Refrigerator surface anti-condensation control method based on machine learning

Abstract

The invention provides a refrigerator surface anti-condensation control method based on machine learning, and belongs to the technical field of refrigeration. The mathematical model of the surface temperature of the refrigerator is established through machine learning, the heating duty ratio of the heating wire is adjusted in real time, and compared with a fixed heating duty ratio, the heating duty ratio of the heating wire can be increased or decreased according to the actual working environment, so that the working efficiency is improved. The surface temperature of the refrigerator is more accurately controlled within a narrow temperature range slightly higher than the dew point temperature, and the problem of energy waste of a traditional anti-condensation control technology is solved.

Description

Technical Field

[0001] The present invention relates to the field of refrigeration technology, and in particular to a refrigerator surface anti-condensation control method based on machine learning. Background Art

[0002] During the working cycle of commercial or household refrigerators, condensation on the outer surfaces such as door frames and panels has become a common and urgent problem that needs to be solved. This phenomenon not only threatens food safety, but also increases the burden of daily maintenance. In order to deal with this problem, the traditional practice is to embed electric heating wires on the inner side of the outer surface prone to condensation and maintain it in a continuous heating state, aiming to ensure that the outer surface temperature is always higher than the ambient dew point temperature, thereby eliminating condensation. However, in the face of dynamic changes in ambient temperature and humidity, the dew point temperature also fluctuates, which causes traditional anti-condensation technology to often go too far, making the surface temperature much higher than the dew point temperature, which not only feels warm or even hot to the touch, but also causes significant energy waste. Taking commercial kitchen refrigerators as an example, the energy consumption of traditional anti-condensation technology accounts for as much as 1 / 7 to 1 / 6 of the total energy consumption of the refrigerator, which is a significant waste of energy. Summary of the invention

[0003] According to the technical problems of touch warmth and energy waste mentioned in the above background technology, a method for controlling condensation on the refrigerator surface based on machine learning is provided. The present invention constructs a prediction model for the dynamic change of the refrigerator surface temperature through machine learning; and based on the established prediction model, the heating duty cycle parameter of the heating wire is adjusted to control the surface temperature within a narrow range slightly higher than the dew point temperature, which can effectively prevent the occurrence of condensation and significantly reduce energy consumption, thereby achieving the dual goals of low energy consumption and high-precision control.

[0004] The technical means adopted by the present invention are as follows:

[0005] A refrigerator surface anti-condensation control method based on machine learning comprises the following steps:

[0006] Model building phase: Establish a dynamic change prediction model for refrigerator surface temperature s ;

[0007] Learning stage: Determine the refrigerator surface temperature dynamic change prediction model t s All coefficients in t are used to establish the final refrigerator surface temperature dynamic change prediction model t s ;

[0008] Control stage: Control the refrigerator surface temperature within a narrow temperature range [t d +l,t d +l+0.5], where t d Indicates the dew point temperature of the current environment; l indicates the bias constant;

[0009] The refrigerator surface temperature dynamic change prediction model is:

[0010] t s =f(t i ,t o ,r,u)

[0011] Among them, t s Indicates the surface temperature of the refrigerator; t i Indicates the internal temperature of the refrigerator; t o represents the external environment temperature; u represents the working voltage of the heating wire; r represents the heating duty cycle of the heating wire, that is, the ratio of the heating wire power-on time per unit time, 0≤r≤100%.

[0012] Furthermore, the learning phase includes the following steps:

[0013] Step 1.1: Determine the experimental parameters; use the refrigerator surface temperature dynamic change prediction model t s Each input parameter in is set to three levels, and an orthogonal test is used to obtain L9(3 4 )Orthogonal test table;

[0014] Step 1.2: Data collection: The L9 (3 4 ) Each row of parameters in the orthogonal test table is set as the actual operating parameters of the refrigerator. After a certain period of operation in a constant temperature and humidity refrigeration laboratory, the surface temperature of the refrigerator is collected after the refrigerator reaches a balanced state;

[0015] Step 1.3: Based on the collected surface temperature data Determine the model t through machine learning technology s =f(t i ,t o ,r,u), that is, to determine the prediction model of the dynamic change of the surface temperature of the refrigerator.

