Refrigerator fault intelligent diagnosis method and system based on artificial intelligence

Abstract

The invention discloses a refrigerator fault intelligent diagnosis method and system based on artificial intelligence. The method comprises the steps of model training; training the model by adopting a deep learning algorithm; the method specifically comprises the steps of selecting a proper deep learning algorithm training model, constructing a network structure, defining a loss function and optimizer parameters, and performing fault prediction by using the trained model; performing model prediction; and inputting a standardized data set subjected to data preprocessing into the model, performing feature extraction and classification prediction through a deep learning algorithm, and outputting a prediction result and a maintenance suggestion of a refrigerator fault reason. According to the refrigerator fault intelligent diagnosis method and system based on artificial intelligence, technologies such as deep learning, the convolutional neural network and natural language processing are adopted, intelligent diagnosis and prediction of refrigerator faults are achieved by analyzing refrigerator fault data, the accuracy and efficiency of refrigerator fault diagnosis are improved, the labor cost is reduced, and the working efficiency is improved. And the user experience is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence and intelligent diagnosis of refrigerator faults, and in particular to an artificial intelligence-based intelligent diagnosis method and system for refrigerator faults. Background Art

[0002] Generative artificial intelligence - AIGC (Artificial Intelligence Generated Content) refers to a technology based on artificial intelligence technical methods such as generative adversarial networks and large pre-trained models, which generates relevant content with appropriate generalization capabilities through learning and identifying existing data.

[0003] The core idea of ​​AIGC technology is to use artificial intelligence algorithms to generate content with certain creativity and quality. Through training models and learning from a large amount of data, AIGC can generate relevant content based on the input conditions or guidance. For example, by entering keywords, descriptions or samples, AIGC can generate articles, images, audio, etc. that match them.

[0004] As a common refrigeration equipment in households and industrial fields, the normal operation of refrigerators is of great significance to people's lives and industrial production. However, due to long-term use, improper maintenance and other reasons, refrigerators may have various faults. Traditional refrigerator fault diagnosis mainly relies on the experience and technical level of maintenance personnel, which has certain subjectivity and limitations. With the development of artificial intelligence technology, people have begun to try to apply artificial intelligence technology to refrigerator fault diagnosis to improve diagnostic accuracy and efficiency. Summary of the invention

[0005] The purpose of the present invention is to provide an artificial intelligence-based refrigerator fault intelligent diagnosis method and system to solve the problem that refrigerator fault diagnosis in the prior art depends on the experience and technical level of maintenance personnel and has certain subjectivity and limitations.

[0006] The present invention solves the above problems through the following technical solutions:

[0007] The intelligent diagnosis method of refrigerator fault based on artificial intelligence includes:

[0008] Model training: Use deep learning algorithms to train models; specifically, select appropriate deep learning algorithm training models, build network structures, define loss functions and optimizer parameters, and use the trained models for fault prediction;

[0009] And model prediction: the standardized data set that has undergone data preprocessing is input into the model, feature extraction and classification prediction are performed through the deep learning algorithm, and the prediction results and maintenance suggestions for the cause of refrigerator failure are output.

[0010] As a further improvement, the method further includes, before model training:

[0011] Data collection: Collect the operating data and fault information of the refrigerator, as well as the historical data of maintenance records, and establish a fault case library to provide a data basis for subsequent fault prediction; for each fault case, record the fault type, fault manifestation, maintenance record and other information, and establish a corresponding data set.

[0012] Data preprocessing: Clean, label and format the collected data, convert the data into a format suitable for deep learning algorithm processing, and establish a standardized data set.

[0013] As a further improvement, the method further comprises, after data preprocessing:

[0014] User interaction: The prediction results and maintenance suggestions of the model trained by the deep learning algorithm are displayed to users, and user feedback is collected, and optimization and improvement are carried out based on user feedback.

[0015] As a further improvement, the deep learning algorithm adopted is one of a convolutional neural network and a recurrent neural network.

[0016] As a further improvement, the deep learning algorithm selects the convolutional neural network based on spatial structure characteristics, translation invariance, parameter sharing and feature extraction; and selects the recurrent neural network based on time series data, dynamic data, order sensitivity and state transfer.

