Refrigerator defrosting control method based on BP neural network and fuzzy control coupling

Abstract

The invention discloses a refrigerator defrosting control method based on BP (Back Propagation) neural network and fuzzy control coupling, which is characterized by comprising the following steps: realizing deep perception of dynamic characteristics of a refrigerating system through data-driven modeling: learning actual data signals of a compressor and an evaporator by the BP neural network, and indirectly estimating recessive parameters such as refrigerant charge and frost layer form distribution; and the fuzzy controller dynamically adjusts a defrosting strategy according to a neural network prediction result, and collaborative optimization of refrigeration cycle parameters and the frosting process is achieved. Environment parameters are learned online through a neural network, and the fuzzy rule base is optimized in real time; the neural network directly maps bottom layer physical characteristics such as a compressor operation mode and evaporator structure parameters, and outputs a dynamic compensation signal; and the model is continuously iterated by accumulating historical data to adapt to regional / seasonal differences.

Description

Technical Field

[0001] The present invention relates to an intelligent control technology for refrigerators, in particular to a refrigerator defrosting control method based on BP neural network and fuzzy control coupling. Background Art

[0002] When it comes to food preservation and refrigeration / freezing functions in household refrigerators, evaporator frost is a key factor affecting performance. Due to high humidity inside the refrigerator and the evaporator surface temperature being below the dew point, water vapor condenses on the evaporator surface, forming a frost layer with poor thermal conductivity over time. This not only reduces heat exchange efficiency and affects food preservation, but also causes wear and tear on evaporator components, increases energy consumption, and shortens the life of the equipment due to long-term frost.

[0003] Existing intelligent defrosting technologies utilize fuzzy control to achieve adaptive regulation of nonlinear, multivariable, and complex systems. They can dynamically optimize defrost strategies based on parameters such as temperature and humidity, offering greater precision than traditional timed control. For example, patent publication number CN115371338B describes a refrigerator defrost control method. The method first calculates and corrects the initial time step for a step-by-step power reduction based on parameters such as compressor operating time, initial defrost power, and ambient temperature. The method then gradually reduces the defrost power, controlling the time step by setting the temperature rise amplitude of the defrost sensor. Finally, the power and temperature rise amplitude in subsequent time steps are continuously adjusted based on the operating time and temperature rise rate within the previous time step, continuously performing a step-by-step defrost reduction of the electric heater power until the set conditions are met and defrost mode is exited. Another example is patent publication number CN105004127B, which proposes a refrigerator defrost control method. A temperature and humidity sensor is used to monitor the temperature and humidity around the evaporator. The sensor transmits the detected temperature and humidity signals to a control board, which compares the received temperature and humidity signals with preset values to control the on / off of the heater and compressor.

[0004] The limitation of these fuzzy controls in the disclosed technology in refrigerator defrost control is that they can only passively respond to the surface parameters of evaporator frosting, such as frost thickness and temperature changes, but lack the ability to directly control the underlying dynamic characteristics of the refrigeration system.

[0005] These deep-seated influencing factors include: refrigeration cycle parameters (refrigerant pressure, flow rate and throttling device status), compressor operating characteristics (speed, efficiency and start-stop frequency), evaporator physical structure (fin spacing, material thermal conductivity), system ventilation conditions (fan speed, air duct resistance), etc. Traditional fuzzy control relies on a preset rule base, which makes it difficult to capture the coupling relationship between refrigerant phase change and heat and mass transfer in the frosting process, and it is also unable to compensate in real time for changes in fin gap caused by compressor performance degradation or evaporator frosting. Therefore, it is necessary to introduce a control method that can break through the limitations of single fuzzy control experience rules and significantly improve defrosting accuracy and system energy efficiency under complex working conditions. Summary of the Invention

[0006] The purpose of the present invention is to solve the above problems and provide a refrigerator defrosting control method based on BP neural network coupled with fuzzy control, which has the characteristics of dynamic adaptability, bottom-level control capability, multi-objective optimization, data-driven characteristics, etc.

