Predictive maintenance optimization method of intelligent household electrical appliance based on artificial intelligence technology

Abstract

The invention discloses a predictive maintenance optimization method for an intelligent household appliance based on an artificial intelligence technology, and the method comprises the steps: collecting the data of the household appliance, dividing the data into a training set and a test set, and carrying out the preprocessing; training a predictive maintenance optimization model of the intelligent household electrical appliance by using the training set by using a gradient boosting decision tree GBDT algorithm; applying the trained GBDT model to a test set, and evaluating the prediction accuracy and generalization ability of the model by calculating an evaluation index mean absolute error MAE and a decision coefficient R2; the feature importance of the model is analyzed, the contribution degree of each feature to a prediction result is known, and the structure and parameters of the model are optimized according to an evaluation result and feature importance analysis; and deploying the optimized GBDT model into the intelligent household electrical appliance. According to the innovative solution based on the GBDT model, the intelligent level and the maintenance efficiency of household appliance management are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of predictive maintenance optimization of smart home appliances, and in particular to a predictive maintenance optimization method for smart home appliances based on artificial intelligence technology. Background Art

[0002] With the rapid development of the Internet of Things, artificial intelligence and big data technologies, the field of smart home appliances has ushered in unprecedented changes. The maturity and popularization of the Internet of Things technology has built a network foundation for the interconnection of home appliances, making remote control and real-time monitoring possible. The in-depth application of artificial intelligence technology, especially machine learning and deep learning technology, has given smart home appliances the ability to self-learn and adapt, and can accurately identify user habits, automatically adjust working parameters, and realize intelligent control and optimization of home life. At the same time, big data technology provides strong data support for smart home appliances. Through in-depth mining and analysis of massive data, it realizes the accurate push of personalized services and intelligent experience.

[0003] As the core component of the household electricity management system, smart home appliances not only bear the heavy responsibility of intelligent control and automation, but also actively respond to the needs of the power grid to achieve the optimal distribution and consumption of energy. They can automatically adjust the working mode according to the user's living habits and the real-time status of the power grid to promote energy conservation and emission reduction; through real-time monitoring and reporting of energy usage, they help users to scientifically manage energy consumption; and with the help of secure encryption technology, they ensure the security and privacy protection of user data.

[0004] Although smart home appliance technology has made significant progress, there are still technical gaps and challenges in equipment status monitoring, fault prediction, energy efficiency evaluation and maintenance optimization. Current research focuses on risk monitoring, but rarely applies artificial intelligence technology to the status modeling, fault prediction, energy efficiency evaluation and maintenance demand prediction of smart home appliances. In particular, the construction of the relationship model between the characteristic parameters of the equipment and the above key indicators, as well as the in-depth analysis of the impact of these parameters on equipment performance and maintenance strategies, is still insufficient.

[0005] Therefore, it is particularly important to build an accurate relationship model to analyze the intrinsic relationship between equipment characteristic parameters and their status, maintenance requirements, fault prediction and energy efficiency evaluation, so as to provide scientific basis and technical support for the intelligent management, fault warning and efficient maintenance of smart home appliances, and further improve the convenience, comfort and intelligence level of home life. Summary of the invention

[0006] In view of this, the present invention proposes a predictive maintenance optimization method for smart home appliances based on artificial intelligence technology to solve the limitations of smart home appliance AI applications in the prior art and the problem that fault prediction and maintenance optimization urgently need to be strengthened.

[0007] The specific technical solutions of the present invention are as follows:

[0008] The predictive maintenance optimization method of smart home appliances based on artificial intelligence technology includes:

[0009] Step 1: Collect the temperature, humidity, voltage, current, usage time, working mode, user operation record, equipment status, maintenance requirements, fault prediction and energy efficiency evaluation data of household appliances, divide the collected data into training set and test set, and perform preprocessing, including checking and processing missing values, to ensure the integrity and accuracy of the data;

[0010] Step 2: Using the gradient boosting decision tree (GBDT) algorithm and the training set, multiple decision trees are constructed through iteration. Each decision tree is fitted on the residual of the previous tree, and the prediction accuracy of the model is gradually improved until the preset number of iterations is reached or other stopping conditions are met, thereby training a predictive maintenance optimization model for smart home appliances.