[0017] As a further improvement, when the deep learning algorithm adopts a convolutional neural network, the convolutional neural network is used for feature extraction and classification tasks when processing time series data;

[0018] The network structure of the convolutional neural network includes:

[0019] Input layer: used to receive refrigerator sensor data sequence;

[0020] Convolutional layer: Use convolution kernels of different sizes to extract features at different levels;

[0021] Pooling layer: reduces the dimension of the feature map and retains key information;

[0022] Fully connected layer: converts feature maps into probabilities of fault classification;

[0023] The model prediction methods of convolutional neural networks include:

[0024] Perform convolution operation on the input data;

[0025] Perform pooling operation on the convolutional layer output;

[0026] Map the pooled features to the classification results.

[0027] As a further improvement, when the deep learning algorithm adopts a recurrent neural network, the recurrent neural network can consider the time dependency of the data when processing time series data;

[0028] The network structure of the recurrent neural network includes:

[0029] Input layer: used to receive refrigerator sensor data sequence;

[0030] Recurrent layers: These include the hidden layers of the recurrent neural network, which connect the data through time steps;

[0031] Output layer: The final output layer is used for classification or regression tasks;

[0032] Model prediction methods for recurrent neural networks include:

[0033] Hidden layer state update;

[0034] The final output of the output layer is based on the hidden state.

[0035] The refrigerator fault intelligent diagnosis system based on artificial intelligence is used to implement the above refrigerator fault intelligent diagnosis method based on artificial intelligence, including: a model training module and a model prediction module. The model training module uses a deep learning algorithm to train a model to predict the cause of refrigerator failure and provide maintenance suggestions; the model prediction module uses the trained model to predict failures.

[0036] As a further improvement, it also includes a data acquisition module and a data preprocessing module.

[0037] The data acquisition module installs sensors and monitoring equipment on the refrigerator, collects data on the operation and fault information of the refrigerator by the sensors and monitoring equipment, and collects historical data of maintenance records, and establishes a fault case library to provide a data basis for subsequent fault prediction; for each fault case, records information such as the fault type, fault manifestation, and maintenance record, and establishes a corresponding data set;

[0038] The data preprocessing module is used to clean, annotate and format the collected data, convert the data into a format suitable for deep learning algorithm processing, and establish a standardized data set.

[0039] As a further improvement, the system also includes a user interaction module, which is used to display the model prediction results and maintenance suggestions trained by the deep learning algorithm to the user, collect user feedback, and optimize and improve based on the user feedback.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] The present invention discloses an artificial intelligence-based refrigerator fault intelligent diagnosis method and system, which adopts deep learning, convolutional neural network, natural language processing and other technologies, and realizes intelligent diagnosis and prediction of refrigerator faults by analyzing refrigerator fault data, thereby improving the accuracy and efficiency of refrigerator fault diagnosis, reducing labor costs and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of an artificial intelligence-based refrigerator fault intelligent diagnosis method of the present invention;

[0043] Figure 2 The present invention is a schematic diagram of an artificial intelligence-based refrigerator fault intelligent diagnosis system. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below in conjunction with the drawings in the preferred embodiments of the present application. In the drawings, the same or similar reference numerals throughout represent the same or similar parts or parts with the same or similar functions. The described embodiments are part of the embodiments of the present application, not all of the embodiments. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0045] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0046] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, or it can be an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0047] In the description of the present application, it should be understood that the terms "upper", "lower", "front", "back", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are orientations or positional relationships based on the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0048] In addition, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product, or display that includes a series of steps or elements is not necessarily limited to those steps or elements explicitly listed but may include other steps or elements not explicitly listed or inherent to such process, method, product, or display.

[0049] The following will be combined Figure 1 , a refrigerator fault intelligent diagnosis method and system based on artificial intelligence involved in the embodiment of the present application are described in detail. It is worth noting that the following embodiments are only used to explain the present application and do not constitute a limitation on the present application.

[0050] Embodiment 1:

[0051] Combined with Figure 1 As shown, an intelligent diagnosis method for refrigerator faults based on artificial intelligence includes the following specific steps:

[0052] A. Data Collection:

[0053] Collect the operating data and fault information of the refrigerator, as well as historical data such as maintenance records. For each fault case, record the fault type, fault manifestation, maintenance record and other information, and establish a corresponding data set.