[0016] Furthermore, the control stage comprises the following steps:

[0017] Step 2.1: Collect real-time data of refrigerator operation at a certain sampling frequency, including: the internal temperature of the refrigerator t i , external environment temperature t o 、Relative humidity of external environment h o And the working voltage u of the heating wire;

[0018] Step 2.2: Calculate the dew point temperature; the dew point temperature is approximately determined by the Magnus-Tetens formula, that is:

[0019]

[0020] in, a=17.27,b=237.7℃;

[0021] Step 2.3: Based on the refrigerator surface temperature dynamic change prediction model established in the learning phase, calculate the optimal heating duty cycle r of the heating wire * , the refrigerator surface temperature t s Controlled in a narrow temperature range [t d +I,t d +l+0.5], so that the surface temperature of the refrigerator is not lower than the dew point temperature, thereby preventing condensation; where l represents the bias constant;

[0022] Step 2.4: Control the mainboard through the refrigerator and set the optimal heating duty cycle r * , to achieve precise heating of the heating wire.

[0023] Furthermore, the L9(3 4 ) The orthogonal test table is:

[0024] Test No. <![CDATA[Internal temperature t i > <![CDATA[Ambient temperature t o > Duty cycle r Voltage u 1 1 1 1 1 2 1 2 3 2 3 1 3 2 3 4 2 1 3 3 5 2 2 2 1 6 2 3 1 2 7 3 1 2 2 8 3 2 1 3 9 3 3 3 1

[0025] The setting values ​​of each parameter in the orthogonal experiment table, namely the numbers 1, 2, 3, represent the setting level of each parameter.

[0026] Furthermore, the equilibrium state is the internal temperature of the refrigerator t i , external environment temperature t o , voltage u, heating duty cycle r and surface temperature t s All remain stable, that is, the change within 20 minutes does not exceed 0.2.

[0027] Furthermore, the certain period of time is 4 to 5 hours.

[0028] Furthermore, the refrigerator surface temperature is the temperature of the lowest point on the refrigerator surface.

[0029] Furthermore, the certain sampling frequency is 1 to 2 seconds.

[0030] Compared with the prior art, the present invention has the following advantages:

[0031] The present invention establishes a mathematical model of the refrigerator surface temperature through machine learning, and realizes real-time adjustment of the heating duty cycle of the heating wire. Compared with a fixed heating duty cycle, the heating duty cycle of the heating wire can be increased or decreased according to the actual working environment, so as to more accurately control the refrigerator surface temperature within a narrow temperature range slightly higher than the dew point temperature, thereby avoiding the energy waste problem of traditional anti-condensation control technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0033] Figure 1 This is a flow chart of a refrigerator surface temperature control method based on machine learning according to the present invention;

[0034] Figure 2 It is a fitting curve diagram of the refrigerator surface temperature dynamic change prediction model of the present invention. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] like Figure 1-2 As shown, the present invention provides a refrigerator surface anti-condensation control method based on machine learning, comprising the following steps:

[0038] A. Model building stage: Establish a dynamic change prediction model for refrigerator surface temperature s ; The refrigerator surface temperature dynamic change prediction model is:

[0039] t s =f(t i ,t o ,r,u)

[0040] Among them, t s Indicates the surface temperature of the refrigerator; t i Indicates the internal temperature of the refrigerator; t o represents the external environment temperature; u represents the working voltage of the heating wire; r represents the heating duty cycle of the heating wire, that is, the ratio of the heating wire power-on time per unit time, 0≤r≤100%. Model t s =f(t i ,t o ,r,u) is fixed. For different types of refrigerators, model t s =f(t i ,t o ,r,u) only have different values ​​in some coefficients of the function expression.

[0041] B. Learning stage: Determine the refrigerator surface temperature dynamic change prediction model t s All coefficients in; specifically, comprising the following steps:

[0042] Step 1.1: Determine the experimental parameters; use the refrigerator surface temperature dynamic change prediction model t s Each input parameter in is set to three levels, and an orthogonal test is used to obtain L9(3 4 ) orthogonal test table; In this application, the L9(3 4 ) The orthogonal test table is:

[0043] Test No. <![CDATA[Internal temperature t i > <![CDATA[Ambient temperature t o > Duty cycle r Voltage u 1 1 1 1 1 2 1 2 3 2 3 1 3 2 3 4 2 1 3 3 5 2 2 2 1 6 2 3 1 2 7 3 1 2 2 8 3 2 1 3 9 3 3 3 1

[0044] The setting values ​​of each parameter in the orthogonal experiment table, i.e., the numbers 1, 2, and 3, represent the setting level of each parameter. Specifically, in this application, as a preferred embodiment, each parameter is set as follows:

[0045] Internal temperature i :For different refrigerators, due to different uses, the internal temperature t i The range is also different. At the same time, in the experiment, the internal temperature t i To keep the temperature constant, a variable frequency compressor is required to adjust the working frequency to achieve constant temperature control of the internal temperature (if the refrigerator is a non-variable frequency refrigerator, a variable frequency technology can be used to make an experimental refrigerator), that is: the internal temperature t i Converted into the compressor operating frequency freq, if the compressor operating frequency range freq is [25,70], then its three levels are set to 25, 47, 70; For example: when the compressor operating frequency freq is set to 25, the actual refrigerator internal temperature t i Indicates the internal temperature t i The value of level 1 of the factor.