[0011] Step 3: Apply the trained GBDT model to the test set and evaluate the prediction accuracy and generalization ability of the model by calculating the mean absolute error (MAE) and the coefficient of determination (R2).

[0012] Step 4: Analyze the feature importance of the model to understand the contribution of each feature to the prediction results, and optimize the model structure and parameters based on the evaluation results and feature importance analysis to improve the prediction performance and interpretability of the model.

[0013] Furthermore, the method also includes: step 5, deploying the optimized GBDT model to smart home appliances, timely discovering potential risks by real-time monitoring of the operating status and performance of the equipment, and collecting user feedback and actual equipment operation data to continuously monitor and optimize model performance and achieve predictive maintenance optimization of smart home appliances.

[0014] Furthermore, the GBDT algorithm in step 2 uses the K-fold cross-validation method to determine the optimal hyperparameter combination.

[0015] Furthermore, step 4 uses visualization tools to display the decision path and feature importance of the model, improves the interpretability of the model, and helps users better understand the model prediction results.

[0016] Furthermore, the independent variables of the data set in step 1 include the current temperature of the home appliance, the humidity level in the environment, the current consumption of the home appliance, the voltage input of the home appliance, the operating time of the home appliance, the current working mode of the home appliance, and the user's operation record of the home appliance; the dependent variables of the data set focus on the operating status, maintenance requirements, fault prediction and energy efficiency evaluation of the home appliance.

[0017] Furthermore, in step 2, the parameters involved in the training process of the GBDT model include the number of decision trees, the depth of the decision trees, and the learning rate.

[0018] Furthermore, step 5 also includes: setting a model adaptive update module; the model adaptive update module can automatically adjust and optimize model parameters according to newly collected user feedback and actual device operation data to adapt to changes in device performance and changes in user needs.

[0019] Furthermore, step 4 adopts a feature contribution evaluation method to determine which features have the greatest impact on the prediction results of the model.

[0020] Furthermore, step 4 adopts To calculate the global importance of feature j, and Respectively represent the global importance of feature j and the importance of feature j in subtree T n The local importance of , and N represents the number of subtrees.

[0021] Furthermore, in step 4, Define the local importance of feature j in a single subtree, which is determined based on the reduction in the residual sum of squares when the node in the subtree is split and whether the feature is selected as the basis for splitting; where M is the subtree T n The number of leaf nodes; is the reduction of the residual sum of squares of non-leaf nodes after t split; I(v t =j) is the discriminant function. When the feature that the non-leaf node depends on at t split is feature j, the function value is 1, otherwise, the function value is 0.

[0022] The beneficial effects of the present invention are:

[0023] 1. Construction of high-precision prediction model: By constructing a relational model based on the GBDT model, the present invention successfully establishes a close connection between the characteristic parameters of household appliances and the equipment status, maintenance requirements, fault prediction and energy efficiency evaluation. With its high accuracy and powerful prediction capabilities, the model provides users with intelligent equipment management tools, effectively improving the reliability and overall performance of the equipment. At the same time, the data-driven approach makes household appliance risk monitoring and prediction more accurate, thereby promoting the dual improvement of household appliance efficiency and lifespan, and contributing to the realization of sustainable development goals.

[0024] 2. Improvement of intelligent safety management: This invention deeply analyzes the importance of characteristic parameters of household appliances in evaluating equipment status, predicting maintenance needs, failure risks and energy efficiency performance, and provides a scientific basis for the safety management of intelligent household appliances. By comprehensively considering the indicators of each characteristic dimension, this invention can fully reveal the influencing factors of household appliance risk scores, and propose targeted improvement measures accordingly, effectively prevent potential safety hazards, and significantly improve the safety management level of intelligent household appliances.

[0025] 3. Comprehensive consideration of multi-dimensional characteristic parameters: Different from traditional methods, this invention innovatively considers the characteristic parameters of different dimensions of household appliances at the same time and constructs a comprehensive and accurate relationship model. This model can deeply capture the multi-faceted characteristics of equipment operation, thereby achieving accurate prediction of equipment status, maintenance requirements, failure risks and energy efficiency performance. This method of comprehensively considering characteristic parameters of different dimensions not only provides users with more comprehensive and personalized equipment management and maintenance services, but also greatly improves the overall performance of the equipment and user experience, and promotes the management and maintenance of household appliances towards intelligence and personalization. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application, 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 only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0027] Figure 1 The present invention is a flowchart of a method for optimizing predictive maintenance of smart home appliances based on artificial intelligence technology. DETAILED DESCRIPTION