[0054] The collected data includes but is not limited to the temperature, humidity, pressure and other parameters of the refrigerator, as well as various fault manifestations of the refrigerator, such as failure to refrigerate, excessive noise, etc. In addition, it is also responsible for collecting the user's description of the refrigerator failure in order to make subsequent fault predictions and repair suggestions.

[0055] In terms of technical implementation, various sensors and monitoring equipment, such as temperature sensors, pressure sensors, sound sensors, etc., can be used to monitor the operating status of the refrigerator in real time. At the same time, a fault case library can be established by collecting historical data such as maintenance records to provide a data basis for subsequent fault prediction.

[0056] B. Data preprocessing:

[0057] Clean, label and format the collected data to establish a standardized data set. Specifically, this includes deduplication, classification, labeling, and sorting of the data to convert the data into a format suitable for deep learning algorithm processing.

[0058] Clean, label and format the collected data to establish a standardized data set. Specifically, data cleaning mainly includes removing duplicate data, processing missing values, eliminating outliers, etc.; data labeling mainly includes classifying and labeling fault types and fault manifestations; data formatting mainly includes converting data in various formats into a unified format to facilitate subsequent model training and prediction.

[0059] In terms of technical implementation, some common data processing methods can be used, such as data filtering, data labeling, data formatting, etc. At the same time, some data processing tools, such as Python, MATLAB, etc., can be used to implement data preprocessing.

[0060] C. Model training:

[0061] Use deep learning algorithms to train models. Specifically, this includes selecting appropriate deep learning algorithm training models, building network structures, defining loss functions, optimizer and other parameters, and using trained models for fault prediction.

[0062] Preferably, the model can be a convolutional neural network CNN, a recurrent neural network RNN, etc.

[0063] A deep learning algorithm is used to train a model to predict the cause of refrigerator failure and provide maintenance suggestions. Specifically, it is necessary to first select a suitable model such as a convolutional neural network (CNN) or a recurrent neural network (RNN) based on the standardized data set after data preprocessing, build a network structure, define loss functions, optimizer and other parameters; then use the trained model to predict failures and generate maintenance suggestions.

[0064] In terms of technical implementation, some common deep learning frameworks, such as TensorFlow, PyTorch, etc., can be used to implement model training. At the same time, the model needs to be optimized and adjusted to improve the prediction accuracy and efficiency of the model.

[0065] D. Model prediction:

[0066] The standardized data set that has undergone data preprocessing is input into the model, and feature extraction and classification prediction are performed through a deep learning algorithm, and the prediction results and maintenance suggestions for the cause of refrigerator failure are output.

[0067] Responsible for inputting the pre-processed standardized data set or the fault information input by the user into the model, extracting features and classifying and predicting through deep learning algorithms, and outputting the cause of the refrigerator failure and repair suggestions. Specifically, the user can input the fault information into the model through user interaction, and the model will extract features and classify and predict based on this information, and output the prediction results and repair suggestions.

[0068] In terms of technical implementation, some common deep learning algorithms, such as convolutional neural network (CNN) and recurrent neural network (RNN), can be used to generate fault prediction and maintenance suggestions. At the same time, a suitable model input interface needs to be designed so that users can easily input fault information into the model.

[0069] E. User Interaction:

[0070] The prediction results and maintenance suggestions of the model trained by the deep learning algorithm are presented to the users, and user feedback is collected. Specifically, this includes designing the user interface of the interactive system, presenting the prediction results and maintenance suggestions to the users, and receiving user feedback. The interactive system is optimized and improved based on user feedback.

[0071] Responsible for interacting with users and displaying results. Specifically, it is necessary to design the user interface of the interactive system so that users can easily input fault information into the model; at the same time, the prediction results and maintenance suggestions are displayed to users. In addition, this module also needs to collect user feedback in order to optimize and improve the interactive system.

[0072] In terms of technical implementation, some common user interface design tools, such as HTML, CSS, JavaScript, etc., can be used to design the user interface; at the same time, some display tools can be used to display the prediction results and maintenance suggestions. In addition, a suitable feedback mechanism needs to be designed to collect user feedback.

[0073] Taking refrigerator fan fault diagnosis as an example, an intelligent refrigerator fault diagnosis method based on artificial intelligence is illustrated.