[0046] Ambient temperatureo :For different refrigerators, due to different uses, the ambient temperature t o The range is also different. If the ambient temperature t o The range of is [10,40], and its three levels are set to 10, 25, 40.

[0047] Heating duty cycle r: The operating range is [0,100], and its three levels are set to 0, 50, and 100.

[0048] Voltage u: The actual working environment of the refrigerator is different, and the working range of voltage u is slightly different. If the working range of voltage u is [200,240], its three levels are set to 200, 220, and 240.

[0049] Step 1.2: Data collection: The L9 (3 4 ) Each row of parameters in the orthogonal test table is set as the actual operating parameters of the refrigerator. After a certain period of operation (4 to 5 hours) in a constant temperature and humidity refrigeration laboratory, the surface temperature of the refrigerator is collected after the refrigerator reaches a equilibrium state; in this application, the equilibrium state is the internal temperature of the refrigerator t i , external environment temperature t o , voltage u, heating duty cycle r and surface temperature t s All remain stable, that is, the change within 20 minutes does not exceed 0.2.

[0050] Step 1.3: Based on the collected surface temperature data Determine the model t through machine learning technology s =f(t i ,t o ,r,u), that is, to determine the dynamic change prediction model of the refrigerator surface temperature. In this application, the machine learning technology refers to the use of function mining technology and tools to divide the original data set into a learning set and a test set, complete machine learning on the learning set, and apply the learning results to the test set for verification.

[0051] C. Control stage: Control the surface temperature of the refrigerator within a narrow temperature range [t d +l,t d +l+0.5], the final refrigerator surface temperature dynamic change prediction model t is generated s ; where t d represents the dew point temperature of the current environment; l represents the bias constant, which is usually set to 2. Specifically, the following steps are included:

[0052] Step 2.1: Collect real-time data of refrigerator operation at a certain sampling frequency (usually 1 to 2 seconds), including: the internal temperature of the refrigerator t i(Unit: °C), external environment temperature t o (Unit: °C), relative humidity of external environment h o (Unit: %) and the working voltage u of the heating wire (unit: V).

[0053] Step 2.2: Calculate the dew point temperature; the dew point temperature is approximately determined by the Magnus-Tetens formula, that is:

[0054]

[0055] in, a=17.27,b=237.7℃;

[0056] Step 2.3: Based on the refrigerator surface temperature dynamic change prediction model established in the learning phase, calculate the optimal heating duty cycle r of the heating wire * , the refrigerator surface temperature t s Controlled in a narrow temperature range [t d +l,t d +l+0.5], so that the surface temperature of the refrigerator is not lower than the dew point temperature, thereby preventing condensation;

[0057] Step 2.4: Control the mainboard through the refrigerator and set the optimal heating duty cycle r * , to achieve precise heating of the heating wire.

[0058] As a preferred embodiment, in this application, the refrigerator surface temperature mentioned is the temperature of the lowest point of the refrigerator surface temperature. It can be understood that in other embodiments, the surface temperature may be the average temperature of all temperatures or the highest temperature or a value within a preset range. At the same time, the refrigerator internal temperature mentioned in this application refers to the temperature of the middle area inside the refrigerator.

[0059] Embodiment 1:

[0060] The dynamic change prediction model of refrigerator surface temperature established in the learning stage t s =f(t i ,t o ,r,u) The technical indicators achieved are: learning accuracy R 2 =0.9906; the average error is 0.3704, the minimum error is 0.01593, and the maximum error is 1.27676. This is the prediction model t in the laboratory environment s =f(t i ,t o ,r,u) calculated or fitted surface temperature and measured surface temperature comparison chart, from Figure 2 It can be seen that the prediction model t s =f(t i,t o ,r,u) has high fitting accuracy.