[0007] The above technical problems of the present invention are mainly solved by the following technical solutions: a refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, characterized by: achieving deep perception of the dynamic characteristics of the refrigeration system through data-driven modeling: the BP neural network learns the actual data signals of the compressor and evaporator to indirectly estimate implicit parameters such as the refrigerant charge amount and the frost layer morphology distribution; the fuzzy controller dynamically adjusts the defrost strategy based on the neural network prediction results, achieving coordinated optimization of the refrigeration cycle parameters and the frosting process, including the following contents:

[0008] S1-data acquisition and preprocessing; S2-construction of BP neural network prediction model for refrigerator defrost control; S3-BP neural network model training; S4-BP neural network model performance evaluation; S5-construction of BP neural network and fuzzy control coupling model; S6-weight adjustment of fuzzy control rule base; S7-dynamic adjustment of refrigerator defrost; S8-construction of comprehensive defrost evaluation function.

[0009] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, preferably, S1-data acquisition and preprocessing: acquiring refrigerator operation data, cleaning and normalizing the data, and dividing the data into a training set, a validation set, and a test set;

[0010] Normalization is to scale the data to a specified range: [0,1] or [-1,1]; the formula is:

[0011]

[0012] Where: x′ is the normalized data, x is the input feature; min(x) is the minimum value of the feature; max(x) is the maximum value of the feature;

[0013] Use the training set to train the model through continuous iterative backpropagation and parameter update processes; use the validation set to adjust the model's hyperparameters; and use the test set to evaluate the model's performance.

[0014] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, the refrigerator operation data includes: evaporator temperature, ambient temperature and humidity, door switch frequency, historical defrost records, refrigeration cycle parameters, compressor operation characteristics, evaporator physical structure parameters, and system ventilation condition parameters.

[0015] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, as an optimal method, S-2, constructing a BP neural network prediction model for refrigerator defrost control: the BP neural network consists of an input layer, a hidden layer and an output layer, that is, constructing a BP neural network model with evaporator temperature, environmental parameters, and user behavior as inputs, and with frost layer thickness prediction value and defrost time as outputs.

[0016] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, the ReLU activation function and Adam optimizer are used for training the BP neural network model in S-3. In the mathematical expression of the ReLU activation function, for the input value x, its output y can be expressed by the following piecewise function:

[0017] y = ReLU(x) = max(0, x)

[0018] In the aforementioned refrigerator defrost control method based on coupling of BP neural network and fuzzy control, as a preferred embodiment, when evaluating the performance of S-4 and BP neural network models, the model hyperparameters are adjusted using a grid search algorithm and the mean square error (MSE) model is used for evaluation. The mean square error is a loss function used to measure the difference between the predicted output of the neural network and the true label. The mean square error is represented by σ:

[0019]

[0020] Where: n is the number of input layer nodes, x i is the predicted output of the i-th sample, is the average.

[0021] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, S-5, when constructing the BP neural network and fuzzy control coupling model, a fuzzy control rule base is established, the neural network prediction results are converted into fuzzy language variables, and the defrost heating power and duration are determined through fuzzy reasoning.

[0022] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, S6-fuzzy control rule base weight adjustment is preferred. According to the actual needs and experience of refrigerator defrosting, the output of the BP neural network prediction model is used as the input of the fuzzy control rule model, and the fuzzy rule weight is dynamically adjusted through online learning to form a closed-loop optimization system; the coupling model refers to a collaborative decision-making system formed by the neural network prediction module and the fuzzy control module through data interaction.

[0023] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, as a preferred method, S7-when dynamically adjusting the refrigerator defrost, statistics on the accumulated defrost data and comparison of the states at adjacent moments are the key to the refrigerator defrost system switching from passive response to active optimization.

[0024] In the aforementioned refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, as a preferred step, S8-constructs a comprehensive defrost evaluation function, including constructing a comprehensive evaluation function including defrost efficiency, energy consumption, and temperature fluctuation, inputs a correction instruction to calculate the evaluation index at time t, outputs a control signal if the index at time t is better than that at time t-1, otherwise retrains the model;

[0025] Construct a comprehensive evaluation function J:

[0026] J=αE+βΔT+γD

[0027] Where, E: defrost energy consumption (Wh), calculated in real time by the current sensor; ΔT: temperature fluctuation amplitude inside the defrost device refrigerator (°C); D: residual frost thickness (mm), obtained by weighing or laser ranging; α, β, γ: weight coefficients;

[0028] Optimization goal: minimize J, that is:

[0029] J * =min(αE+βΔT+γD)

[0030] This technical solution leverages a BP neural network (back propagation neural network) coupled with fuzzy control technology to achieve deep perception of the refrigeration system's dynamic characteristics through data-driven modeling. The BP neural network learns signals such as compressor current and the evaporator inlet and outlet temperature difference, indirectly estimating implicit parameters such as refrigerant charge and frost layer distribution. The fuzzy controller dynamically adjusts the defrost strategy based on the neural network's predictions, achieving coordinated optimization of refrigeration cycle parameters and the frosting process. This composite control architecture transcends the empirical rule limitations of single fuzzy control and significantly improves defrost accuracy and system energy efficiency under complex operating conditions.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] (1) Dynamic adaptability: Through online learning of environmental parameters such as temperature and humidity, door opening frequency and the nonlinear relationship between frost rate, the fuzzy rule base is optimized in real time.