[0074] Suppose a model is built to predict possible fault types and provide maintenance suggestions based on the operating status data of the wind turbine. Here, it is assumed that a sequence classification model based on a recurrent neural network (RNN) is used.

[0075] Data preprocessing: The operating status data of the refrigerator fan may be time series data, including parameters such as wind speed, temperature, and current. First, the data needs to be preprocessed by cleaning, labeling, and formatting to handle missing values, outliers, etc. Then, the data is divided into sequences of fixed length for input into the model.

[0076] Feature extraction: In this case, feature extraction may not be necessary since the raw time series data is used directly as input to the model. However, it may be necessary to normalize or standardize the data to ensure that the values ​​of different parameters are in a similar range.

[0077] Model prediction: You can build an RNN-based sequence classification model, such as a long short-term memory network (LSTM) or a gated recurrent unit (GRU). Input the processed time series data into the RNN-based sequence classification model, and the model will learn the patterns and trends in the sequence. Ultimately, the model outputs the probability distribution of each possible fault type.

[0078] When using deep learning algorithms for fault prediction, you can consider using models such as convolutional neural networks (CNN) or recurrent neural networks (RNN).

[0079] When choosing a convolutional neural network (CNN):

[0080] 1.1 Spatial structural characteristics

[0081] When the input data has obvious spatial structure characteristics, such as images, videos or audio data, CNN is the first choice. CNN effectively captures local features and spatial relationships through convolutional layers and pooling layers, and is suitable for processing two-dimensional and three-dimensional data.

[0082] 1.2 Translation Invariance

[0083] If the task is important for translation invariance, such as image recognition, object detection, etc., the local connection and weight sharing characteristics of CNN can provide effective feature extraction and recognition.

[0084] 1.3 Parameter Sharing

[0085] CNN's parameter sharing enables it to have smaller memory consumption and computational overhead when processing large-scale data, making it suitable for scenarios that need to process large amounts of data.

[0086] 1.4 Feature Extraction

[0087] When the task focuses on extracting local features from the input data, such as edges and textures in images, CNN can effectively complete this task by extracting features layer by layer and gradually combining these features.

[0088] If a convolutional neural network (CNN) is used, it is often used for feature extraction and classification tasks when processing time series data (such as sensor data). The following is the convolutional neural network structure and formula:

[0089] The network structure includes:

[0090] Input layer: Receives refrigerator sensor data sequence.

[0091] Convolutional layer: Use convolution kernels of different sizes to extract features at different levels.

[0092] Pooling layer: Reduces the dimension of the feature map and retains key information.

[0093] Fully connected layer: Converts feature maps into probabilities of fault classification.

[0094] formula:

[0095] Convolution operation: Perform convolution operation on the input data. For example, the convolution operation of the \(l\)th layer can be expressed as:

[0096] \[\text{Conv}^{(l)}=\sigma\left(\sum_{i}^{N^{(l)}}W^{(l)}_i*X^{(l-1)}+b^{(l)}\right)\];

[0097] Among them, \(W^{(l)}_i\) is the \(i\)th convolution kernel of the \(l\)th layer, \(*\) represents the convolution operation, \(\sigma\) is the activation function, and \(b^{(l)}\) is the bias term.

[0098] Pooling operation: Pooling operation is performed on the output of the convolutional layer, such as maximum pooling:

[0099] \[\text{Pool}^{(l)}=\max(X^{(l)})\];

[0100] Fully connected layer: maps the pooled features to the classification results:

[0101] \[\hat{y}=\text{softmax}(W^{(L)}\cdot\text{Pool}^{(L-1)}+b^{(L)})\];

[0102] Among them, the \(\text{softmax}\) function is used for multi-category classification, \(W^{(L)}\) is the weight of the fully connected layer, and \(b^{(L)}\) is the bias term.

[0103] When choosing a recurrent neural network (RNN):

[0104] 2.1 Time Series Data

[0105] When the input data has a time order or sequence relationship, such as text data in natural language processing and audio data in speech recognition, RNN is a more appropriate choice. RNN can process sequence inputs of variable length and can capture the temporal dependencies in the data.

[0106] 2.2 Dynamic Data

[0107] If the input data length of the task is not fixed and the output of each time step depends on the previous time step, RNN can process sequence data in a cyclic manner and is suitable for processing dynamic and continuous input-output relationships.

