is very important in image binarization. In this invention, since the boundary of the frost layer is not a regular straight line but an irregular curve, the frost layer image needs to be segmented and analyzed to obtain the threshold value of each region. Then, the maximum grayscale value of the i-th region is gi,max, the minimum grayscale value is gi,min, and the median value of the region is gi,mid. This method achieves the judgment of the frost layer boundary by binarizing the frost layer image, and thus the judgment of the dimensionless gap between frost layers. The determination of the threshold value is the key in the binarization process of this method. The so-called threshold value is the boundary between two grayscale values ​​in the binarization process. All pixels in the image with pixel values ​​less than or equal to the threshold value are set to one pixel value, and all pixels in the image with pixel values ​​greater than the threshold value are set to another pixel value.

[0056] By binarizing the frost layer image, the boundaries of the frost layers can be determined, thereby enabling the determination of the dimensionless gaps between frost layers. To narrow down the area for determining the threshold, the median value is used as the boundary to obtain two gray-level intervals for the i-th region, which are:

[0057] I i,1 =(g i,min g i,mid )

[0058] I i,2 =(g i,mid g i,max )

[0059] Since the grayscale ranges of different regions are not uniform, the grayscale ranges of all regions are combined, and the expected value of region Ii,1 is calculated and used as the threshold for binarization, that is:

[0060]

[0061] Where pj is the probability of the grayscale value gj appearing.

[0062] This yields a binarized image of each region of the image. For each region's binarized image, the positions in the matrix that are all 0s and all 255s are compared; the midpoint between these two values ​​is taken as the boundary position.

[0063] L 边 = mid(L0,L 255 )

[0064] To better describe the state of frost, this invention introduces a dimensionless gap. Generally, describing the gap of a frost layer only reflects the magnitude of the gap, not the relationship between the frost layer and adjacent fins, making it impossible to accurately determine the degree of frost formation. Introducing a dimensionless gap overcomes this shortcoming. The dimensionless gap is the ratio of the gap between adjacent frost layers to the distance between adjacent fins, thus linking the frost layer thickness to the fin spacing. After determining the boundaries, the distance 6 between the upper and lower boundaries of the frost layer between each fin is determined based on the pixel count and magnification. Figure 3 Δ in the equation, and then the dimensionless gap δ between the frost layers in the fins can be obtained:

[0065]

[0066] The average of the dimensionless gaps of the frost layer between all fins in each region is used as the parameter for determining defrosting.

[0067]

[0068] Where N is the total number of fin spacings, n is the number of fin spacings in the i-th region, and k is the number of regions.

[0069] Step S4: During defrosting, the image acquisition device monitors the frost layer status in real time. and At this time, defrosting ends and normal cooling mode resumes.

[0070] This invention automatically and in real-time collects temperature, humidity, and air pressure parameters on the fin side using an image acquisition device. It can monitor a range of data, including ambient temperature and humidity, and then autonomously determines whether to initiate defrosting, achieving a high degree of automation. Four observation areas are set up for comprehensive observation of the evaporator's frosting status. A dimensionless gap between adjacent fins of the evaporator is introduced as a standard for detecting frost thickness. This invention not only detects frost on the evaporator but also provides early warnings and defrosting mechanisms. During binarization, this invention calculates the expected value of the grayscale range of each region on the frost image of the same fin as a binarization threshold. To determine whether defrosting has ended, this invention sets two criteria: the dimensionless gap between adjacent fins and the rate of change of pressure difference across the evaporator. These two criteria allow for precise control of the defrosting effect.

[0071] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes that can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention are all within the protection scope of the claims of the present invention.

Claims

1. A visual recognition-based intelligent detection method for frost formation on cold storage evaporators, characterized in that, Includes the following steps: Step S1: Monitor the ambient temperature t, relative humidity RH, and surface temperature t of the evaporator in real time using data sensors installed on the evaporator. surface And the pressure difference ΔP between the air before and after passing through the evaporator; Step S2: By collecting the surface temperature t of the evaporator surface With dew point temperature t f For comparison, when t surface ≤t f At +1℃, the image acquisition device is activated to monitor frost layer in multiple image acquisition areas distributed on the side of the evaporator fins. Step S3: The frost layer thickness is processed using a visual recognition module to analyze the frost layer photos uploaded by the image acquisition device. This processing includes: identifying the frost layer boundaries through grayscale image binarization, determining the distance Δ between the upper and lower boundaries of the frost layer between each fin, and then calculating the dimensionless gap δ between the fins using the formula δ = Δ / l, where l is the distance between two adjacent fins; and then calculating the dimensionless gap... When conducting analysis, When this happens, the system issues an early warning and continues monitoring; when At that time, defrosting should begin; Step S4: During defrosting, the image acquisition device monitors the frost layer status in real time. and At this time, defrosting ends and normal cooling mode resumes.

2. The intelligent detection method for frost formation on a cold storage evaporator based on visual recognition as described in claim 1, characterized in that, The dew point temperature is calculated using the ambient temperature t and relative humidity RH, using the following formula: Among them, ambient temperature t and dew point temperature t f The unit is degrees Celsius, the relative humidity RH is a percentage, ln is the natural logarithm, and the constants a and b are a = 17.27 and b = 237.7℃, respectively.

3. The intelligent detection method for frost formation on a cold storage evaporator based on visual recognition as described in claim 1, characterized in that, The image acquisition area is set to four, which are distributed on the side of the evaporator fins. The positional relationship between the image acquisition area and the upper, lower, left, and right sides of the evaporator fins is as follows: Where L and H are the length and width of the evaporator fin side, respectively, and L1, L2, L3, L4, L5, H1, H2, H3, and H4 are the distances of the four image acquisition areas relative to the top, bottom, left, and right sides of the evaporator.

4. The intelligent detection method for frost formation on a cold storage evaporator based on visual recognition as described in claim 3, characterized in that, The method for obtaining the dimensionless gap of the frost layer is as follows: The frost layer boundary is identified using a visual recognition module to determine the frost layer thickness. After determining the boundary, the distance Δ between the upper and lower boundaries of each fin is determined based on the pixels and magnification. Then, the dimensionless gap δ of the frost layer between the fins can be obtained. Where l is the distance between two adjacent fins, the average of the dimensionless gaps of the frost layer between all fins in the four image acquisition areas is used as the parameter for judging defrosting. Where N is the total number of fin spacings, n is the number of fin spacings in the i-th region, and k is the number of regions.

5. The intelligent detection method for frost formation on a cold storage evaporator based on visual recognition according to claim 4, characterized in that, The visual recognition frost layer thickness module identifies the frost layer boundary through grayscale image binarization processing; after grayscale processing, the threshold value of each region of the frost layer image is obtained; Define the maximum gray level of the i-th region as g. i,max The minimum gray level is g i,min The median value of this region is g. i,mid .

6. The intelligent detection method for frost formation on a cold storage evaporator based on visual recognition according to claim 5, characterized in that, To improve accuracy, the median grayscale value of all pixels in the i-th region is used as the boundary to obtain two grayscale intervals for the i-th region, namely: I i,1 =(g i,min ,g i,mid ) I i,2 =(g i,mid ,g i,max ) For region I i,1 Calculate the expected value and use it as the threshold for binarization, i.e. Where p j grayscale value g j The probability of occurrence; thus, a binarized map of each region is obtained. For each region's binarized map, the positions in the judgment matrix that are all 0s and all 255s are selected, and the position in between is taken as the frost layer boundary position, i.e.: L 边 <mid(L0,L 255 )。

Images