[0033] (2) Bottom-level control capability: The neural network directly maps the underlying physical characteristics such as the compressor operating mode and evaporator structural parameters, and outputs dynamic compensation signals, such as adjusting the opening of the throttling device.

[0034] (3) Multi-objective optimization: Simultaneously optimize multi-dimensional indicators such as defrosting efficiency, reducing energy consumption, and improving component life.

[0035] (4) Data-driven features: Continuously iterate the model by accumulating historical data to adapt to regional / seasonal differences. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a general flow chart of the present invention.

[0037] Figure 2 This is a flow chart for constructing a BP neural network prediction model of the present invention. DETAILED DESCRIPTION

[0038] The technical solution of the present invention will be further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0039] This embodiment provides a refrigerator defrost control method based on the coupling of BP neural network and fuzzy control, which achieves deep perception of the dynamic characteristics of the refrigeration system through data-driven modeling: the BP neural network learns the actual data signals of the compressor and evaporator, and indirectly estimates implicit parameters such as the refrigerant charge amount and the frost layer morphology distribution; the fuzzy controller dynamically adjusts the defrost strategy according to the neural network prediction results, thereby achieving coordinated optimization of the refrigeration cycle parameters and the frosting process.

[0040] Specifically, by obtaining parameters such as evaporator temperature, ambient temperature and humidity, door switch frequency, historical defrost records, refrigeration cycle parameters, compressor operating characteristics, evaporator physical structure parameters, and system ventilation conditions, a BP neural network prediction model for refrigerator defrost control is constructed. The data is cleaned, preprocessed, and normalized, and the data set is divided for training. Then, the grid search algorithm is used to adjust the model hyperparameters, and the mean square error (MSE) model is used to evaluate the performance of the BP neural network model. The frost layer thickness prediction value and defrost time predicted by the neural network are used as the input of fuzzy control, and the defrost strategy - defrost start time, defrost heating power, and defrost interval - is dynamically and actively adjusted through fuzzy rules.

[0041] Including the following content, such as Figure 1 As shown:

[0042] S1. Data acquisition and preprocessing

[0043] The refrigerator operation data is obtained, including evaporator temperature, ambient temperature and humidity, door switch frequency, historical defrost records, refrigeration cycle parameters (refrigerant pressure, flow and throttling device status), compressor operation characteristics (speed, efficiency and start-stop frequency), evaporator physical structure parameters (fin spacing, material thermal conductivity), system ventilation condition parameters (fan speed, air duct resistance), etc. The data is cleaned and normalized, and divided into training set, validation set and test set.

[0044] Normalization is to scale the data to a specified range: [0,1] or [-1,1]; the formula is:

[0045]

[0046] Where: x′ is the normalized data, x is the input feature; min(x) is the minimum value of the feature; max(x) is the maximum value of the feature.

[0047] Dataset Splitting: Use the training set to train the model, minimizing the loss function through iterative backpropagation and parameter updates. Use the validation set to adjust the model's hyperparameters. Use the test set to evaluate the model's performance.

[0048] S2. Constructing a BP neural network prediction model for refrigerator defrosting control

[0049] The BP neural network consists of an input layer, a hidden layer, and an output layer. This constructs a BP neural network model that takes evaporator temperature, environmental parameters, and user behavior as inputs and outputs frost thickness predictions and defrost time. The steps for constructing a BP neural network are as follows:

[0050] (1) Network initialization: determine the system input (X, Y), the number of nodes in the input layer, hidden layer, and output layer n, l, m, and initialize the connection weights ω between neurons in each layer ij and ω jk , given the learning rate and neuron activation function, and initialize the thresholds a, b of the hidden layer and output layer.

[0051] (2) Hidden layer output calculation:

[0052]

[0053] Where: X: input variable; ω ij : connection weight; a: threshold; H: hidden layer output.