[0108] 2.3. Order Sensitivity

[0109] For tasks that need to consider input order or contextual information, such as machine translation and text generation, RNN can effectively utilize the information in the sequence to produce more accurate and coherent output.

[0110] 2.4. State Transfer The hidden state of RNN can transfer information between time steps, which is suitable for tasks that require memory and long-term dependence, such as language modeling, sentiment analysis, etc.

[0111] If a recurrent neural network (RNN) is used, it is suitable for processing time series data and can consider the time dependency of data. The following is the structure and formula of the recurrent neural network:

[0112] The network structure includes:

[0113] Input layer: Receives refrigerator sensor data sequence.

[0114] Recurrent layers: These comprise the hidden layers of a recurrent neural network, connecting data across time steps.

[0115] Output layer: The final output layer is used for classification or regression tasks.

[0116] formula:

[0117] Hidden layer state update: For time step \(t\), the update of the hidden state \(h_t\) can be expressed as:

[0118] \[h_t=\text{tanh}(W_{hh}h_{t-1}+W_{xh}x_t+b_h)\];

[0119] Among them, \(W_{hh}\) is the weight matrix of the hidden layer itself, \(W_{xh}\) is the weight matrix input to the hidden layer, \(b_h\) is the bias term of the hidden layer, and \(\text{tanh}\) is the hyperbolic tangent function.

[0120] Output layer: The final output \(\hat{y}_t\) can be calculated based on the hidden state, such as the Softmax layer for classification:

[0121] \[\hat{y}_t=\text{softmax}(W_{hy}h_t+b_y)\];

[0122] Among them, \(W_{hy}\) is the weight matrix connecting the hidden layer to the output layer, and \(b_y\) is the bias term of the output layer.

[0123] Loss Function and Optimizer

[0124] When choosing an optimizer, common algorithms include stochastic gradient descent (SGD), stochastic gradient descent with momentum (SGDwithmomentum), Adam (AdaptiveMomentEstimation), etc.

[0125] 1. Stochastic Gradient Descent (SGD)

[0126] Stochastic gradient descent is the most basic optimization algorithm, which updates parameters according to the negative direction of the gradient. Its update rules are as follows:

[0127] \[\theta_{t+1}=\theta_{t}-\eta\nabla_{\theta}J(\theta_t)\]

[0128] in:

[0129] -\(\theta_t\) is the model parameter after the \(t\)th iteration.

[0130] -\(\eta\) is the learning rate, which controls the update step size.

[0131] -\(\nabla_{\theta}J(\theta_t)\) is the gradient of the loss function \(J(\theta_t)\) with respect to the parameters \(\theta_t\).

[0132] 2. Stochastic Gradient Descent with Momentum (SGDwithmomentum)

[0133] Stochastic gradient descent with momentum enhances the stability and acceleration of gradient updates by introducing momentum terms, which helps to jump out of the local optimal point of the loss function and converge faster in the steep direction. Its update rules are as follows:

[0134] \[v_{t+1}=\gammav_t+\eta\nabla_{\theta}J(\theta_t)\]

[0135] \[\theta_{t+1}=\theta_{t}-v_{t+1}\]

[0136] in:

[0137] -\(v_t\) is the velocity or momentum, which is initially 0.

[0138] -\(\gamma\) is the momentum coefficient, usually set to a value between 0 and 1.

[0139] -The meanings of other symbols are the same as those of stochastic gradient descent.

[0140] 3.Adam(AdaptiveMomentEstimation)

[0141] Adam is an optimization algorithm with adaptive learning rate, which combines first-order momentum (exponential moving average of gradient) and second-order momentum (exponential moving average of squared gradient) with bias correction. Its update rule is as follows:

[0142] \[m_{t+1}=\beta_1m_t+(1-\beta_1)\nabla_{\theta}J(\theta_t)\]

[0143] \[v_{t+1}=\beta_2v_t+(1-\beta_2)(\nabla_{\theta}J(\theta_t))^2\]

[0144] \[\hat{m}_{t+1}=\frac{m_{t+1}}{1-\beta_1^{t+1}}\]

[0145] \[\hat{v}_{t+1}=\frac{v_{t+1}}{1-\beta_2^{t+1}}\]

[0146] \[\theta_{t+1}=\theta_{t}-\eta\frac{\hat{m}_{t+1}}{\sqrt{\hat{v}_{t+1}}+\epsilon}\]

[0147] in:

[0148] -\(m_t\) and \(v_t\) are estimates of momentum and squared gradient, initially 0.