[0028] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0029] The present invention proposes a predictive maintenance optimization method for smart home appliances based on artificial intelligence technology. By building a highly accurate prediction model, it realizes real-time monitoring and proactive maintenance of key information such as the status and fault conditions of smart home appliances. This method fully utilizes the advantages of artificial intelligence technology and promotes the management of smart home appliances towards intelligence and efficiency. It not only significantly improves the performance and reliability of the equipment, but also effectively extends the service life of the equipment, bringing users an unprecedented convenient and smart home life experience.

[0030] Specifically, the implementation process of the present invention first focuses on the careful preparation of a data set, which covers a wide range of multi-dimensional operating information of household appliances, including but not limited to temperature, humidity, voltage stability, usage time, specific working mode, user operation habit records, current device status, historical maintenance needs, fault prediction labels, and energy efficiency evaluation results, etc. This comprehensive and meticulous data collection work has laid a solid foundation for the construction of subsequent models.

[0031] Subsequently, the data set was scientifically divided into training and test sets to ensure that the model can fully learn the data features during the training process and accurately reflect its generalization ability in the subsequent evaluation stage. Next, the importance of data preprocessing is self-evident. By reasonably filling in missing values, identifying and processing outliers, and standardizing or normalizing data, the data quality is effectively improved, providing a strong guarantee for the accurate training of the model.

[0032] In the model building stage, the GBDT algorithm stood out with its powerful integrated learning ability. By iteratively building multiple decision trees, each tree is committed to further fitting and optimizing based on the prediction residual of the previous tree, thereby achieving continuous improvement in the model prediction accuracy. At the same time, through fine-tuning of key hyperparameters such as the number of trees, depth, and learning rate, the performance of the model is further optimized, ensuring its efficient performance in complex household appliance status prediction tasks.

[0033] Finally, the trained GBDT model was fully evaluated using the test set to verify its generalization ability on unseen data. The test results show that the model can accurately predict the state change trend, potential failure risk and energy efficiency performance of home appliances, providing strong data support for the management and maintenance of smart home appliances.

[0034] It is worth mentioning that after the model training is completed, the feature importance analysis function can be used to further explore the contribution of each feature to the prediction result. This function not only helps smart home appliance users and builders to more clearly understand the internal laws of device operation, but also provides important directional guidance for them to timely discover potential risks and take targeted measures, further improving the overall operating efficiency and user experience of the smart home system.

[0035] The specific steps of the method of the present invention are as follows:

[0036] Step 1: Data collection and construction of predictive maintenance optimization model for smart home appliances.

[0037] In the predictive maintenance optimization process of smart home appliances based on the gradient boosting decision tree (GBDT) technology, it is first necessary to build a comprehensive data set that covers a variety of operating parameters of home appliances and their corresponding maintenance status information. Specifically, the independent variables of these data sets include the current temperature of the home appliance, the humidity level in the environment (i.e., the humidity of the home appliance), the current consumption of the home appliance, the voltage input of the home appliance, the operating time of the home appliance, the current working mode of the home appliance, and the user's operation record of the home appliance. These parameters fully reflect the operating status of the equipment and the external environmental conditions, providing rich input features for the model.

[0038] On the other hand, the dependent variables of the data set focus on the operating status, maintenance needs, fault prediction and energy efficiency evaluation of household appliances. The operating status of the equipment is divided into different levels such as normal, abnormal, and faulty, which are used to evaluate the immediate health status of the equipment; maintenance needs directly indicate whether the equipment needs further maintenance or repair operations; fault prediction further refines the possible types of faults and the expected time of occurrence, providing an important basis for preventive maintenance; and energy efficiency evaluation quantitatively evaluates the energy efficiency of the equipment through indicators such as energy consumption.

[0039] By using these parameters as input features of the GBDT model, we can use the powerful learning ability of the GBDT algorithm to train a model that can accurately predict the status and performance of the equipment. This model can quickly identify potential failure risks based on the real-time operation data of the equipment and plan maintenance plans in advance, thereby achieving optimized operation and predictive maintenance of smart home appliances.