[0054] f: hidden layer activation function, its expression is as follows:

[0055]

[0056] Where e is a natural constant.

[0057] (3) Output layer calculation:

[0058]

[0059] Where: jk : connection weight; b: threshold; O: output layer output.

[0060] S3. Use BP neural network model to train the collected data

[0061] The ReLU activation function and Adam optimizer are used for training. In the mathematical expression of the ReLU activation function, for an input value x, its output y can be expressed as the following piecewise function:

[0062] y = ReLU(x) = max(0, x)

[0063] Adam optimization combines the ideas of momentum and adaptive learning rate. It simultaneously calculates the exponentially weighted average of the first-order moment (mean) and second-order moment (variance) of the gradient and uses this information to adjust the learning rate. It can adaptively adjust the learning rate for different parameters and performs well when dealing with sparse gradients. Figure 2 The prediction model construction process of the BP neural network shown in FIG.

[0064] S4: Use grid search algorithm to adjust model hyperparameters and use mean square error (MSE) model to evaluate BP neural network model performance

[0065] The mean squared error is a loss function that measures the difference between the predicted output of a neural network and the true label. The mean squared error is generally represented by σ:

[0066]

[0067] Where: n: number of input layer nodes; x i : The predicted output of the i-th sample; Average.

[0068] S5. Establish a fuzzy control rule base to form a BP neural network and fuzzy control coupling model

[0069] A fuzzy control rule base is established to convert the neural network prediction results into fuzzy language variables, such as "high frost layer" and "low humidity", and the defrost heating power and duration are determined through fuzzy reasoning.

[0070] Fuzzy control is an intelligent control method based on fuzzy set theory that transforms expert knowledge and experience into fuzzy rules. Fuzzy control rules contain the following logic:

[0071] IF predicted frost thickness > threshold AND ambient humidity is high THEN extend defrost time;

[0072] IF the door switching frequency is high THEN shorten the defrost interval.

[0073] Defuzzification is performed using the center of gravity method to output specific control parameters.

[0074] S6. Fuzzy control rule base weight adjustment

[0075] Based on the actual needs and experience of refrigerator defrosting, the output of the BP neural network prediction model is used as the input to the fuzzy control rule model. The fuzzy rule weights are dynamically adjusted through online learning to form a closed-loop optimization system. A coupled model is a collaborative decision-making system formed by the interaction of data between the neural network prediction module and the fuzzy control module.

[0076] S7. Dynamic adjustment strategy for refrigerator defrosting based on accumulated data

[0077] Counting the accumulated defrost data and comparing the status at adjacent moments is the key to transforming the refrigerator defrost system from "passive response" to "active optimization".

[0078] By collecting real-time statistics on the cumulative duration and number of defrost cycles, and comparing defrost parameters (such as heating power and duration) with operating conditions (such as compressor start and stop frequency) at adjacent times (t and t-1), the system automatically triggers policy adjustments when the cumulative value exceeds a preset threshold (such as more than three defrosts per day or a duration exceeding two hours) or detects an abnormality (such as a sensor failure or abnormally thick frost layer). Adjustments include extending the defrost interval, increasing heating power, and sending maintenance reminders, forming a closed-loop control mechanism to optimize defrost efficiency, reduce energy consumption, and ensure safe equipment operation.

[0079] S8. Constructing a comprehensive defrost evaluation function

[0080] The comprehensive defrosting evaluation function includes building a comprehensive evaluation function including defrosting efficiency, energy consumption, and temperature fluctuation. The correction instruction is input to calculate the evaluation index at time t. If the index at time t is better than that at time t-1, the control signal is output, otherwise the model is retrained. Constructing the comprehensive evaluation function J:

[0081] J=αE+βΔT+γD

[0082] Where, E: defrost energy consumption (Wh), calculated in real time by the current sensor;

[0083] ΔT: Temperature fluctuation range inside the defrost device refrigerator (°C);

[0084] D: residual thickness of frost layer (mm), obtained by weighing or laser ranging;

[0085] α, β, γ: weight coefficients.

[0086] Optimization goal: minimize J, that is:

[0087] J * =min(αE+βΔT+γD)

[0088] The above embodiments are intended to illustrate the present invention, not to limit it. The embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments derived from the present invention by persons of ordinary skill in the art without inventive effort shall fall within the scope of protection of the present invention.