[0149] -\(\beta_1\) and \(\beta_2\) are the exponential decay rates of the first-order momentum and the second-order momentum, usually set to 0.9 and 0.999 respectively.

[0150] -\(\epsilon\) is a small constant added for numerical stability (such as \(10^{-8}\)).

[0151] -\(\eta\) is the learning rate.

[0152] The Adam optimization algorithm combines the characteristics of momentum and the ability of adaptive learning rate, and can usually achieve good results in many cases.

[0153] In practical applications, these optimization algorithms are selected according to the characteristics of the problem and the distribution of the data, which can effectively accelerate the training process of the model and improve the convergence speed.

[0154] Output results: Based on the probability distribution of the model output, the fault type with the highest probability can be selected as the prediction result, and corresponding maintenance suggestions can be provided. For example, if the model predicts that the fault type is "bearing fault", it can be recommended to check whether the bearing is worn and whether it needs to be replaced.

[0155] The present invention provides an intelligent diagnosis method for refrigerator faults based on artificial intelligence, which realizes intelligent diagnosis and prediction of refrigerator faults through artificial intelligence technology, improves the accuracy and efficiency of refrigerator fault diagnosis, reduces labor costs, and improves user experience. At the same time, it can also be optimized and improved according to user feedback, and continuously improve and enhance the performance and quality of the system.

[0156] Embodiment 2:

[0157] Combined with Figure 2 As shown, an intelligent diagnosis system for refrigerator faults based on artificial intelligence is used to implement an intelligent diagnosis method for refrigerator faults based on artificial intelligence in Example 1, including: a data acquisition module, a data preprocessing module, a model training module, a model prediction module and a user interaction module connected in sequence, and these modules are usually connected in sequence according to a functional link, and the output of each module is the input of the next module. For example, the output of the data acquisition module is the input of the preprocessing module, and the output of the preprocessing module is the input of the model training module. There is a dependency relationship between the modules, that is, the output of one module is crucial to the input of the downstream module. For example, if the data acquisition module fails to provide high-quality data, the results of the preprocessing module and the model training module may be affected. The synergy between the modules is the basis for the normal operation of the system. The quality and timeliness of the data directly affect the training quality and prediction effect of the model, and the accuracy and reliability of the model directly affect the final application effect and user experience.

[0158] Data acquisition module

[0159] Sensors and monitoring equipment are installed on refrigerators to collect data such as refrigerator operation and fault information, as well as historical data such as maintenance records. For each fault case, the fault type, fault manifestation, maintenance record and other information are recorded, and a corresponding data set is established.

[0160] Data preprocessing module

[0161] Clean, label and format the collected data to establish a standardized data set. Specifically, this includes deduplication, classification, labeling, and sorting of the data to convert the data into a format suitable for deep learning algorithm processing.

[0162] Model training module

[0163] A deep learning algorithm is used to train the model to predict the cause of refrigerator failure and provide maintenance suggestions. Specifically, this module first needs to select a suitable deep learning model (such as convolutional neural network CNN, recurrent neural network RNN, etc.) based on the data set after data preprocessing, and build the network structure, define loss function, optimizer and other parameters; then use the trained model to predict failures and generate maintenance suggestions.

[0164] In terms of technical implementation, this module can use some common deep learning frameworks, such as TensorFlow, PyTorch, etc., to implement model training. At the same time, the model needs to be optimized and adjusted to improve the prediction accuracy and efficiency of the model.

[0165] Model prediction module

[0166] The deep learning algorithm is used to train the model, and the trained model is used for fault prediction. It is used to predict the cause of refrigerator failure and provide repair suggestions. The fault information entered by the user is input into the model, and the prediction results and repair suggestions are output.

[0167] User interaction module

[0168] The prediction results and maintenance suggestions are presented to the user, and user feedback is collected for interaction with the user and presentation of results.

[0169] Although the present invention is described herein with reference to the illustrative embodiments of the present invention, the above embodiments are only preferred embodiments of the present invention, and the embodiments of the present invention are not limited to the above embodiments. It should be understood that those skilled in the art can design many other modifications and embodiments, which will fall within the scope and spirit of the principles disclosed in this application.