[0040] The specific algorithm steps are as follows:

[0041] Input: A training set containing data such as temperature, humidity, current, voltage, usage time, working mode, user operation records of household appliances, as well as corresponding label information such as device status, maintenance requirements, fault prediction, energy efficiency evaluation, etc., and a loss function is specified to evaluate the accuracy of the model prediction.

[0042] Input: training set T = {(x1, y1), (x2, y2), ..., (x N ,y N )},x i＝1，……，N ∈R n ,y i ∈R, where R n represents n-dimensional real number space; R represents one-dimensional real number space; x i A vector composed of values ​​of temperature, humidity, current, voltage, usage time, working mode, user operation records, etc. of household appliances (ie: x iIt represents a vector, that is, an independent variable. The value of each dimension in the vector represents the value of the temperature, humidity, current, voltage, usage time, working mode, user operation record and other data of the home appliance). i For x i The corresponding equipment status / maintenance requirements / fault prediction / energy efficiency evaluation is as follows. L(y,f(x)) is the loss function, which will be used in subsequent steps.

[0043] Output: Predictive maintenance optimization model for smart home appliances

[0044] Step 11: Initialization c represents the value of the residual that needs to be fitted (similar to linear regression). The meaning of this formula is to choose the residual value that minimizes the total loss, that is, to traverse all samples and estimate a better c so that the total loss calculated for all samples on c is minimized, that is, the total residual fit is the best.

[0045] Step 12: for m=1, 2, ..., M, where m is the number of iterations;

[0046] For each sample x i (i=1,2,…,N), calculate the residual r mi (The residual here is approximated by the negative gradient value of the loss function):

[0047]

[0048] Wherein, m=1,2,…,M, and M is the number of iterations.

[0049] Step 121: Based on the previous step mi Fit a new regression tree, whose leaf node area is denoted by R mj ,j=1,2,…,J, where J represents the number of leaf nodes of the mth tree.

[0050] Step 122: For j=1, 2, ..., J, calculate:

[0051] Step 123: Update the current learner and obtain: I is the discriminant function, which has the same meaning as I in the following text; R mj It represents the leaf node area, where j represents the number of leaf nodes of the mth tree.

[0052] Step 13: Get the final predictive maintenance optimization model for smart home appliances:

[0053]

[0054] Data preprocessing: When using the gradient boosted decision tree (GBDT) method to build a predictive maintenance optimization model for smart home appliances, data preprocessing is a key step to ensure model stability and performance improvement. For a given data set, the following detailed data preprocessing steps are crucial:

[0055] Step a: Handling missing values.

[0056] First, it is necessary to carefully check the integrity of each feature in the data set, identify and handle missing values. The presence of missing values ​​may seriously interfere with the model training process, so effective measures must be taken to fill them. For numerical features, such as temperature, humidity, current, voltage, etc. of household appliances, if there are missing values, the mean value (or median, determined by the data distribution characteristics) of the feature can be selected for filling. This strategy aims to maintain the integrity of the data while avoiding introducing excessive deviations. By filling in missing values, it is ensured that each data point contains complete information, providing a solid foundation for subsequent model training.

[0057] Step b: Divide into training set and test set.

[0058] After completing the missing value processing, the next step is to divide the dataset into a training set and a test set. This step is crucial for evaluating the performance of the model. Usually, the data is divided into 80% training set and 20% test set to ensure that the model can fully learn the rules and patterns in the data during the training process, while verifying its generalization ability on the test set. Randomly dividing the data samples can ensure the independence of the training set and the test set, thereby more accurately evaluating the performance of the model.

[0059] It is worth noting that although data standardization is one of the common preprocessing steps in the field of machine learning, the gradient boosting decision tree (GBDT) does not require standardization when processing data. This is because the GBDT model is a nonlinear model based on a tree structure, and its construction process mainly depends on the relative order of features rather than the specific numerical size. When dividing nodes, the decision tree focuses on the relative importance of features rather than the absolute value of the features. Therefore, data standardization is not necessary for the construction and prediction of the GBDT model. This feature makes GBDT more robust when dealing with features with different scales and ranges, without the need for complex linear transformations of features, so that it can capture complex nonlinear relationships in data more flexibly and effectively.

[0060] Step 2: Training of the predictive maintenance optimization model for smart home appliances based on the gradient boosting decision tree.