[0061] Embodiment 2:

[0062] The refrigerator is placed in a natural environment (ambient temperature, humidity, and cabinet temperature are all changing). The duty cycle of the heating wire is precisely regulated based on the real-time data collected, and the refrigerator surface temperature is stably controlled within a narrow temperature range [t d +2,t d +2.5]; at the same time, the prediction model t s =f(t i ,t o The surface temperature calculated by the ,r,u) method is not much different from the measured surface temperature. In most cases, it is less than 0.3 degrees, and occasionally there is a difference of about 0.5 degrees.

[0063] The data in Table 2 show that: in an environment with a temperature of 25.15°C and a relative humidity of 57%, the duty cycle is 65%, and the energy saving efficiency is 35% compared with the previous heating scheme.

[0064] Table 2: Control accuracy of surface temperature

[0065]

[0066]

[0067] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0068] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments. In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A refrigerator surface anti-condensation control method based on machine learning, characterized in that: The following steps are involved: Model building phase: Establish a dynamic change prediction model for refrigerator surface temperature s ; Learning stage: Determine the refrigerator surface temperature dynamic change prediction model t s All coefficients in t are used to establish the final refrigerator surface temperature dynamic change prediction model t s ; Control stage: Control the refrigerator surface temperature within a narrow temperature range [t d +l,t d +l+0.5], where t d Indicates the dew point temperature of the current environment; l indicates the bias constant; The refrigerator surface temperature dynamic change prediction model is: t s =f(t i ,t o ,r,u) Among them, t s Indicates the surface temperature of the refrigerator; t t Indicates the internal temperature of the refrigerator; t o represents the external environment temperature; u represents the working voltage of the heating wire; r represents the heating duty cycle of the heating wire, that is, the ratio of the heating wire power-on time per unit time, 0≤r≤100%.

2. The method for controlling refrigerator surface condensation based on machine learning according to claim 1, characterized in that: The learning phase includes the following steps: Step 1.1: Determine the experimental parameters; use the refrigerator surface temperature dynamic change prediction model t s Each input parameter in is set to three levels, and an orthogonal test is used to obtain L9(3 4 )Orthogonal test table; Step 1.2: Data collection: The L9 (3 4 ) Each row of parameters in the orthogonal test table is set as the actual operating parameters of the refrigerator. After a certain period of operation in a constant temperature and humidity refrigeration laboratory, the surface temperature of the refrigerator is collected after the refrigerator reaches a balanced state; Step 1.3: Based on the collected surface temperature data Determine the model t through machine learning technology s =f(t i ,t o ,r,u), that is, to determine the prediction model of the dynamic change of the surface temperature of the refrigerator.

3. The method for controlling refrigerator surface condensation based on machine learning according to claim 1, characterized in that: The control phase includes the following steps: Step 2.1: Collect real-time data of refrigerator operation at a certain sampling frequency, including: the internal temperature of the refrigerator t i , external environment temperature t o 、Relative humidity of external environment h o And the working voltage u of the heating wire; Step 2.2: Calculate the dew point temperature; the dew point temperature is approximately determined by the Magnus-Tetens formula, that is: in, Constant a = 17.27, constant b = 237.7 °C; Step 2.3: Based on the refrigerator surface temperature dynamic change prediction model established in the learning phase, calculate the optimal heating duty cycle r of the heating wire * , the refrigerator surface temperature t s Controlled in a narrow temperature range [t d +l,t d +l+0.5], so that the surface temperature of the refrigerator is not lower than the dew point temperature, thereby preventing condensation; where l represents the bias constant; Step 2.4: Control the mainboard through the refrigerator and set the optimal heating duty cycle r * , to achieve precise heating of the heating wire.

4. The method for controlling refrigerator surface condensation based on machine learning according to claim 2, characterized in that: The L9(3 4 ) The orthogonal test table is: The setting values ​​of each parameter in the orthogonal experiment table, namely the numbers 1, 2, 3, represent the setting level of each parameter.

5. The method for controlling refrigerator surface condensation based on machine learning according to claim 2, characterized in that: The equilibrium state is the internal temperature of the refrigerator t i , external environment temperature t o , voltage u, heating duty cycle r and surface temperature t s All remain stable, that is, the change within 20 minutes does not exceed 0.

2.

6. The method for controlling refrigerator surface condensation based on machine learning according to claim 2, characterized in that: The certain period of time is 4 to 5 hours.

7. A refrigerator surface anti-condensation control method based on machine learning according to claim 1 or 2, characterized in that: The refrigerator surface temperature is the temperature at the lowest point on the refrigerator surface.

8. The method for controlling refrigerator surface condensation based on machine learning according to claim 3, characterized in that: The certain sampling frequency is 1 to 2 seconds.

Images