Claims

1. An intelligent diagnosis method for refrigerator faults based on artificial intelligence, characterized in that: include: Model training: Use deep learning algorithms to train models; specifically, select appropriate deep learning algorithm training models, build network structures, define loss functions and optimizer parameters, and use the trained models for fault prediction; And model prediction: the standardized data set that has undergone data preprocessing is input into the model, feature extraction and classification prediction are performed through the deep learning algorithm, and the prediction results and maintenance suggestions for the cause of refrigerator failure are output.

2. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to claim 1 is characterized in that: The method further comprises, before model training: Data collection: Collect the operating data and fault information of the refrigerator, as well as the historical data of the maintenance records, and establish a fault case library to provide a data basis for subsequent fault prediction; for each fault case, record the fault type, fault manifestation and maintenance record information, and establish a corresponding data set; Data preprocessing: Clean, label and format the collected data, convert the data into a format suitable for deep learning algorithm processing, and establish a standardized data set.

3. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to claim 1 is characterized in that: The method further comprises, after data preprocessing: User interaction: The prediction results and maintenance suggestions of the model trained by the deep learning algorithm are displayed to users, and user feedback is collected, and optimization and improvement are carried out based on user feedback.

4. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to any one of claims 1 to 3, characterized in that: The deep learning algorithm adopted is one of a convolutional neural network and a recurrent neural network.

5. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to claim 4 is characterized in that: The deep learning algorithm selects the convolutional neural network based on spatial structure characteristics, translation invariance, parameter sharing and feature extraction; and selects the recurrent neural network based on time series data, dynamic data, order sensitivity and state transfer.

6. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to claim 5 is characterized in that: When the deep learning algorithm adopts a convolutional neural network, the convolutional neural network is used for feature extraction and classification tasks when processing time series data; The network structure of the convolutional neural network includes: Input layer: used to receive refrigerator sensor data sequence; Convolutional layer: Use convolution kernels of different sizes to extract features at different levels; Pooling layer: reduces the dimension of the feature map and retains key information; Fully connected layer: converts feature maps into probabilities of fault classification; The model prediction methods of convolutional neural networks include: Perform convolution operation on the input data; Perform pooling operation on the convolutional layer output; Map the pooled features to the classification results.

7. The refrigerator fault intelligent diagnosis method based on artificial intelligence according to claim 5 is characterized in that: When the deep learning algorithm adopts a recurrent neural network, the recurrent neural network can consider the time dependency of data when processing time series data; The network structure of the recurrent neural network includes: Input layer: used to receive refrigerator sensor data sequence; Recurrent layers: These include the hidden layers of the recurrent neural network, which connect the data through time steps; Output layer: The final output layer is used for classification or regression tasks; Model prediction methods for recurrent neural networks include: Hidden layer state update; The final output of the output layer is based on the hidden state.

8. An artificial intelligence-based refrigerator fault intelligent diagnosis system, used to implement the artificial intelligence-based refrigerator fault intelligent diagnosis method as claimed in claim 1, characterized in that: include: A model training module and a model prediction module, wherein the model training module uses a deep learning algorithm to train a model for predicting refrigerator failure causes and providing maintenance suggestions; The model prediction module uses the trained model to perform fault prediction.

9. The refrigerator fault intelligent diagnosis system based on artificial intelligence according to claim 8 is characterized in that: It also includes data acquisition module and data preprocessing module. The data acquisition module installs sensors and monitoring equipment on the refrigerator, collects data on the operation and fault information of the refrigerator by the sensors and monitoring equipment, and collects historical data of maintenance records, and establishes a fault case library to provide a data basis for subsequent fault prediction; For each fault case, record the fault type, fault manifestation, maintenance record and other information, and establish a corresponding data set; The data preprocessing module is used to clean, annotate and format the collected data, convert the data into a format suitable for deep learning algorithm processing, and establish a standardized data set.

10. The refrigerator fault intelligent diagnosis system based on artificial intelligence according to claim 8 or 9, characterized in that: The system also includes a user interaction module, which is used to display the model prediction results and maintenance suggestions trained by the deep learning algorithm to the user, collect user feedback, and optimize and improve according to the user feedback.

Images