[0061] The main parameters involved in the training process of the GBDT model are as follows:

[0062] (1) n_estimators represents the number of decision trees, that is, the number of model iterations. Generally speaking, the larger the n_estimators value, the stronger the model fit, but too large a n_estimators value may cause the model to overfit. Therefore, when selecting n_estimators, it is necessary to comprehensively consider the model's fit and efficiency. The default value of n_estimators is 100.

[0063] (2) learning_rate represents the learning rate of the model. The learning rate represents the contribution weight of each base learner in the entire iteration. The default value is generally 0.1. It is used to control the step size of each tree model update, also known as the step size factor. Generally speaking, the smaller the learning rate, the slower the model converges, and more iterations are required to achieve the desired effect. Therefore, when adjusting the learning rate, it is necessary to comprehensively consider the number of iterations and the size of the learning rate to determine the appropriate parameter combination.

[0064] (3) subsample is used to control the subsample ratio used by each base learner, and to prevent overfitting by sampling the samples. The value range of subsample is (0,1], and it is generally recommended to be between [0.5,0.8]. The default value is 1. When the value is 1, it means that during the training process of the model, the training set used each time is the same as the original sample set, that is, the samples are not sampled.

[0065] (4) max_depth represents the depth limit of each decision tree, which is used to control the complexity of the model. When max_depth is small, the model's fitting ability is weak and it is prone to underfitting; when max_depth is too large, the model is prone to overfitting. The default value of max_depth is 3. When the depth is 1, the generated trees are all binary trees with only two leaf nodes, so there is no interaction between the variables. When the depth is 2, the generated tree model has interaction between two variables. By analogy, deeper base trees can explore very complex variable interactions, but at the same time they are also more likely to cause overfitting.

[0066] In the process of model parameter adjustment, the K-fold cross-validation method is usually used to find the optimal parameters. K-fold cross-validation is a commonly used model evaluation technique, which divides the data set into K non-overlapping equal subsets and uses K training and testing to evaluate the performance of the model. In each cross-validation, one subset is used as the test set, and the remaining K-1 subsets are used as training sets to train the model. The final result is the average of all cross-validation results. This study used 10-fold cross-validation and grid search methods for parameter optimization. First, the range of parameter combinations was determined through grid search, and then the model was evaluated in combination with 10-fold cross-validation to find the optimal parameter combination.

[0067] Step 3, evaluation of the predictive maintenance optimization model for smart home appliances.

[0068] In the predictive maintenance optimization project of smart home appliances, model evaluation is the core link to ensure the accuracy and reliability of model prediction. A reasonable evaluation process can not only help judge the pros and cons of the model, but also provide a clear direction for subsequent model optimization. Therefore, according to the task characteristics of the model, that is, regression tasks such as predicting the status of home appliances, identifying potential faults, and evaluating energy efficiency, appropriate evaluation indicators are carefully selected.

[0069] According to different model tasks, evaluation indicators can be divided into classification indicators and regression indicators. Classification indicators are usually used to measure the performance of classification algorithms, such as accuracy, recall rate, ROC (receiver operating characteristic) curve, AUC (area under the curve), etc.; while regression indicators are used to evaluate the performance of regression algorithms, including MSE (mean square error), MAE (mean absolute error), MAPE (mean absolute percentage error), R 2 (coefficient of determination), etc. Different models and application scenarios require the selection of appropriate evaluation indicators to measure the performance and applicability of the model.

[0070] Specifically, the present invention uses MAE (mean absolute error) and R 2 (coefficient of determination) is used as the main evaluation indicator. MAE can intuitively reflect the average deviation between the model prediction value and the actual value, which is of great significance for capturing the absolute error in the prediction. 2 It is an indicator that comprehensively measures the goodness of fit of the model. The closer its value is to 1, the closer the model's prediction result is to the actual value, and the better the fitting effect. The combination of these two indicators can comprehensively and accurately evaluate the comprehensive ability of the model in predicting the status of household appliances, identifying potential faults, and evaluating energy efficiency.

[0071] During the evaluation process, a test set that is completely independent of the training set was first used to ensure the objectivity and accuracy of the evaluation results. Then, the trained model was predicted using the test set data, and the MAE and R2 values ​​were calculated. By comparing and analyzing the results of these evaluation indicators, we can clearly understand the performance of the model in different aspects, including its prediction accuracy, stability, and possible problems.