Claims

1. A refrigerator defrosting control method based on BP neural network coupled with fuzzy control, characterized by: Data-driven modeling enables deep understanding of the dynamic characteristics of the refrigeration system. A BP neural network learns actual data signals from the compressor and evaporator to indirectly estimate implicit parameters such as refrigerant charge and frost layer distribution. A fuzzy controller dynamically adjusts the defrost strategy based on the neural network's predictions, achieving coordinated optimization of refrigeration cycle parameters and the frosting process. This includes the following: S1-data acquisition and preprocessing; S2-construction of BP neural network prediction model for refrigerator defrost control; S3-BP neural network model training; S4-BP neural network model performance evaluation; S5-construction of BP neural network and fuzzy control coupling model; S6-weight adjustment of fuzzy control rule base; S7-dynamic adjustment of refrigerator defrost; S8-construction of comprehensive defrost evaluation function.

2. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 is characterized in that: S1-Data acquisition and preprocessing: Obtain refrigerator operation data, clean and normalize the data, and divide it into training set, validation set, and test set; Normalization is to scale the data to a specified range: [0,1] or [-1,1]; the formula is: Where: x′ is the normalized data, x is the input feature; min(x) is the minimum value of the feature; max(x) is the maximum value of the feature; Use the training set to train the model through continuous iterative backpropagation and parameter update processes; use the validation set to adjust the model's hyperparameters; and use the test set to evaluate the model's performance.

3. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 2, characterized in that: Refrigerator operation data: including evaporator temperature, ambient temperature and humidity, door switch frequency, historical defrost records, refrigeration cycle parameters, compressor operation characteristics, evaporator physical structure parameters, and system ventilation condition parameters.

4. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 is characterized in that: S-2. Construct a BP neural network prediction model for refrigerator defrost control: The BP neural network consists of an input layer, a hidden layer, and an output layer. That is, a BP neural network model is constructed with evaporator temperature, environmental parameters, and user behavior as inputs, and frost thickness prediction value and defrost time as outputs.

5. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 is characterized in that: When training the BP neural network model in S-3, the ReLU activation function and the Adam optimizer are used for training. In the mathematical expression of the ReLU activation function, for the input value x, its output y can be expressed by the following piecewise function: y=ReLU(x)=max(0,x).

6. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 is characterized in that: S-4. When evaluating the performance of the BP neural network model: use the grid search algorithm to adjust the model hyperparameters and use the mean square error (MSE) model for evaluation. The mean square error is a loss function that measures the difference between the predicted output of the neural network and the true label. The mean square error is represented by σ: Where: n is the number of input layer nodes, x i is the predicted output of the i-th sample, is the average.

7. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 is characterized in that: S-5. When constructing the BP neural network and fuzzy control coupling model, a fuzzy control rule base is established, the neural network prediction results are converted into fuzzy language variables, and the defrosting heating power and duration are determined through fuzzy reasoning.

8. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1 or 7, characterized in that: S6-Fuzzy control rule base weight adjustment: Based on the actual needs and experience of refrigerator defrosting, the output of the BP neural network prediction model is used as the input of the fuzzy control rule model, and the fuzzy rule weight is dynamically adjusted through online learning to form a closed-loop optimization system; the coupling model refers to a collaborative decision-making system formed by the neural network prediction module and the fuzzy control module through data interaction.

9. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1, characterized in that: S7- When dynamically adjusting the refrigerator defrost, statistics on the accumulated defrost data and comparison of the status at adjacent moments are the key to the refrigerator defrost system's transition from passive response to active optimization.

10. The refrigerator defrosting control method based on BP neural network and fuzzy control coupling according to claim 1, characterized in that: S8-Build a comprehensive defrosting evaluation function, including building a comprehensive evaluation function including defrosting efficiency, energy consumption, and temperature fluctuation. Input the correction instruction to calculate the evaluation index at time t. If the index at time t is better than that at time t-1, output the control signal, otherwise retrain the model; Construct a comprehensive evaluation function J: J=αE+βΔT+γD Where, E: defrost energy consumption (Wh), calculated in real time by the current sensor; ΔT: temperature fluctuation amplitude inside the defrost device refrigerator (°C); D: residual frost thickness (mm), obtained by weighing or laser ranging; α, β, γ: weight coefficients; Optimization goal: minimize J, that is: J * = min(αE+βΔT+γD).

Images