[0072] Based on the evaluation results, we further analyzed the differences between the model prediction results and the actual values, and explored the reasons behind these differences. This includes comprehensive considerations of data quality, model parameters, feature selection, and other aspects. Through this series of analysis work, we can not only find out the problems and shortcomings of the model, but also provide strong support and guidance for subsequent model optimization.

[0073] Step 4: Interpretability analysis of the predictive maintenance optimization model results of smart home appliances.

[0074] In the smart home appliance predictive maintenance optimization project of the present invention, the GBDT model is selected as the core algorithm not only because of its excellent performance in prediction accuracy, but also because it is an integrated learning algorithm based on decision trees and has excellent interpretability. The GBDT model integrates multiple decision trees so that the decision logic of each tree can be clearly displayed. At the same time, with the help of rich visualization tools, it can intuitively show the importance of each feature to the model prediction results, as well as the degree of influence of the interaction between features on the model dependent variable.

[0075] In order to deeply understand the prediction basis of household appliance status, maintenance requirements, fault prediction and energy efficiency evaluation, the present invention conducts a detailed feature importance analysis. The GBDT algorithm builds the model in an iterative manner. Each round of iteration optimizes the new decision tree according to the residual of the current model, and updates the prediction ability of the overall model based on the contribution of these trees. In this process, the role of each feature in the model construction process, that is, its global importance, can be evaluated by calculating the average value of the local importance of the feature in all subtrees.

[0076] Specifically, using To calculate the global importance of feature j, and Respectively represent the global importance of feature j and the importance of feature j in subtree T n The local importance of , and N represents the number of subtrees.

[0077] The local importance of feature j in a single subtree is defined based on the reduction in the residual sum of squares when a node in the subtree is split and whether the feature is selected as the basis for splitting.

[0078] Among them, M is the subtree T n The number of leaf nodes, because the GBDT model is a binary tree, so the subtree T n The number of leaf nodes is represented by M-1. is the reduction in the residual sum of squares of non-leaf nodes after t splits. I(v t =j) is the discriminant function. When the feature that the non-leaf node depends on at t split is feature j, the function value is 1, otherwise, the function value is 0.

[0079] Through this analysis process, it is possible to quantify the importance of parameter characteristics such as temperature, humidity, current, voltage, usage time, working mode, and user operation records of household appliances in predicting equipment status, evaluating maintenance needs, predicting failures, and energy efficiency. This information not only reveals the key factors that affect the performance of household appliances, but also provides valuable references for improving user experience, promoting energy conservation and emission reduction, extending equipment life, enhancing equipment reliability, realizing intelligent management, and saving user time and energy. Therefore, in-depth analysis and understanding of the interpretability of the results of the smart home appliance predictive maintenance model is not only a technological advancement, but also an important driving force for promoting the development of the smart home field towards a more intelligent and efficient direction.

[0080] Step 5: Deploy the optimized GBDT model to smart home appliances, monitor the operating status and performance of the equipment in real time, discover potential risks in a timely manner, and collect user feedback and actual equipment operation data to continuously monitor and optimize model performance and achieve predictive maintenance optimization of smart home appliances. Set up a model adaptive update module, which can automatically adjust and optimize model parameters according to newly collected user feedback and actual equipment operation data to adapt to changes in equipment performance and changes in user needs.

[0081] In summary, the present invention proposes an innovative solution based on the GBDT model for the management and maintenance of household appliances, which significantly improves the intelligence level of equipment management and maintenance efficiency, which is specifically reflected in the following three aspects:

[0082] 1. Construction of high-precision prediction model: By constructing a relational model based on the GBDT model, the present invention successfully establishes a close connection between the characteristic parameters of household appliances and the equipment status, maintenance requirements, fault prediction and energy efficiency evaluation. With its high accuracy and powerful prediction capabilities, the model provides users with intelligent equipment management tools, effectively improving the reliability and overall performance of the equipment. At the same time, the data-driven approach makes household appliance risk monitoring and prediction more accurate, thereby promoting the dual improvement of household appliance efficiency and lifespan, and contributing to the realization of sustainable development goals.

[0083] 2. Improvement of intelligent safety management: This invention deeply analyzes the importance of characteristic parameters of household appliances in evaluating equipment status, predicting maintenance needs, failure risks and energy efficiency performance, and provides a scientific basis for the safety management of intelligent household appliances. By comprehensively considering the indicators of each characteristic dimension, this invention can fully reveal the influencing factors of household appliance risk scores, and propose targeted improvement measures accordingly, effectively prevent potential safety hazards, and significantly improve the safety management level of intelligent household appliances.

[0084] 3. Comprehensive consideration of multi-dimensional characteristic parameters: Different from traditional methods, this invention innovatively considers the characteristic parameters of different dimensions of household appliances at the same time and constructs a comprehensive and accurate relationship model. This model can deeply capture the multi-faceted characteristics of equipment operation, thereby achieving accurate prediction of equipment status, maintenance requirements, failure risks and energy efficiency performance. This method of comprehensively considering characteristic parameters of different dimensions not only provides users with more comprehensive and personalized equipment management and maintenance services, but also greatly improves the overall performance of the equipment and user experience, and promotes the management and maintenance of household appliances towards intelligence and personalization.

[0085] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A predictive maintenance optimization method for smart home appliances based on artificial intelligence technology, characterized in that: The method comprises: Step 1: Collect the temperature, humidity, voltage, current, usage time, working mode, user operation record, equipment status, maintenance requirements, fault prediction and energy efficiency evaluation data of household appliances, divide the collected data into training set and test set, and perform preprocessing, including checking and processing missing values, to ensure the integrity and accuracy of the data; Step 2: Using the gradient boosting decision tree (GBDT) algorithm and the training set, multiple decision trees are constructed through iteration. Each decision tree is fitted on the residual of the previous tree, and the prediction accuracy of the model is gradually improved until the preset number of iterations is reached or other stopping conditions are met, thereby training a predictive maintenance optimization model for smart home appliances. Step 3: Apply the trained GBDT model to the test set and calculate the evaluation indicators mean absolute error (MAE) and determination coefficient (R). 2 , evaluate the model’s predictive accuracy and generalization ability; Step 4: Analyze the feature importance of the model to understand the contribution of each feature to the prediction results, and optimize the model structure and parameters based on the evaluation results and feature importance analysis to improve the prediction performance and interpretability of the model.

2. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: The method also includes: step 5, deploying the optimized GBDT model to smart home appliances, timely discovering potential risks by real-time monitoring of the operating status and performance of the equipment, and collecting user feedback and actual equipment operation data to continuously monitor and optimize model performance and achieve predictive maintenance optimization of smart home appliances.

3. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: The GBDT algorithm in step 2 uses the K-fold cross-validation method to determine the optimal hyperparameter combination.

4. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: Step 4 uses visualization tools to display the decision path and feature importance of the model, thereby improving the interpretability of the model and helping users better understand the model prediction results.

5. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: The independent variables of the data set in step 1 include the current temperature of the home appliance, the humidity level in the environment, the current consumption of the home appliance, the voltage input of the home appliance, the operating time of the home appliance, the current working mode of the home appliance, and the user's operation record of the home appliance; The dependent variables of the dataset focus on the operating status, maintenance requirements, failure prediction, and energy efficiency evaluation of household appliances.

6. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: In step 2, the parameters involved in the training process of the GBDT model include the number of decision trees, the depth of the decision trees, and the learning rate.

7. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 2, characterized in that: The step 5 also includes: setting a model adaptive update module; the model adaptive update module can automatically adjust and optimize model parameters according to newly collected user feedback and actual device operation data to adapt to changes in device performance and changes in user needs.

8. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 1, characterized in that: The step 4 adopts a feature contribution evaluation method to determine which features have the greatest impact on the prediction results of the model.

9. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 8, characterized in that: The step 4 adopts To calculate the global importance of feature j, and Respectively represent the global importance of feature j and the importance of feature j in subtree T n The local importance of , and N represents the number of subtrees.

10. The predictive maintenance optimization method for smart home appliances based on artificial intelligence technology as claimed in claim 9, characterized in that: In step 4, Define the local importance of feature j in a single subtree, which is determined based on the reduction in the residual sum of squares when the node in the subtree is split and whether the feature is selected as the basis for splitting; where M is the subtree T n The number of leaf nodes; is the reduction of the residual sum of squares of non-leaf nodes after t split; I(v t =j) is the discriminant function. When the feature that the non-leaf node depends on at t split is feature j, the function value is 1, otherwise, the function value is 0.

Images